Remote Sensing & Dike Quality - 2012.11.01.1, 2, 4 & 6; Final report. FloodControl 2015

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6 FLOOD CONTROL 2015 > STIELTJESWEG 2 > 2628 CK DELFT > THE NETHERLANDS > T: +31 (0)88 33 57 446 > F: +31 (0)15 261 0821 [email protected] > WWW.FLOODCONTROL2015.COM

6

INFO Title Final report RSDYK: REMOTE SENSING & DIKE QUALITY - 2012.11.01.1-6 Commissioned by Stichting Flood Control Number of pages 82 Status Final

Version

1

Date

Author

2012-12-14

H.R.G.K. Hack

Signature

Reviewer

Signature

J. Maccabiani

Research team:  University Twente/ITC, Dr. H.R.G.K. Hack (project leader)  University Twente/ITC, S, Cundill, M.Sc.  University Twente/ITC, Dr. M. van der Meijde  Univeristy Twente/ITC, Dr. M. Noomen  Fugro, M. van der Meer  Fugro, L. Zwang  Haskoning, J. van der Schrier, M.Sc.  Stichting IJkdijk, W. Zomer Disclaimer While every effort has been made to ensure that the information herein is accurate, Stichting Flood Control does not accept liability for error or fact or opinion which may be present, nor for the consequences of any financial decision based on this information. The reports and material submitted by the various research providers, which are contained within the publication, have been prepared in accordance with reasonable standards of scientific endeavor. The research providers also have no control over its use by third parties, and shall likewise have no liability to a third party arising from their use of this information.

MANAGEMENT SUMMARY FC2015/RSDYK In the context of the Flood Control 2015 project, this project RSDYK2012 is done to establish the possible correlations between terrestrial remote sensing techniques, geological information of the surrounding subsurface, geophysical details of a dike and the quality of peat dikes. The project is done at two sites at the Reeuwijk location, and at one site, Veenderij, at the Amsterdam location, The Netherlands; the later in the context of the “Droogteproef Experiment 2012, Veenderij, Amsterdam”.

REEUWIJK LOCATION Spatial and temporal variations in the radiation temperatures measured by remote sensing have been established (Reeuwijk location). These thermal responses of the dikes are mainly related to the seasonal variation and to the distribution in the moisture content of the topsoil. The subsurface geology and stratigraphic profile have been obtained from interpolated pseudo-sections of the 2-D and 3D electrical imaging surveys and from boreholes and Dutch Penetration testing (CPT). The lateral and vertical variation as well as the heterogeneity of the dike material is obvious. In addition, three so-called “gas CPT” soundings have been done at the “Tempeldijk-North” site. The results show that at various depths gas is present in the subsurface. The gas-CPTs likely indicate that active decomposition (“rotting”) of the peat takes place in the subsurface. Soil moisture is one of the most important parameter affecting surface stability in soil structures. This is because in peat soil, the effective stresses and shear strength are directly related to water content, and even pre-failure deformations are largely controlled by the moisture content. The problems as seepage (“kwel”) and possibly subsidence in the dike Tempeldijk-South are identified by nearly all investigation methods, however often not unambiguously with just one single method. However, integration of the different investigation methods gives often a very clear indication of the quality and possible problems of the dike.

AMSTERDAM LOCATION Only one fieldwork campaign has been executed because the weather during 2012 was such that the conditions of the dike with respect to moisture content in the dike material did not change. This also limited the options to analyze the data because data of a dry dike could not be collected and are thus not available for comparison. It is expected that during 2013 a dataset under dry conditions can be collected and compared with the data collected in 2012. Speculatively, based on experience with the Reeuwijk location peat dike, it is likely that the remote sensing data can be correlated with the moisture content in the material at the dike surface and probably indicates subsurface inhomogeneity. However, the subsurface and processes between Reeuwijk and Amsterdam are different.

PUBLICATIONS In 2012 two scientific articles have been published, one article is in the process of being reviewed, and two presentations with extended abstract or poster have been presented and published.

CONCLUSIONS & RECOMMENDATIONS Main conclusions of this project are: Reeuwijk location: 

The comparison of the reference site “Tempeldijk-North” with “Tempeldijk-South” (a known “problem” location) shows that in all surface and subsurface investigations the Tempeldijk-South surface and

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

3

subsurface structure are more irregular which are likely due to or indicate “problems’ such as seepage (“kwel”) and subsidence. 

The thermal infrared images of Tempeldijk-South showed a layered structure which may be reflecting the subsurface structure of the dike. The layered structure is detectible likely because excess water is present in some of the layers. However, illumination differences may also be causing a layering effect. This is still subject of further investigation.



Visual images showed differences in vegetation cover at locations where excess water is likely present.



The data from the Algemeen Hoogtebestand Nederland may show patterns indicating deficiencies in a dike.

Thermal infrared in combination with near infrared imaging and in particular hyper-spectral imaging should be able to locate problem areas in dikes accurately. The near infrared or hyper-spectral imaging will likely be a supporting tool to be used to compensate the thermal infrared interpretation for vegetation, and environment and climate changes. A combination with LIDAR data would probably be an advantage, even the data of the Algemeen Hoogtebestand Nederland may already give sufficient accurate data for dike investigations. Amsterdam location: The purpose of the data collection is a comparison between remote sensing and ground data collected during wet and dry conditions of the site. That has not been achieved because of the weather conditions. In most years, the weather is such that long dry periods occur during the year. This was however not the case during 2012, but it is expected that during 2013 weather conditions are such that data under dry conditions can be collected. Speculatively, based on experience with the peat dike at the Reeuwijk location, it is likely that the remote sensing data can be correlated with the moisture content in the material at the dike surface and probably indicates subsurface inhomogeneity.

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

4

TABLE OF CONTENTS Management summary ............................................................................................. 3  FC2015/RSDyK ....................................................................................................... 3  Reeuwijk location ..................................................................................................... 3  Amsterdam location ................................................................................................. 3  Publications .............................................................................................................. 3  Conclusions & recommendations............................................................................. 3  Table of contents....................................................................................................... 5  1 

2 

3 

4 

Introduction ......................................................................................................... 7  1.1 

Context ........................................................................................................ 7 

1.2 

Progress ...................................................................................................... 7 

1.3 

Deliverables ................................................................................................. 7 

1.4 

Activities in 2012 .......................................................................................... 8 

1.5 

Embargo ...................................................................................................... 9 

1.6 

Parties and persons involved in the project ................................................. 9 

Background ....................................................................................................... 10  2.1 

Dikes .......................................................................................................... 10 

2.2 

Remote sensing and dike quality ............................................................... 10 

2.3 

Flood Control 2015 .................................................................................... 10 

2.4 

Objectives .................................................................................................. 11 

2.5 

Research methodology .............................................................................. 11 

2.6 

Climate ...................................................................................................... 12 

Literature ........................................................................................................... 13  3.1 

Introduction ................................................................................................ 13 

3.2 

Peat and clay as dike and dike foundation ................................................ 13 

3.3 

Water content and homogeneity ................................................................ 13 

3.4 

Remote sensing ......................................................................................... 14 

Geology ............................................................................................................. 19  4.1  Tertiary (Paleocene, Eocene, Oligocene, (Paleogene) & Miocene, Pliocene (Neogene)) (65 - 2.6 Ma ago) ................................................................................................... 19 

5 

4.2 

Pleistocene (Early- Quaternary) (2.6 Ma – 10,000 year ago) .................... 20 

4.3 

Holocene (Late-Quaternary) (10,000 year to present)............................... 21 

4.4 

Subsidence ................................................................................................ 23 

Reeuwijk location ............................................................................................. 25  5.1 

Introduction ................................................................................................ 25 

5.2 

Peat excavation ......................................................................................... 25 

5.3 

Tempeldijk ................................................................................................. 26 

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

5

6 

5.4 

Present situation ........................................................................................ 28 

5.5 

Surface lithology ........................................................................................ 28 

5.6 

Reeuwijk, Tempeldijk-South ...................................................................... 28 

5.7 

Reeuwijk, Tempeldijk-North ....................................................................... 51 

5.8 

Dike cover quality ...................................................................................... 53 

5.9 

Discussion & Conclusions Reeuwijk location ............................................ 56 

Amsterdam “STOWA Droogteproef” location................................................ 58  6.1 

Introduction ................................................................................................ 58 

6.2 

Data collection ........................................................................................... 59 

6.3 

Data collection requirements ..................................................................... 59 

6.4 

Data sets and quality ................................................................................. 59 

6.5 

Type of data ............................................................................................... 60 

6.6 

Evaluation .................................................................................................. 61 

6.7 

Discussion ................................................................................................. 63 

6.8 

Conclusions Amsterdam location .............................................................. 63 

7 

Communication; presentations and articles .................................................. 64 

8 

Conclusions ...................................................................................................... 65 

Appendix 1 :  1.1 

Ground surface data sensor equipment .................................................... 66 

1.2 

Geophysical subsurface investigation equipment...................................... 69 

Appendix 2 :  9 

Equipment .................................................................................... 66 

Droogteproef experiment 2012, Veenderij, Amsterdam ........... 72 

References ........................................................................................................ 76 

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

6

1 INTRODUCTION 1.1

CONTEXT

This report describes the activities of the project RSDYK2012 (FC2015) and gives the results achieved in 2012 in the context of the Flood Control 2015 research and development program (FC2015, 2013). RSDYK2012 is the 4th year of a research program and follows on the FC2015 projects RSDYK2008/2009, RSDYK 2010, and RSDYK2011. It is a combination of the deliverables of RSDYK2012 with the numbers 2012.11.01.1, 2, 4 & 6 as listed in the proposal. The activities and results are part of a scientific research to study the possibilities of using remote sensing techniques for identifying differences in quality of dikes with remote sensing emphasizing on passive remote sensing. Part of the research is done in the context of the research done for of a 4-year PhD program at the University of Twente. The activities in 2012 are a follow-up on the results in the forgoing years and therefore the results are integrated in the results of forgoing years.

1.2

PROGRESS

The research and development in the RSDYK2012 project is for a large part integrated in a PhD research. The PhD research will continue for another expected 14 months as planned. Therefore, not all results are yet final and work on integration and analyses of data will continue. In particular, the last phase of a research project is expected to deliver most results. At present (December 2012), most parts of the project are in the final phases of the research and development stages, or enter testing/validation and demonstration stages.

1.3

DELIVERABLES

The deliverables are listed in Table 1. The results of the various analyses in deliverables 2012.11.01.1, 2, 4 & 6 are interrelated and therefore discussed in one report (this report). The articles published on the research (deliverable 2012.11.01.6) are not included because copyright does not allow this. For the articles, refer to the publication media (see chapter 7). During the year, the results of the analyses of the data obtained in 2012 and previous years, and the weather during the year at the various test locations, lead to various changes in the work done during 2012.

1.3.1

Modification of deliverables

A main addition to the project is the execution of three so-called “Membrane Interface Probe CPTs” done at the “North-Tempeldijk” site in Reeuwijk. These CPTs have been done to establish the presence of gas in the subsurface. The report on these CPT measurements is included in deliverable 2012.11.01.3 & 5 (reported separately). The planned workshops have not been held during 2012 because already a major congress (Flood Risk Conference, Rotterdam) was organized in which also the research of this project was included. Table 1. Deliverables RSDYK for 2012.

Number Deliverable 2012.11.01.1

2012.11.01.2

Description English

Dutch

Report on the results and analyses of relations between remote sensing and material characteristics of dikes. Report on parameter and property determination

Eindrapport resultaten van analyses van de relaties tussen remote sensing en materiaal eigenschappen van de dijken Rapport over parameter en eigenschappen identificatie

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

7

2012.11.01.3 (*1)

Prototype for imbedding relations

2012.11.01.4

Report with analyses on the developed methodology

2012.11.01.5 (*1)

Report with description to use deliverables in a DIS environment

Eerste prototype voor inbedding van de gevonden relaties Rapport met analyses over de relevantie van de ontwikkelde methodologie Rapport met beschrijving methodiek om remote sensing technieken te kunnen gebruiken in DIS-modules Artikel

2012.11.01.6 (*2) Published articles Notes: *1) Deliverables 2012.11.01.3 and 2012.11.01.5 are reported in a separate deliverable “2012.11.01.5 Remote sensing for dike strength: DIS module for secondary levees”. *2) Articles cannot be included because of copyright; see publication details in chapter 7.

1.4

ACTIVITIES IN 2012

In 2012, the main activities are listed in the Table 2 below. Analyses of data have in 2012 mainly been done on data from the Reeuwijk location (chapter 0) and Veenderij location near Amsterdam (chapter 6). The data has been partially processed and analyzed on the options for dike quality detection and is summarized in this report. Table 2. Activities in 2012 in the RSDYK project.

date 2012.01.01-12.31

2012.01.19 2012.03.29 2012.03.30

2012.04.24 2012.04.25 2012.05.09 2012.07.22

Activity

Location

Processing and analyzing existing and newly obtained data, preparing articles and presentations, and PhD research and preparing PhD thesis. Project meeting with HHR Rijnland STOWA Kennisdag Inspecties Waterkering Presentation: “Remote Sensing Data for use in Regional Dike inspection”

Enschede

Project meeting “Droogteproef”, Veenderij, Amsterdam Project & “Sterke Cluster” meeting Project meeting with City of Reeuwijk Poster presentation: “Investigation of Remote Sensing for Dike Inspection”

2012.08.17-18

Field campaign

2012.11.01 2012.11.20-22 2012.12.12

Project meeting with HHR Rijnland Floodrisk conference Field campaign Gas CPTs

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

Leiden Arnhem Nederlands Aardwetenschappelijk Congres 11, Koningshof, Veldhoven, The Netherlands ((Cundill et al., 2012c) Amersfoort Nieuwegein Reeuwijk IEEE International Geoscience and Remote Sensing Symposium (IGARSS 2012) : Remote sensing for a dynamic earth, Munich, Germany, 22-27 July 2012. IEEE, p. 1. (Cundill et al., 2012a) Stowa “Droogteproef” test site, Veenderij, Amsterdam Leiden Rotterdam Tempeldijk-North test site, Reeuwijk

8

1.5

EMBARGO

The data and text in this report are (in part) to be included into scientific articles and in a PhD thesis at the University of Twente. Therefore, the data and text is under embargo until after the PhD examination. It should not be distributed outside the FC2015 consortium members and the client of FC2015. The data and text cannot be used in publications before the PhD study has been completed with the PhD examination.

1.6

PARTIES AND PERSONS INVOLVED IN THE PROJECT

1.6.1

Parties involved in execution of the project

Organization: University of Twente, Faculty of Geo-Information Science and Earth Observation (ITC): S, Cundill, M.Sc. Dr. H.R.G.K. Hack (project leader) Dr. M. van der Meijde Dr. M. Noomen

Organization: Royal Haskoning: J. van der Schrier, M.Sc.

Organization: Fugro: M. van der Meer L. Zwang

Organization: Stichting IJkdijk: W. Zomer

Organization: Delft University of Technology: Dr. D. J. M. Ngan-Tillard

1.6.2

Stakeholders involved

Organization: City of Bodegraven & Reeuwijk (formerly City of Reeuwijk): Contact person: Dr. Jan Rupke

Organization: Hoogheemraadschap Rijnland: Contact person: Onno van Logchem, M.Sc.

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

9

2 BACKGROUND 2.1

DIKES

Dikes (also called ‘levees’, ‘embankments’, or ‘dams’) protect a large part of the low-lying areas in the world. Breaches of dikes such as happened on an extensive scale in 1953 in The Netherlands show that dike breaches can result in many deaths and cause massive economic damage. More recent dike breaches in The Netherlands (Wilnis and Terbregge in 2003, and the near-failure of the dikes in Stein in 2004 and Tolbert in 2012) show that also on a smaller scale dike (near-) breaches can lead to major economic damage and to dangerous situations for those living behind the dikes. Increasing numbers of people living and working in lowlying areas and sea level changes due to subsidence and possibly climate change, make dikes increasingly more important. To assure the quality of dikes, dikes have to be inspected regularly.

2.2

REMOTE SENSING AND DIKE QUALITY

In many countries, dikes are inspected by a dike inspector walking the dike looking for irregularities (Givehchi et al., 2002; Moser and Zomer, 2006; Swart, 2007). It has been proposed that remote sensing is a possible tool for increasing the efficiency, objectivity and coverage of dike inspection with potential problematic areas being identified for further inspection (Givehchi et al., 2002; Hack et al., 2008; Swart, 2007). Two key features that dike inspectors evaluate are the quality of the dike covering and the soil moisture content of the dike (Moser and Zomer, 2006; Van Baars and Van Kempen, 2009). The covering of many dikes consists of grass. Multispectral, and more recently hyperspectral, remote sensing is used widely in vegetation studies (Blackburn, 2007; Lillesand et al., 2008). This is because different vegetation types and qualities can be distinguished, particularly in the near-infrared part of the electromagnetic spectrum. Additionally, vegetation is highly responsive to changes in soil moisture (Hopkins and Hüner, 2009). Remote sensing, especially hyperspectral and thermal, is actively researched for detection of water deficit stress in agricultural plants because this affects the quality and quantity of the harvest. The RSDYK project described in this report investigates the possibilities to establish with remote sensing the quality of the dike cover and the quality of the functionality of dikes as water retaining structure. The RSDYK project is executed under the Dutch research and development program Flood Control 2015.

2.3

FLOOD CONTROL 2015

Dikes are a flooding protection structure in the Netherlands and some other counties. According to Van Baars (2005), the primary (3,200 km) and secondary (14,000 km) dikes in the Netherlands protect more than 50% of the country from flooding. To maintain the groundwater level and drain the precipitation of the lower lands, water is pumped from the ditches into the canals and from the canals into the sea. Many of the secondary dikes are so-called “peat dikes”. These dikes consist of peat that has not been excavated while the surrounding peat is excavated. The peat has been excavated for fuel starting from the early middle ages. The peat and clay dikes act as a flooding tempering means in case a large flooding of the Western part of the Netherlands occurs. The flooding is unlikely to be stopped by these dikes but the lowering of the flooding rate may give opportunities to use dikes and the roads that are often on top for evacuation. Due to the large number of dike length, it is impossible to do a thorough investigation over the full length. Presently the quality of the dikes is established by visual inspection. Only at locations where the quality is visually deemed low is further investigation to the quality of the dike done. Apart from the fact that a visual inspection is slow and may be biased and subjective, a more important problem is that a dike in different seasons may behave qualitative differently, even on different days depending on the weather. The visual inspection is generally restricted to once every year (Bakkenist et al., 2012a), or more where is established

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

10

that the safety of the dike is not trusted, but certainly not on a basis that can ascertain that a dike is stable in all environmental conditions. Remote sensing from the air allows for a far faster means of inspection. However, although it has been thought for a long time that remote sensing may be an attractive option it has never been systematically studied. Therefore, this project has been initiated to establish whether remote sensing is a possible option for dike quality assessment before and during flooding situations. Within the context of the Flood Control 2015 (FC2015, 2013) project, the secondary peat dikes have a specific function. Secondary dikes may reduce the flooding rate in the Western part of the Netherlands when the main dikes against the sea and main rivers have failed. Important is then how long these dikes may still be able to function. Obviously, in a time of a major flooding in the Western part of the Netherlands no time will be available to start an investigation to the quality of the dikes. The quality of the dikes has therefore to be established beforehand. In the context of the Flood Control 2015 project, this project is done to establish the possible correlations between terrestrial remote sensing techniques, geological information of the surrounding subsurface, geophysical details of a dike and the quality of peat dikes. The RSDYK project is directed to dikes in The Netherlands and conditions that occur in The Netherlands. The project is done at two locations; one in Reeuwijk and one near Amsterdam, The Netherlands.

2.4

OBJECTIVES

The main objective of this project is to indicate possible relationships between terrestrial remote sensing, geological information of the surrounding subsurface, and weak areas in dikes mainly consisting of peat. Geophysics, boreholes, and Dutch Cone Penetration (CPT) tests have been done to investigate the subsurface of the dike. The project addresses the following specific objectives: 

Identify the spatial and temporal variations of the thermal radiation of the dike materials as well as reflectance features of the grass using thermal infrared (TIR) and near infrared (NIR).



Determine the variation in the composition of a dike, the soil moisture condition, and the material properties using two and three-dimensional (2D and 3D) electrical imaging surveys, boreholes and CPTs.



Indicate possible relationships between thermal infrared, near infrared, visual light remote sensing, and the subsurface model of the dike and possible weak areas in the dike.

2.5

RESEARCH METHODOLOGY

This project comprises of literature and desk studies, field data collection, and post field data analysis works. A literature review has been made on terrestrial remote sensing techniques (TIR and NIR) and physical parameters of peat dikes such as moisture content. Information about the geological setting of the study area has also been gathered from previous works of different researchers who worked in the study area. During the field data collection, field images of TIR, NIR and visual are acquired using ground based sensors at two dike locations (Reeuwijk and Amsterdam). This is done in three different season’s summer, autumn and winter. In addition, 2-D and 3D electrical imaging surveys are conducted at the Reeuwijk dike location. Boreholes and Dutch cone penetration tests have been done in Reeuwijk for referencing the geophysical subsurface model. Figure 1 shows a summarized schematic workflow that is used to achieve the objectives of the project.

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

11

Figure 1. Schematic work flow diagram

2.6

CLIMATE

According to Köppen-Geiger classification (Kottek et al., 2006; World Maps climate classification, 2013), The Netherlands has a temperate sea climate with rain almost throughout the whole year. In general, the winters are mild having an average mean temperature of 1.7°C. The mean temperature may be below zero in the coldest month. In summer, five months have a mean temperature over 10°C with a maximum temperature of 17°C in July. The precipitation is evenly distributed over the year with a yearly average of 760 mm (Ten Cate, 1982). In spring, precipitation is low which causes a deficit in surface water due to evaporation.

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

12

3 LITERATURE 3.1

INTRODUCTION

Continuous monitoring of dikes is vital to secure their stability and to protect from the major impact on the environment and casualties should these fail (McCahon et al., 1987).

3.2

PEAT AND CLAY AS DIKE AND DIKE FOUNDATION

The stability of peat dikes vary with time and therefore monitoring of peat dikes is important (McCahon et al., 1987). The stability of peat mass movements is not fully understood under all circumstances (Carling, 1986; Dykes and Kirk, 2001; Van Baars, 2005; Van Baars and Van Kempen, 2009). Various mechanisms but in particular the hydrological and geotechnical conditions determine the stability of peat dikes (Nterekas et al., 2011; O’Kelly, 2008). These conditions are usually affected by seasonal variations, which is often a main cause of failure in many engineering structures as described, for example, in publications of the International Peat Society (2013), in the conferences of the Mire Research Group (Tallis et al., 1997), and in articles such as Evans et al. (1999). Ward et al. (1955) described the risk of a peat layer under a dike. He indicated that dikes founded on very weak peat might collapse within a short period after construction. Instability can occur in peat dikes even if they are on top of an impervious material like clay (Carling, 1986). This is because peat dikes can have less weight than the resultant water force especially when the crest of the dike dries out (Van Baars, 2005; Van Baars and Van Kempen, 2009). This resultant force can be influenced by a rise of water level in the canals, ditches, or streams.

3.3

WATER CONTENT AND HOMOGENEITY

In The Netherlands, the effective soil water content is maximal at the beginning of the spring and then decreases until the end of summer (Béhaegel et al., 2007). Soil thermal properties are strongly influenced by the soil volumetric water content, volume fraction of solids, and volume fraction of air. The distribution of water content and total unit weight vary in both vertical and horizontal directions in peat layers (Hobbs, 1986; Sikder, 1994). The retention of water in peat is assumed to be free water in large cavities, capillary water in narrower cavities, and water physically or chemically bound (Boylan et al., 2008; Dalton, 1954; MacFarlane and Radforth, 1965). In peat, the effective stresses and shear strength that determine the stability are directly related to the water content. The water content of the topsoil varies with respect to the seasonal variations. The reduction of the water content of the topsoil during the dry conditions in the summer can result in drying and shrinkage of the peat layer. This will cause new cracking, reactivation of old cracks, and opening of peat fuel cuttings (Long and Jennings, 2006). During a following intense rainfall, water can rapidly percolate to the base of the peat through the new and old cracks without saturation of the peat layers in the top of the dike. Hence, the weight of the peat in the top layers is still low whereas the pore pressures in the peat at lower level will have increased significantly. This reduces the effective stresses and the resistance to sliding. It is also possible to speculate that repeated drying and wetting cycles caused shrinkage and swelling movements in the peat (Warburton et al., 2004). The soil moisture content is also a key parameter in computing the surface energy balance and important in many applications including hydrology, agriculture and meteorology (Petrone et al., 2004; Szatyłowicz et al., 2007). Other factors like specific gravity, organic content, heterogeneity in soil texture, vegetation, land use, topography and surface temperature also affect the stability of peat dikes (Li and Islam, 1999; Tansey et al., 1999).

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

13

3.4

REMOTE SENSING

Remote sensing in all ranges of the electro-magnetic spectrum (Figure 2) has many applications in geotechnical investigations. It is also used for mapping the top soil moisture over a varying landscape (Famiglietti et al., 1999; Li and Islam, 1999), identifying engineering structures, and detection of hazardous gas leakage of pipelines from the reflectance spectral signature of stressed vegetation (Van der Werff et al., 2007). The Ministry of Infrastructure and the Environment (Rijkswaterstaat) in The Netherlands, has made an inventory of the possibilities of remote sensing for the purpose of dike quality assessment (Swart, 2007). In this publication, the possible options for using remote sensing are described based on a literature review. Any vegetation present on and around the dikes is likely to be influenced by changes in groundwater table or moisture content of the material and vice versa. The health of the vegetation can be affected as the groundwater table becomes too shallow or too deep. The most likely changes are expected to occur in the chlorophyll concentrations in the vegetation, which are an indicator of the health state. In stressed vegetation, the absorption efficiency of the chlorophyll decreases. In very stressed vegetation, when physical changes start to manifest, the IR reflectance decreases due to changes in the cell structure of the plant. This leads to a reduction in reflectance in the IR simultaneous with an increase in reflectance in the red and green. Hence, if the stability of peat and to a certain extend also clay dikes depends on the moisture content, and the health of the vegetation on a dike is dependent on the moisture content, and it is possible to establish the health of the vegetation by remote sensing, it should then be possible to establish a relation between remotely sensed images and the quality of the peat and probably clay dikes.

Figure 2. The electro-magnetic spectrum (figure courtesy of Pion, 2013). A difference in the reflectance of grass, which covers a peat dike, might relate to the soil moisture variation of the material. The health of plants is reflected in its chlorophyll content (Van der Meijde et al., 2006). Adams et al. (1999) demonstrated that healthy vegetation shows high reflectance in the green and NIR region (Figure 3). However, the chlorophyll of stressed or dry vegetation decreases its absorption efficiency and increase in reflectance in the red (Adams et al., 1999; Gausman, 1974; Tucker, 1979). Environmental factors such as soil, geomorphology, and vegetation apparent roughness influence the reflectance values. Variations in climatic

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

14

factors in particular, precipitation and temperature, have therefore a strong influence on variation in the reflectance.

Figure 3. This general diagram shows the stress indicated by a progressive decrease in Near-IR reflectance accompanied by a reversal in Short-Wave IR reflectance (source: NOAA-CSC, 2008).

3.4.1

LIDAR

A remote sensing technique, which is already operational in dike management, is LIDAR (Laser Imaging Detection And Ranging), or laser altimetry as it is often referred to. LIDAR is used to generate elevation information, and more particularly digital elevation models (DEMs) from which the height and profile information for dikes can be derived. The level of detail of aerial LIDAR is such that subsidence and deformations can be detected in dikes. Rijkswaterstaat and the Dutch water boards arranged for the generation of a nationwide elevation model from LIDAR data, which is flown every year from 2008. This is called the Actueel Hoogtebestand Nederland (in 2013: version 2) (AHN, 2013) and is ideal for change detection (Moser and Zomer, 2006; Swart, 2007). With high point density and multiple reflections, it is even possible to detect higher grasses due to seepage (Swart, 2007).

3.4.2

Radar interferometry

Related to LIDAR, is radar interferometry. Although this technique operates in the microwave wavelengths and not the optical like LIDAR, radar interferometry also detects displacements in the x, y and z-directions. It thus should be possible to use it for dike geometry and in particular dike height, profile, subsidence, or deformations. Radar interferometry is operational for dike management (Hanssen et al., 2007; Hanssen and Van Leijen, 2008; Moser and Zomer, 2006; Swart, 2007). Radar interferometry requires a stable platform and thus obtains better results when the sensor is on a satellite. However, this often results in low spatial resolution and so LIDAR is superior for DEM creation for dike management. An advanced radar satellite technique, called PS-InSAR or Persistent Scatterers Interferometric Synthetic Aperture Radar (PSI), can detect deformations of 1mm/yr; movements over wide areas can be detected and monitored with even greater sensitivity. PSI typically works best with dikes with a hard cover, such as stone or asphalt. Although research projects (Terrafirma, 2013) are still ongoing, the technology is likely very useful for detecting small, order of millimeters, dike deformation and trends in dike deformation (Bermon, 2008; Cooksley et al., 2013; ESA; Satellites keep an eye on Dutch dikes, 2012; Gernhardt et al., 2010). Excessive dike deformation may often be the onset of dike failure.

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

15

3.4.3

Visible photography

Visible light (VIS) photography is also operational in dike management and digital aerial photography has replaced film. Visible photography can be used to observe some visual inspection parameters and for landscape orientation (Moser and Zomer, 2006). Visible photography is ideally suited for change detection (Moser and Zomer, 2006; Swart, 2007). It could also be used to identify discoloration of ditch water due to displaced soil particles as a result of seepage (Swart, 2007). In addition, analysis of the individual bands in visible photography, namely red, green and blue, provide further useful information. The decrease in reflectance in the red band but not in the green could be indicative of increasing leaf area index (LAI) (Asner, 1998) or water deficit stress (Kim et al., 2010). Using the ratio Green:Red, low values could be indicative of the factors that result in an increase in overall brightness (see Hyperspectral Imaging). Plant stress, in general, reduces chlorophyll concentration (Blackburn, 2007; Carter and Knapp, 2001; Carter and Spiering, 2002) and chlorophyll concentration is strongly correlated to reflectance in the green wavelengths (Carter and Knapp, 2001; Jensen, 2007).

3.4.4

Near infrared photography

Near infrared (NIR) photography or imaging has been used extensively in vegetation studies due to vegetation’s high reflectance in this portion of the electromagnetic spectrum. Healthy, lush vegetation reflects strongly in the near infrared portion of the electromagnetic spectrum. Other factors can also result in higher reflectance values in the near infrared. These include lower water content and more standing litter which both result in higher overall reflectance but with different shaped spectral signatures (Asner, 1998; Jensen, 2007). Standing litter has a disproportionate effect on grass canopy reflectance, including the NIR (Asner, 1998). Band ratios, including the ratio NDVI (Normalized Difference Vegetation Index), usually correct for illumination differences (Lillesand et al., 2008) which makes them valuable in evaluating areas with differing illuminations. The NIR/Green ratio can be used to indicate stress-related changes in chlorophyll concentrations. The reason for this is that NIR is relatively stable for slight changes in chlorophyll concentration but an increase in green reflectance is indicative of chlorophyll reduction (Carter and Knapp, 2001). NIR photography is ideal for evaluating the quality and condition of vegetation (Moser and Zomer, 2006; Swart, 2007). Although it is operational for determining the quality of, for example, trees and shrubs, there are no known examples of NIR photography being applied to the grass coverings of dikes (Moser and Zomer, 2006). Swart (2007) proposes that NIR photography is suited to mapping water, including ditches, and could be used to detect movement in duckweed, indicating possible seepage.

3.4.5

Hyperspectral imaging

Hyperspectral imaging has largely been ignored by dike managers and consultants, with their focus more on NIR imaging. However, Moser and Zomer (2006) propose that hyperspectral imaging would be more suited to certain applications, such as detecting plant deterioration which results in an alteration of the reflectance spectrum of the plant. These alterations can be subtle and not being detected by NIR imaging. Two advantages of using hyperspectral data are that the shape of the spectral signature can be examined and that unique spectral features in non-traditional wavelengths or narrow-bands can be discovered. An increase in overall brightness or increased reflectance can be attributed to increased standing litter, a decrease in mean leaf incident angle (MLA) (Asner, 1998), water deficit stress (Carter, 1991; Jensen, 2007; Kyllo, 2003; Nowatzki et al., 2004) or a decrease in chlorophyll concentration (Carter and Knapp, 2001). As discussed in the section on visible and near infrared photography, information about vegetation condition can be obtained from the green, red and near-infrared wavelengths. Additionally, Carter and Knapp (2001) found that as chlorophyll concentration increases so the spectral reflectance at 683nm (red limit) decreases. Not only can vegetation condition be determined but vegetation types can be differentiated based on their unique spectral signatures, with better discrimination at the start of the growing season (Swart, 2007).Thus

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

16

hyperspectral imaging is also suitable for species differentiation, which may be an indication of seepage. In addition, the discoloration of ditch water by suspended soil particles from seepage may also be more easily detected using hyperspectral imaging if the discoloration is slight.

3.4.6

Thermal infrared

Thermal infrared remote sensing is widely used for many applications where different surface temperatures indicate a feature in the subsurface, for example, detecting and monitoring fires in underground coal beds (Yang, 1995), hydrothermal features (Heasler et al., 2009), archaeological surveys (Heitger, 2006), and military use (Howard, 2001). The presence of moisture and water also change the surface temperature of a structure and therefore more moist areas or water seepages can be found by thermal remote sensing such as leakage in tunnels (Inagaki and Okamoto, 1997), pipelines (Chunli et al., 2005), and dams and canals (Nellis, 1982). Thermal remote sensing is based on the thermal infrared range of the electro-magnetic spectrum. According to Planck’s Radiation law, all objects above 0°K emit thermal electromagnetic energy in the 3.0 - 14 μm wavelength region. The emissive power of a black body at any wavelength as well as the amount of emitted energy per wavelength depends on the object’s temperature. Different materials can have widely different emissivity values within the range of 0 to 1. The range of emissivity for ground components in situ, i.e. soil, vegetation and rocks, varies at a given wavelength according to their physical properties and water content (Blumberg et al., 2002; Blumberg et al., 2000; Fuchs and Tanner, 1966; Van de Griend et al., 1991). Planck's law gives the spectral radiance of electromagnetic radiation at all wavelengths from a black body at temperature T as a function of wavelength λ:

I  , T  

 

2 h 3 c2

1 e

h kT

1

I ν,T is the energy per unit time (or the power) radiated per unit area of emitting surface in the normal direction per unit solid angle per unit frequency by a black body at temperature T

[Eq.1]

h is the Planck constant; c is the speed of light in a vacuum; k is the Boltzmann constant ν is the frequency of the electromagnetic radiation; T is the temperature of the body in kelvins

The emissivity power increases with temperature at each wavelength and the position of the maximum emissive power shifts towards shorter wavelengths, i.e. relatively more energy is emitted at shorter wavelengths (Figure 4).

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

17

Figure 4. Blackbody emission vs. wavelength for different temperatures (modified after Schubert, 2006). Many researchers (e.g. Idso et al., 1975; Price, 1980; Reginato et al., 1976) assessed and mapped soil moisture by thermal infrared using satellite images and airborne sensors for studying biophysical processes. Jackson et al. (1981) showed the difficulties for retrieval of soil moisture due to the influence of surface variables like vegetation cover. A thermo tracer is a highly sensitive radiometric camera that measures the infrared radiation in the thermal region emitted from objects. The camera used for this project is a NEC thermo tracer (NEC TH9100, 2013). The thermal infrared images by a thermo tracer are shown to be useful to map areas characterized by different soil moisture content (Mora et al., 2007).

3.4.7

Passive Microwave Radiometry (PMR)

Passive Microwave Radiometry (PMR) is a remote sensing technique that measures radiation emitted between a few millimeters to decimeters in wavelength. Due to atmospheric, ionosphere, and microwave communication influences, measurements are usually only made between 0.8 and 21 cm. Radiation in these wavelengths is mainly a function of water content of the soil, since the radiation is a function of the dielectric constant and is thus related to conductivity (Moser and Zomer, 2006). However, the radiation is also influenced by the height of the groundwater table, biomass, salinity, and the temperature of open water (Moser and Zomer, 2006; Swart, 2007). The wavelength of the radiation detected depends on the depth of the conductive (i.e. moisture or water) features in the subsoil, but also depends on soil moisture levels (Haarbrink and Shutko, 2008; Moser and Zomer, 2006). PMR has been used for dike management to establish dryer and more moist areas in dikes (Miramap, 2013).

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

18

4 GEOLOGY The Western part of The Netherlands, including the Reeuwijk and Amsterdam locations (see section 0 and section 6 respectively) is influenced by the geology mainly starting from the Tertiary (65 Ma ago) in the context of the tectonic North Sea Basin. A large part of the details in this section is based the book “Ondergrond van Nederland” (De Mulder et al., 2003).

4.1

TERTIARY (PALEOCENE, EOCENE, OLIGOCENE, (PALEOGENE) & MIOCENE, PLIOCENE (NEOGENE)) (65 - 2.6 MA AGO)

The North Sea Basin developed in northwestern Europe at the end of the early Tertiary (Figure 5). The area now known as the Netherlands is located at the southern tip of the basin. During the Tertiary, the basin subsided gradually and continuously filled up with sediments (Bosch and Kok, 1994; De Mulder et al., 2003; Ten Cate, 1982). At the end of the Pliocene (end of Tertiary), the coastline was situated over the Southeastern part of the Netherlands (Figure 6). The base of the Tertiary is at about 700 m below surface in the Reeuwijk area and around 900 to 1,000 m under the area of the Amsterdam location. The thickness of Tertiary sediments in around 450 m below Reeuwijk and 650 m below the Amsterdam location and consists of layers of originally clay, silt, sand, and minor quantities of calcareous and organic material. The degree of consolidation and cementation varies, but generally is higher in the deeper layers. Many of the layers would be regarded geotechnically as rock rather than soil, i.e. clay-, silt-, sand-, and limestone, and coal.

Figure 5. Development of the North Sea Basin during the Tertiary (modified from De Mulder et al., 2003).

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

19

Figure 6. Coastline at end of Pliocene (end of Tertiary) (modified from De Mulder et al., 2003).

4.2

PLEISTOCENE (EARLY- QUATERNARY) (2.6 MA – 10,000 YEAR AGO)

During the Pleistocene (from 2.6 Ma to 10,000 year ago), multiple glaciations spread over Northern Europe. As far as known, the Reeuwijk area has never been glaciated during the Pleistocene, but the presence of the glaciations to the north and east of Reeuwijk influenced the morphology of the landscape and sedimentation pattern in the Reeuwijk area. The ice during the Saalian glaciation came nearest and the maximum extent of the ice reached to some 40 km north and northeast of the Reeuwijk area (Figure 7). The Amsterdam location was just on the Southern edge of the Ice sheet during the Saalian glaciation. The glaciations forced the rivers Rhine and Meuse into westerly courses. During the glaciations, the area of Reeuwijk remained in the peri-glacial zone and in the warmer periods in between the glaciations, the climate was more moderate. The base of the Pleistocene is about 250 m below surface in the Reeuwijk area and 300 m below the Amsterdam location. The thickness of the Pleistocene is about 240 m in the Reeuwijk area and 290 m in the area of the Amsterdam location. The Pleistocene consists mainly of marine and fluvial sandy deposits with some more silty and clayey layers.

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

20

Figure 7. Maximum extent of ice cover during Saalian (modified from De Mulder et al., 2003).

4.3

HOLOCENE (LATE-QUATERNARY) (10,000 YEAR TO PRESENT)

At the end of the Pleistocene (10,000 year ago), i.e. at the end of the Weichselian glaciation, part of the (later) North Sea and the western Netherlands was a continental gently westward sloping plain mainly consisting of sandy deposits (Ten Cate, 1982). Climate change caused a very rapid sea level rise accompanied by a rise of regional groundwater table (Figure 8a). In this period, three zones of sedimentation can be distinguished: a littoral sandy zone of coastal barriers and dunes, a clayey zone of tidal flats, salt marshes, and brackish lagoons, and, at a greater distance from the sea, a zone of peat formation in a fresh water environment. This system of different sedimentation zones shifted towards the east as the sea gradually flooded the former dry North Sea floor (Figure 8b) (De Mulder et al., 2003; Ten Cate, 1982).

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

21

Figure 8. (a) Sea level during Holocene; (b) Coastline during Holocene (modified from De Mulder et al., 2003). Sedimentation in the Holocene period started with the formation of peat (basal peat). The coastline moved further to the east and reached the present Netherlands in about 8,000 BP. The rate of sea level rise reduced to 27 cm/100 years until 5,000 BP and the extension of marine deposits reduced significantly (Bijlsma, 1982). Figure 9 shows the development of the Netherlands during the Middle- and Late-Holocene. The groundwater level was still high which allowed the development of a thick peat layer over the marine and fluvial deposits. This peat forming process continued until 700 BP in the central part of the Netherlands. When the peat layer was inundated and/or eroded by the water, marine or fluvial sediments were deposited over it. The base of the Holocene is about 7 – 9 m below the surface in the non-excavated areas around Reeuwijk and Amsterdam areas (which is about -10 m NAP).

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

22

Figure 9. Paleography of The Netherlands during Middle- and Late-Holocene (modified from De Mulder et al., 2003).

4.4

SUBSIDENCE

The whole of The Netherlands is part of a subsiding tectonic sedimentary basin (North Sea Basin - see above). The subsidence during the Quaternary (i.e. up to today) is not everywhere the same and is not constant in time. The average tectonic subsidence in the Reeuwijk area over the last 2.5 Ma is 3 - 4 mm/100 year. The area is also subsiding by isostatic compensation of around 5 - 6 mm/100 year due to increasing load of sediments, varying load by ice due to the glaciations, and varying load by seawater. The sediments and in particular recent sediments undergo compaction due to overlying loads that results in a subsidence of around 1.5 - 2 mm/100 year. The tectonic subsidence rate in the Amsterdam area is larger explaining the thicker Tertiary and Pleistocene deposits. Water extraction by man for drinking and agricultural water in the top 10 to 200 m layers increases this last cause for subsidence. Moreover, the peat deposits decay (oxidise) rapidly and cause a substantial decrease of the terrain level when oxygen is allowed to enter. In particular, the peat excavations and lowering of groundwater level for agricultural land use cause a very fast peat decay due to oxygen entering from the air or

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

23

groundwater flow with dissolved oxygen. The overall subsidence in the area of Reeuwijk is in the order of 1 cm per year but locally it maybe 4 – 10 cm/year.

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

24

5 REEUWIJK LOCATION 5.1

INTRODUCTION

The research for the identification of dike quality by remote sensing has concentrated on the Tempeldijk, Reeuwijk (Figure 10). The Tempeldijk is a dike located north of the village of Reeuwijk-Dorp in the province of Zuid Holland, in the central western part of The Netherlands. The dike is located in a so-called “polder area” (Figure 10 & Figure 14).

Figure 10. Test site locations Reeuwijk-Dorp (Google Maps, 2012).

5.2

PEAT EXCAVATION

In the area, as in large parts in the West of The Netherlands, excavation of peat for fuel has taken place since the early Middle Ages (first reported around 800 AD) and on a large and systematic scale between the 10th and mid-19th century (Nienhuis, 2008). The groundwater level in the area was near surface and excavation took mostly place underwater leaving the excavated areas inundated (Figure 11). Between the excavated areas, strips were left for roads, housing, and some local farming. Wind and resulting water wave action caused the strips to erode, and increased decay by oxidation of the peat in the strips caused subsidence (section 4.4). The excavated areas became connected and formed (large) lakes, which continuously increased in size by wave erosion. This became a hazard for surrounding villages and towns. Moreover, the value of the land for agricultural use increased and starting from the 16th and especially during the 18th and 19th centuries, the land was reclaimed by pumping out the water by windmills and constructing a system of ditches and canals to drain the water to rivers and sea (Figure 12).

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

25

Figure 11. Peat-cutter (Luyken, 1694).

Figure 12. Sketch of peat excavation (vertical scale exaggerated) (from Cundill et al., 2012b).

5.3

TEMPELDIJK

The Tempeldijk is the boundary between a “high” in-situ peat deposit area where peat has not been excavated, and a “low” area where peat has been excavated. The elevation of the “high” area (to the east of the Tempeldijk) is about 1.5 m below NAP (National Mean Sea-Level Reference – M.S.L.) and “low” (to the west of the Tempeldijk) is about 6.5 m below NAP (Figure 13). It functions as a dike (e.g. dam – in Dutch: “boezem kade”) for a de-watering canal (ditch) (Figure 12). Two test sites are used; one on each end, i.e. Tempeldijk-North and Tempeldijk-South (Figure 10 & Figure 14).

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

26

Figure 13. Digital Elevation Model (DEM) of the area of Figure 14.

Figure 14. Reeuwijk location test sides. The urban area in the bottom middle is Reeuwijk-Dorp (aerial photo: Google Earth/Aerodata International Surveys; image date 2005, 2013) Grid: UTM (WGS84, zone 31 NH). REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

27

5.4

PRESENT SITUATION

At present, land use is mainly meadows for dairy production (milk and cheese; original “Gouda Cheese” – the town of Gouda is about 5 km south), and flower, shrub, and tree horticulture. Especially the horticulture has a high economic value. Both agricultural uses are very sensitive to the groundwater level. Too high groundwater level results in a too low bearing capacity of the terrain for the cows, while a too low level reduces the grass growth. The horticulture is also very sensitive for groundwater level. Foundations of old houses and structures (> 50 years old) vary, but are mostly shallow strip foundations consisting of materials such as wood, waste, bricks, stones, or are founded on (short) mostly wooden piles into clayey or onto more sandy/silty layers. New houses and structures are founded on concrete piles down to Pleistocene sand at about 7-10 m below NAP (geology see section 4). Most roads are on an embankment of sand, gravel, stones, or any other strong and more or less durable material available at the time of construction. The embankments are directly on in-situ peat and clay. These roads often suffer large differential settlement requiring regular maintenance. Some newer roads are made with either very light-weight embankments (polystyrene) that “float” on the water-saturated peat and clay, or by a (reinforced) concrete deck on concrete piles onto Pleistocene sand.

5.5

SURFACE LITHOLOGY

The resulting Holocene lithology in the Tempeldijk-Reeuwijk area follows the “new” lithological divisions with the formerly common stratigraphic names in brackets (De Mulder et al., 2003). The former common stratigraphic names were determined on borehole cores but the new layer identification is based on the descriptions in the literature only. The Holocene lithological formation of “Nieuwkoop; layer Hollandveen” (formerly Hollandveen) consisting of a fairly consistent peat layer on and partially interbedded with the formation of “Naaldwijk; layer Wormer” (formerly Calais III) consisting of mainly clay layers and marine and fluvial fine sand channel fill, and with the formation of Echteld (formerly Gorkum) consisting of sandy and silty clay with thin sand layers and lenses. The underlying Pleistocene formation of Boxtel consists of sand (Bosch and Kok, 1994; De Mulder et al., 2003).

5.6

REEUWIJK, TEMPELDIJK-SOUTH

Tempeldijk-South site (Figure 14 & Figure 15) is reported to have problems due to seepage (“kwel”) and possibly subsidence. For a more detailed description of Tempeldijk-South refer to the final report of the pilot project RSDYK2008 (Hack et al., 2008).

Figure 15. Tempeldijk-South site panorama photo, Reeuwijk (direction ditch at the left about NorthSouth and fence on the right about East-West).

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

28

5.6.1

Subsurface investigation

The test site Tempeldijk-South measures about 100 by 50 m along the Tempeldijk (Figure 16). The test site is the west side of the dike. The top of the dike is at about -2 m while the bottom of the dike is at about -5 m NAP. The area is covered with grass that is regularly cut in summer. Seepage (in Dutch “kwel”) occurs in the ditch and at the foot of the dike at the western side of the dike. Possibly a part of the dike has (slightly) subsided as indicated by the elevation contour lines (not shown). The elevations are based on the data of the “Actueel Hoogtebestand Nederland” (AHN, 2013). At the site of Tempeldijk-South, 2 boreholes and 17 CPTs (Dutch Cone Penetration tests) with pore water pressure measurement have been made (Figure 16). The locations, and borehole, including photo logs, and CPT logs are included in Hack et al. (2008). Directly beside a borehole a CPT test has also been done to facilitate interpretation of the CPTs. The boreholes are made with a so-called “Delft Continuous Soil Sampler” (a type of triple-tube core sampler). Borehole logs have been made by visual description of the borehole cores.

Figure 16. Tempeldijk-South test site area Boreholes and CPTs (aerial photo: Google Earth/Aerodata International Surveys; image date 2005, 2013) Grid: UTM (WGS84, zone 31 NH).

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

29

5.6.2

Subsurface model

The borehole and (Dutch) Cone Penetration Tests (CPTs) logs obtained at Tempeldijk-South have been included in a three-dimensional geological model. The interpretation has been done starting with the description of the boreholes coupled to the nearby CPT. The CPTs are interpreted loosely following the standards commonly used in The Netherlands and following international standards (Abu-Farsakh et al., 2008; Robertson, 1990) for CPT interpretation. Interpretation of soil lithology based on CPT data and in particular, details of peat and peat-containing layers is notoriously difficult. Variations in remains of different type of plants or in the coherence of plant remains give changes in CPT values that are difficult to correlate to the visual description of the peat layers in the boreholes and between CPTs. 5.6.2.1

Modelling and model likelihood

The interpolation of the boundaries between the lithology units is done by “Kriging” (details see Table 3). The model has been built up from bottom to top. “Thin sections” where the thickness of a layer is less than 0.03 m are transparent in the model, i.e. a layer less thick than 0.03 m is assumed not to exist. The layers are compared to horizontal distance and amount of information horizontally, relatively thin. This allows interpolation routines to become “creative” which reduces the reliability of the boundaries. The certainty of the depth of a boundary also reduces further away from a borehole or CPT. Near the perimeter of the interpolated area automatic boundary interpolation becomes very unreliable and therefore additional artificial boreholes have been added where required to maintain a geologically logical interpolation (Figure 17). Figure 18 shows the three-dimensional model of the subsurface with the top most (“PEAT/CLAY man-made”) layer removed (for details about the geology see section 4 and 5.6.3). The top layer is removed because it consists of a mixture of peat and clay likely created by the farmers to improve the bearing capacity of the ground (Rupke, 2008). This layer follows the topography and is not part of the lithology as existing after excavation of the peat. Figure 19 shows the plan view of the 3D model of Figure 18. Table 3. Interpolation

RockWorks 15 settings Interpolation method: grid size: settings: Logarithmic High fidelity Polynomial Enhancement Smooth grid:

Densified

Kriging 0.25 x 0.25 x 0.01 m3 (x x y x z) automatic routine to improve interpolation of anomalous data routine to better approximately honor control points routine to better honor a regional trend by fitting a polynomial surface (set to automatic order) settings: filter size = 2 (i.e. average depth is based on 24 surrounding nodes) iterations = 1 (i.e. the model is one time calculated) automatically added Delaunay triangulation midpoints to the source xyz grid

Note: The modeling has been done with the interpolation method and options that gave a geologically realistic model of the subsurface. It is not claimed that these are the only or the best methods and options.

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

30

Figure 17. Artificial boreholes to control automatic interpolation. (aerial photo: Google Earth/Aerodata International Surveys; image date 2005, 2013) Grid: UTM (WGS84, zone 31 NH).

5.6.3

Generalized subsurface conditions

The subsurface from the surface downwards can be generalized for the Tempeldijk-South site (see also the 3D model in Figure 18). The lithology names refer to the names used in the sections and 3D model. The generalized composition of the dike is: 

From the surface, a layer consisting of a mixture of clay and peat (“PEAT/CLAY man-made”) is present with a thickness of about 0.5 m in the East on top of the dike reducing in thickness towards the west, the bottom of the dike. This layer is likely a partially man-made top layer to improve the bearing capacity of the surface for the cattle and agricultural equipment (Rupke, 2008). The thickness of the layer is shown in Figure 20.



A sequence of peat and silty or clayey peat layers with some thin silt and clay layers (PEAT7, CLAY5, PEAT6, SILT3, and PEAT5) is present between the man-made top layer and a depth of about – 5 m. In western direction these layers truncate against the man-made top layer.



A fairly consistent clay and clayey peat layer (CLAY4) is present at -5 m.

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

31



Between about -5 and -9.5 to -10.5 a sequence of peat and silty or clayey peat layers with some thin silt and clay layers is present.



At about -9.5 to -10.5 m a slightly undulating boundary marks the start of a sandy sequence (SAND2, LOAM, SAND1).

Figure 18. Tempeldijk-South, 3D subsurface model without man-made peat/clay top layer; based on boreholes and CPTs. Grid: UTM (WGS84, zone 31 NH).

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

32

Figure 19. Tempeldijk-South, plan view of 3D subsurface model without man-made peat/clay top layer. Grid: UTM (WGS84, zone 31 NH).

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

33

Figure 20. Tempeldijk-South, thickness (in meter) contours man-made top layer. (aerial photo: Google Earth/Aerodata International Surveys; image date 2005, 2013) Grid: UTM (WGS84, zone 31 NH).

5.6.4

3D Resistivity survey

A full three-dimensional resistivity survey has been done on the Tempeldijk-South. A brief layout of the equipment and field procedure is included in Appendix 1.2).

5.6.5

Correlation between 3D resistivity Survey and subsurface model at Tempeldijk-South

The resistivity imaging of the subsurface at Tempeldijk-South can fairly accurately be related to the subsurface lithology model (Figure 21). The low resistivity values correlate to peat and clay layers, whereas the high resistivity values correlate to more silty or sandy layers.

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

34

Figure 21. Tempeldijk-South, Reeuwijk, resistivity and (simplified) lithology data: (a) map of resistivity at the surface, where the dashed line indicates where the bottom part of the dike manifests differently from the rest of the dike and with the location of cross-section A-A′ indicated by the solid line; (b) resistivity cross-section A-A′ showing horizontal layering on the right and higher resistivity values on the left (bottom of the dike) with horizontal layering absent; (c) lithology cross-section A-A′ showing horizontal layering of peat, clay and silt, with sand layers starting at about −11 m NAP. Note: the lithology descriptions are simplified, indicating the most important constituents for this research. (after Cundill et al., 2013b).

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

35

5.6.6

Electromagnetic

In cooperation with the Engineering Geology Section of the Delft University of Technology an electromagnetic survey has been done (Ngan-Tillard, 2011). Electromagnetic data is collected using a Geophex GEM2 sensor (for equipment see appendix 1.2). Data is collected in five frequencies, namely 925, 2,175, 9,825, 20,025 and 40,025 Hz. The data is collected in two grids, one parallel to the slope of the dike and the other perpendicular. The interpolated data measured with the 40,025 frequency is shown in Figure 22. The overall pattern illustrated in Figure 22 does not correspond sharply with the other remote sensing measurements. However, there is an anomaly of higher values around location 7 (for location numbers see Figure 23 and Figure 24). In addition, the diagonal pattern from top left to bottom right of lower values in the center of the measurement area is partially visible in the pre-dawn thermal measurements (section 5.6.10), with warmer temperatures for the same area. This could indicate that the deeper subsurface structure is reflected in pre-dawn thermal measurements.

Figure 22. The interpolated values of the electromagnetic frequency 40,025 Hz smoothed measurements. The 54 black dots are the surface measurement locations (see Figure 24).

5.6.7

Surface data

On the Tempeldijk-South, various remotely sensed data have been collected and direct surface measurements have been done. On the surface of Tempeldijk-South data have been collected following the grid in Figure 23 and Figure 24. Data collected per location consists of: 

Surface soil moisture measured with a soil moisture probe (ThetaProbe Soil Moisture Sensor - ML2x).



Green, red and near-infrared images (Tetracam ADC Camera)



Hyperspectral reflectance using an ASD Fieldspec Pro.



Visible light using a Canon EOS camera.



Temperatures using a NEC thermo tracer.

For the equipment used, refer to Apendix 1.1.

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

36

Measurements are recorded at 54 locations on the southern section of the Tempeldijk (Figure 23 and Figure 24). Soil moisture and hyperspectral measurements are point measurements based on equipment limitations. For data comparison, the remaining measurements, which are image based, are averaged per location to obtain a single measurement per location. The electromagnetic data are collected in a grid offset from the other 54 points to avoid contamination of the 54 locations. Gamma ray radioactivity measurements are reported in Hack et al. (2008).

Figure 23. Measurement grid for the surface data measurement locations at Tempeldijk-South site.

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

37

Figure 24. Locations of boreholes (BH), CPTs (S) and locations for surface measurements (r). (aerial photo: Google Earth/Aerodata International Surveys; image date 2005, 2013) Grid: UTM (WGS84, zone 31 NH).

5.6.8

Soil moisture

Volumetric soil moisture is measured using a Thetaprobe (Appendix 1.1.2). Nine readings are taken at intervals during a 24-hour period and averaged per location. The measurements are not taken at exactly the same position but rather distributed within the measurement area covered by the hyperspectral and thermal sensors. This is done to avoid the disturbed soil that results from removing the probe from the soil as well as to reduce the effect of an anomalous patch of soil affecting the overall results. The overall soil moisture values for the dike are low, with some exceptions (Figure 25). The bottom of the dike has higher soil moisture values, with this extending up the dike on the right-hand side of the measurement area. There is a surprisingly low soil moisture value (location 52) at the foot of the dike. When comparing the visible photo for this location with those surrounding it, there is no obvious reason for this (such as exposed soil) visible in the photo (Figure 26). However, if taken into consideration with locations 4 and 16, there appears to be a drier corridor running from the top to the bottom of the dike. The high soil moisture value at the top left (location 7) of the section of dike is unexpected. It would appear that this wetter pattern follows

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

38

meanderingly down the dike surface towards the higher soil moisture values at the bottom left part of the dike section (locations 44 and 46).

Figure 25. Mean soil moisture values (from measurements over a 24-hour period) per location on the background of the results of natural neighbor interpolation (numbers are location numbers).

Figure 26. Photos from the four bottom-right locations (numbers 51, 52, 53, 54 from left to right).

5.6.9

Near-infrared

Images for every location have been taken using the Tetracam agricultural camera (appendix 1.1.3), in the green, red and near-infrared (NIR) wavelengths. Images are recorded at 13h00, 16h 20, and 19h00 on 15 July 2010. The conditions were cloudy to partly cloudy, cloudy, and sunny to partly cloudy respectively. The values per pixel for each image band are averaged to obtain a single value for the location, which is then analyzed together with the other point data. Healthy, lush vegetation reflects strongly in the near infrared portion of the electromagnetic spectrum. From Figure 27, it can be deduced that the bottom of the dike has lusher and healthier vegetation than the top of the dike. Exceptions are found at locations 7 and 15 at the top of the dike, where these locations have high near infrared reflectance values. This overall trend is also observed in the field. Other factors can also result in higher reflectance values in the near infrared. These include lower water content and more standing litter, which both result in higher overall reflectance but with different shaped spectral signatures (Asner, 1998; Jensen, 2007).

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

39

Figure 27. Near infrared (NIR) values for 16h20 measurements on 15 July 2010 per location on the background of the results of natural neighbor interpolation. The data recorded by the camera is in jpg format and is spread over 256 values. In order to reduce illumination effects, the average of the digital numbers (DN) from the image are used in ratio. The following ratios are used: G/R, NIR/G, NIR/R, and the normalized difference vegetation index (NDVI). Based on all ratios, the top of the dike appears to have more variation in the vegetation, with it becoming more consistent toward the bottom. All ratios have a positive trend increasing with distance from the top of the dike. The NIR/G ratio is indicative of stress related changes in chlorophyll concentrations. The reason for this is that NIR is relatively stable for slight changes in chlorophyll concentration but reduction in green reflectance is indicative of chlorophyll reduction (Carter and Knapp, 2001). This would indicate that the grass on the dike has a general trend of increasing levels of chlorophyll concentrations with increasing distance from the top of the dike. Locations 1, 2, 3, 5, 8, 9, 10, 12, 13, 16, 19, and 27 are below average and locations 4 and 50 are above. Visually, there is little difference between locations 49 and 50, and yet they have very different NIR/G values. Most of the locations with considerably lower NIR/G values have a large proportion of dry grass present. Standing litter has a disproportionate effect on grass canopy reflectance, including the NIR (Asner, 1998). The other three ratios show increasing values with increasing distance from the top of the dike towards the bottom. This corresponds with what is visible in the field and in the visual photos taken of each location, namely that the bottom of the dike had more lush and greener vegetation as well as a denser vegetation cover compared to the top of the dike. Figure 28 shows the NDVI values for the 16h20 measurements. There appear to be some anomalies, with values lower than what would be expected for their location on the dike. These are at locations 10, 13, 16, 27 and 40. For the other ratios these locations are also anomalous, with locations 6, 7 and 45 also included. Other locations that appeared to be slightly anomalous are 4, 28, and 50.

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

40

Figure 28. NDVI values for 16h20 measurements on 15 July 2010. Values for the top of dike appear more varied (circled in orange). Some values appear anomalous (circled in red) and are for locations 10, 13, 16, 27, and 40.

5.6.10 Thermal Images for every location are taken using the NEC TH9100 Thermal Tracer camera (for equipment see appendix 1.1.6), which has a spectral range of 8 to 14µm. Images are recorded almost every hour between 12h00 on 15 July 2010 and 12h00 on 16 July 2010. The conditions varied between sunny, partly cloudy, cloudy and dark clouds with rain drops. The values per pixel for each thermal image are averaged to obtain a single value for the location, which could then be analyzed together with the other point data. Complete datasets have been collected for the following hours (rounded to the whole hour): 12h00, 14h00, 15h00, 16h00, 17h00, 18h00, 20h00, 23h00, 00h00, 03h00, 04h00, 05h 00, and 10h00. 5.6.10.1 Temperature related to cover Considering that the surface of a dike should be fairly uniformly covered in grass, it would be expected that the temperatures should be fairly consistent. However, if there were locations where the cover deviates, it would be expected that this would be reflected in the temperature values. In addition, water availability affects canopy temperature, thus variations in available water should also be reflected in temperature values. In order to visualize the variation of temperatures of each location per time of measurement, the temperatures are plotted on a scatterplot per location for each hourly set of measurements. Outliers are visually identified. Outliers may occur because of underlying vegetation cover, structural or moisture components. They could also be a result of shadowing, illumination, or weather effects. To try to reduce these latter effects, the frequency with which a location is considered an outlier is investigated. The location most frequently identified as an outlier is location 7, with the temperature consistently lower than average during the day and higher than average pre-dawn. Other locations identified as frequent outliers are 1, 16, 45, 46, 2, 19, 25, 35, 24, 42, and 44. During the day, locations 46, 24, 42, and 44 are lower than average and locations 1, 16, 45, 2, 19, 25, and 35 are higher than average. In general, the lower temperatures correspond with more lush vegetation and the higher temperatures with drier patches of vegetation with standing litter (Figure 29 top). However, in some cases there does not appear to be much difference in vegetation cover between two locations and yet one has been identified as an outlier and another not (Figure 29 bottom).

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

41

Figure 29. Top: photographs illustrating the condition of the grass at time of measurement which corresponds to the temperatures measured. The left photograph is for location 1 with some dry patches, which correspond with higher temperatures. The right photograph is of location 46 with lush vegetation, with lower temperatures. Bottom: photographs illustrating the condition of the grass at time of measurement, which do not appear to correspond to the temperatures measured. The left photograph is for location 35, which has a higher than average temperature which could be related to dryness or more standing litter. The right photograph is for location 36, which does not have anomalous temperatures but which does not appear much different from location 35. Each location’s measurements are plotted on a line graph per time of measurement. These are visually inspected for crossovers or generally unusual curves. All outliers listed above are also found in this analysis, with the exception of location 2. In addition, the following locations had anomalies in their curves: 9, 10, 12, 13, 17, 18, 27, 28, 34, 40, 43, and 47. 5.6.10.2 Temperature v. time To investigate the effect of the time of measurement on the temperatures measured, a selection of the complete datasets distributed over the 24 hour period are chosen, namely 15h00, 17h00, 20h00, 23h00, 05h00 and 10h00. These are plotted on a graph, based on location (Figure 30). All datasets show a linear trend with higher temperatures at top the top of the dike and lower values towards the bottom of the dike, except for the pre-dawn measurements (namely 03h00, 04h 00, and 05h00) which show almost no trend. For some locations, their temperature relative to the trend reverses between daytime and pre-dawn measurements. This would imply that for these locations, soil water content is the overriding factor affecting temperature. Other factors that could affect temperature include variation in vegetation density, standing litter or the presence of bare soil. During the daylight hours, some common features are apparent from the graph (Figure 30), although there is considerable variation between the datasets. For example, location 7 has a low relative temperature during daylight hours.

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

42

Figure 30. Comparison of hourly temperatures per location. To reduce the effect of variations in sunlight intensity and weather conditions, all the data is normalized by subtracting the mean of the values per period of measurement. These normalized values are plotted on a line graph to evaluate the trends in the data (Figure 31). As illustrated in Figure 31, locations 7, 24, 46 and 50 have lower day temperatures than the means, with night temperatures approaching the means or rising above. Locations 16, 19 and 27 have lower daytime temperatures, with their night temperatures approaching the means or dropping below. There was a rain event around 16h00, and as expected, the measurements during, and shortly afterwards show distinct deviations from the overall trend. In addition, for some locations, the late afternoon measurements already show alterations in their trends towards the post-sunset values. It is thus decided to focus on the 15h00 measurements as representative for maximal illumination and emittance (Figure 32). The 04h00 measurements are selected as representative for minimal emittance (Figure 33).

Figure 31. Normalized thermal values for selected locations. Apparent sunrise: 05h46, apparent solar noon: 13h38, apparent sunset: 21h30 (Earth System Research Laboratory, 2013). Rain event (in grey) occurred around 16h00. 5.6.10.3 Spatial distribution of temperature The 15h00 measurements (Figure 32) show an overall trend with the bottom of the dike being cooler than the middle and top. Location 7 is considerably cooler than the surrounding locations, with a cooler trend of measurements running from location 7 at the top of the dike to location 47 at the bottom of the dike. In addition, there appears to be a band of cooler temperatures running along the dike in line with the second row of locations (i.e. locations 10-18). There is a grouping of warmer measurements around locations 33 and 34. In the 04h00 measurements (Figure 33) the overall trend is reversed with the top of the dike being cooler than REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

43

the bottom. However, the more detailed patterns are not repeated in the night measurements. Location 7 is warmer than immediately surrounding values but not strongly. The strong banding between the middle and bottom of the dike is not apparent. Location 24 is warm, as is location 1 and 38. There is group of cooler temperatures around location 20.

Figure 32. Measured temperatures at 15h00 per location on the background of the results of natural neighbor interpolation.

Figure 33. Measured temperatures at 04h00 per location on the background of the results of natural neighbor interpolation. 5.6.10.4 Spatial distribution of average temperatures In order to investigate the variation of the thermal spatial pattern of a single set of measurements compared with the overall trend for specific periods of the day and night, means are calculated for two and three measurements per location (Figure 34b: 14h00 and 15h00; Figure 34c: 14h00, 15h00, and 18h00; Figure 34e: 03h00 and 04h00; and Figure 34f: 03h00, 04h00, and 05h00). The overall trend remains the same for both day and night. In addition, the stronger patterns also remain and are even strengthened. From this, it can be concluded that a single time measurement should be sufficiently representative if not taken within a transition period of time. REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

44

Figure 34. Top: comparison of single, two and three temperature measurements during the day: a) 15h00; b) 14h00 & 15h00; and c) 14h00, 15h00 & 18h00; and bottom during the night: d) 04h00; e) 03h00 & 04h00; and f) 03h00, 04h00& 05h00. Temperatures are per location on the background of the results of natural neighbor interpolation.

5.6.11 Visible light Visible light digital photographs, for both orientation and analysis, are taken in RAW format using a Canon EOS 400D with a zoom lens EF-S 17-85mm 1:4-5.6 IS USM (for equipment see Appendix 1.1.5). The photos are taken at around 18h30 on 15 July 2010 with sunny conditions with occasional cloud patches. The photos are taken from below left of the location (north-west of the location). The photos are oblique for orientation and included areas outside of the point location. The images are thus clipped to represent the location area. These clipped images are then used for analysis. The images are first converted from RAW format to 16bit unsigned TIFF files for ease of use but still maintaining radiometric resolution. The images have three layers recording data in the red, green, and blue wavelengths respectively. The values per pixel for each image layer are averaged to obtain a single value per layer for the location, which could then be analyzed together with the other point data. 5.6.11.1 Vegetation v. brightness The values for the three bands are added together to analyze the overall brightness of the locations (Figure 35). An increase in brightness can be attributed to increased standing litter, a decrease in Mean Leaf incident Angle (MLA) (Asner, 1998), water deficit stress (Blum, 2010; Carter, 1991; Jensen, 2007; Nowatzki et al., 2004) or a decrease in chlorophyll concentration (Carter and Knapp, 2001). There is an overall decrease in brightness from the top of the dike towards the bottom. This may be due to a higher presence of dry or yellow grass nearer the top of the dike, with little or none visible at the bottom (Figure 29). Locations 11, 14, 16, 27 and 41 have brightness values higher than average. Locations 21, 32, 35, 39, 47, 49, 50, 51 and 52 have values lower than average. Both the blue and red bands also show an overall trend of decreasing values from the top of the dike to the bottom. However, the green band shows no REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

45

overall trend (Figure 35). The decrease in the red but not in the green could be indicative of increasing leaf area index (LAI) (Asner, 1998) or water deficit stress (Kim et al., 2010). In the red band, locations 16 and 27 are higher than average and locations 49, 50 and 52 are lower than average. In the green band, locations 7, 11, 14, 41 and 54 are higher than average and locations 21, 32, 35 and 50 are lower than average. In the blue band, locations 11, 16, 27 and 36 are higher than average and locations 21, 32, 35, 39, 47, 49, 50, 51 and 52 are lower than average.

Figure 35. Digital number (DN) values for red, green, and blue bands plus the overall brightness (R+G+B) for the color images per location. 5.6.11.2 Normalized values The values are normalized by dividing the value per layer by the sum of all layers for each location to reduce the effects of illumination differences. These normalized values are then graphed on a scatter plot based on location (Figure 36). The red band shows an overall decrease in values from the top of the dike towards the bottom. The green band shows a reversed trend with lower values at the top of the dike and higher values at the bottom, which would correspond with the overall visual appearance that the grass becomes greener and lusher with distance from the top of the dike. This corresponds with the trend that is found in the NIR data. The blue band is fairly constant over the profile of the dike. Locations 7 and 27 are outliers from the surrounding locations in both the red and green bands, with location 6 also being an outlier in the green band and location 4 an outlier in the blue band. When compared to the red band average, locations 1, 2, 3, 5, 10, 16 and 27 are above and locations 46, 47, 49, 50, 52 and 54 are below average. For the green band, locations 46 to 52 and 54 are above average and locations 1, 2, 3, 10, 16, 19 and 27 are below. In the blue band, only location 4 is above average.

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

46

Figure 36. Normalized values for the red, green and blue bands of the color images per location (normalization is done by dividing the band value by the sum of all bands). 5.6.11.3 Correlation between value and location Regardless of whether the data is normalized or not, the pattern of green versus red is consistent per location. This is most likely due to these colors being predominant in the landscape with blue being supplementary. The lush grass has a strong reflectance in the green, whereas bare soil has lower green but higher red reflectance giving it its brown appearance. Dry grass has nearly equal reflectance in the green and red, which corresponds with its yellow appearance. The results of the green/red ratio is shown in Figure 37. Overall, the predominant color of the vegetation on the dike tends towards yellow, indicating the presence of standing litter and dry grass. As mentioned previously, standing litter has a disproportionate effect on grass canopy reflectance (Asner, 1998). The bottom of the dike has higher G/R values that correspond with the lush green appearance of the grass at the bottom of the dike. Location 45 has a noticeably lower value compared to the surrounding locations, which corresponds with what has been seen in the field; with location 45 having patches of dry grass (Figure 38a-b). Locations 7, 28 and 29 have noticeably higher values compared to the surrounding locations. Figure 38c-d show the difference in appearance of locations 7 and 8. This corresponds with trend of the G/R values from the Tetracam agricultural camera. However, the anomalous locations differ with only locations 7 and 28 being in common. In addition, lower G/R values could be indicative of the factors that result in an increase in overall brightness mentioned earlier in this section. Plant stress, in general, reduces chlorophyll concentration (Blackburn, 2007; Carter and Knapp, 2001; Carter and Spiering, 2002) and chlorophyll concentration is strongly correlated to reflectance in the green wavelengths (Carter and Knapp, 2001; Jensen, 2007). Carter and Knapp (2001) found that as chlorophyll concentration increases so the spectral reflectance at 683nm (red limit) decreases.

Figure 37. Green/red ratio values plotted against location. The values around 1 correspond with a yellow color. Higher values move towards green and lower values move towards brown and ultimately red appearances.

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

47

Figure 38. Photographs illustrating the color of the grass at time of measurement. a) Location 45 shows patches of dry grass; b) location 46 shows only lush green grass; c) location 7 has more green grass compared with neighboring location 8 (d) which shows considerably more dry grass. The G/R values are interpolated and visualized spatially (Figure 39). The bottom of the dike is greener than the top with the bottom left of the study area also green towards the middle of the dike. In addition, location 7 is significantly greener than the surrounding values and location 15 is slightly greener.

(a) (c)

Figure 39. The green/red ratio values per location on the background of the results of natural neighbor interpolation.

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

48

5.6.12 Hyperspectral Reflectance spectra for each location are measured using the ASD Fieldspec Pro spectrometer (for equipment see appendix 1.1.4), which has a spectral range of 350 – 2500nm. An 8° fore-optic is used at a height of about 75 cm. The sample count is 25 and three samples are collected as each location per time. The instrument is calibrated at the start of each measurement set and additionally if the illumination conditions changed substantially. Four sets of measurements are taken, namely 13h00, 16h00, and 17h00 on 15 July 2010 and 10h00 on 16 July 2010. The weather conditions are cloudy with strong wind, dark cloud cover including rain, mostly sunny and mostly cloudy with wind respectively. For each data set, the three measurements per location are averaged to form the input datasets for analysis. 5.6.12.1 Spectral signature Two advantages of using hyperspectral data are that the shape of the spectral signature can be examined and that unique spectral features in non-traditional wavelengths or narrow bands can be discovered. It is difficult to inspect 54 spectra visually simultaneously, so the Spectral Angle Mapper (SAM) method, a standard spectral analysis technique, is used to identify the locations that differ the most spectrally. SAM requires an end member with which to compare the other spectra. Using the results from the analysis of the soil moisture, NDVI, thermal, brightness, and G/R ratio data, locations 7, 46, 3, and 16 have been selected as potential end members based on the frequency for which they are considered outliers. Location pairs 7 & 46 and 3 & 16 display similar results. Since all the spectra are of vegetation, the differences in spectral signatures are fairly subtle. Based on the results of the SAM analysis, locations 3, 27, and 46 have been selected as spectrally representative end members for the 19h00 dataset. These spectra have been visually inspected for features and their overall shape or signature have been noted (Figure 40). Location 46 is representative of lush, green vegetation (Asner, 1998; Carter and Knapp, 2001), where the reflectance in the visible wavelengths is quite low (with a small peak in the green) and the reflectance in the NIR is comparatively high. The spectral signature for location 3 shows overall much higher reflectance and the shape is quite different from location 46. The shape of location 3’s spectral signature corresponds with an approximate amount of 40% standing litter (Asner, 1998). This is confirmed with the inspection of the photograph for location 3 (Figure 41). The shape of the spectral signature for location 27 would indicate low amounts of healthy vegetation. The flattened shape and the low overall reflectance is probably a result of the reflectance from the dark, organic soil. Indeed, the examination of the photograph for location 27 (Figure 41) shows that there is very little green vegetation, and substantial amounts of bare soil visible together with dry grass.

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

49

Figure 40. Spectral signatures for selected locations from the 19h00 dataset. Strong atmospheric water absorption near 1400nm and 1900nm prevented measurements in these spectral regions.

Figure 41. Photographs illustrating the grass cover composition at the time of measurement. a) Location 3 shows healthy grass with large quantities of dry grass; b) location 27 shows little healthy grass but large quantities of dry grass and bare soil and c) location 46 which shows only healthy grass. Based on the literature and actual differences visible in the spectra, the following wavelengths are identified as potentially of interest: •

450 nm (blue)

•

550 nm (green)

•

660 nm to 680 nm (red)

•

700 nm (red edge)

•

750 nm (red edge)

•

830 nm (NIR)

•

900 nm (NIR)

•

950 nm (NIR)

•

970 nm (NIR)

•

1076 nm (NIR) REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

50

•

1122 nm (NIR)

•

1155 nm (NIR)

•

1190 nm (NIR)

•

1280 nm (NIR)

•

1350 nm (SWIR)

•

1650 nm (SWIR)

•

2200 nm (SWIR)

5.7

REEUWIJK, TEMPELDIJK-NORTH

Tempeldijk-North site is chosen as reference. The dike seems to function without known problems. Also on the surface of the dike no features have been distinguished that may indicate seepage (‘kwel’), subsidence’, or otherwise features that could be an indication of “problems”.

5.7.1

Subsurface investigation - Dutch Cone Penetration Tests with gas measurement

At the Tempeldijk-North site, three (Dutch) Cone Penetration (CPT) tests with gas measurement (MIP) have been done (Zwang and Monden, 2012). Methane gas in the subsoil can be detected with MIP technology. The locations of the CPT measurements are shown in Figure 42. The measurements are included in the Flood Control 2015 project deliverable “2012.11.01.5; Remote sensing for dike strength: DIS module for secondary levees”.

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

51

Figure 42. Tempeldijk-North test site area with CPT locations (MIP1-3). (photo Google Earth/Aerodata International Surveys, 2009) (Grid: UTM (WGS84, zone 31 NH).

5.7.2

Subsurface model

The CPT measurements have been interpolated following the methodology described in Zwang and Monden (2012) and a subsurface section has been made following the methodology also used for the TempeldijkREMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

52

South site (chapter 5.6.2.1). The profile along the three CPTs (Figure 43) shows a lithology roughly similar to the Tempeldijk-South.

Figure 43. West-East simplified profile along CPT locations (see Figure 42 for locations; depth is in meter below NAP/m.s.l.; depth indication of CPTs is meter below surface).

5.7.3

Summary Tempeldijk-North

Tempeldijk-North some high concentrations of methane gas have been detected in the peat layers with the MIP-CPT technique. The measurements, interpretation, and detailed discussion of the results are included in the Flood Control 2015 deliverable “2012.11.01.5; Remote sensing for dike strength: DIS module for secondary levees”. The effect gas bubbles in the subsoil have on the electrical conductivity is an increase in resistance. Gas saturated water does not cause an increase in resistance. Hence, possibly areas with subsurface gas bubbles may be identified with resistivity surveys.

5.8

DIKE COVER QUALITY

The research done in this project specifically directed to the quality of the dike cover is extensively reported in an article to be published (Cundill et al., 2013b). The main findings relating to dike cover quality are summarized below. Nine damage parameters are evaluated by dike inspectors to establish the quality of grass covered dikes (Bakkenist et al., 2012b). Six of these relate, either entirely or partially, to the quality of the grass covering. Two of the damage parameters relate to the moisture content of the dike, either to wet patches or to dryness. Thus two key criteria that dike inspectors consider when assessing grass covered dikes are the moisture content of the dike and the quality of the dike covering (Moser and Zomer, 2006; Van Baars and Van Kempen, 2009). The quality of the dike covering is important for resistance against erosion and water infiltration in the case of overtopping. In addition, a grass covering can also be influenced by processes taking place within the dike structure, such as seepage or cracking. If remote sensing data can be used as a proxy for dike covering quality, then a faster inspection process for this indicator is possible without limitations on accessibility. The quality of the dike covering for grass-covered dikes is assessed by evaluating the health of the vegetation and the presence of standing litter (dead plant material), flotsam (floating debris), bare soil, and weeds. Multispectral, and more recently hyperspectral, remote sensing is widely used in vegetation studies (Blackburn, 2007; Lillesand et al., 2008; Van der Meer et al., 2006). Vegetation health can be assessed by the amount of biomass and is influenced by water availability, the presence of diseases and nutrient availability. The multispectral based Normalized Difference Vegetation Index (NDVI) is developed and is still used for biomass estimation (Broge and Leblanc, 2001; Rouse et al., 1974). Hyperspectral remote sensing is also found to be useful for the estimation of biomass (Broge and Leblanc, 2001; Darvishzadeh et al., 2009; Haboudane et al., 2004). Remote sensing is researched extensively for assessing vegetation water content

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

53

and plant stress induced by water deficit (Caccamo et al., 2011; Chen et al., 2005; Fensholt and Sandholt, 2003; Hunt and Rock, 1989; Kim et al., 2010). In this project, relationships are studied between 1) soil moisture content and 2) cover quality, and a) thermal, b) broadband multispectral and c) narrowband hyperspectral remote sensing measurements for grass-covered dikes. The study makes use of ground-based, hand-held remote sensing sensors. Moreover, the findings are compared with the subsurface composition of the dike to assist in the understanding of processes that could be manifesting at the surface.

5.8.1

Dike cover quality classes

The quality of the dike covering is evaluated for each location for the areal extent of the remote sensing data. Evaluation criteria include grass density, canopy cover and the presence and quantity of standing litter (dead plant material), flotsam (floating debris), weeds, and bare soil. The grass cover quality for each location is assessed and allocated to either a good, medium, poor or bad quality class (based on the current Dutch dike assessment classes, Bakkenist et al., 2012a). The assessment criteria for the different classes are shown in Table 4. Table 4. Cover quality classes and assessment criteria used for this study (modified after Bakkenist et al., 2012a).

5.8.2

Cover Quality and Thermal Remote Sensing

Only moderate correlations between the thermal remote sensing data and cover quality have been found. A likely explanation is that a single cover quality class is often comprised of several cover types, such as green vegetation, standing litter (dry vegetation) and bare soil, in varying proportions. As discussed before (section 5.8), these various cover types have different thermal responses. Thus, the varying proportions of cover types within a single class could account for some of the differences between the thermal remote sensing data and the cover quality. A more detailed classification system could improve the strength of the relationships.

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

54

5.8.3

Cover Quality and Multispectral Remote Sensing

The multispectral remote sensing shows the strongest correlations to cover quality compared to the other remote sensing datasets. This is consistent with what Broge and Leblanc (2001) found, in that broadband indices performed better than hyperspectral narrowband indices for gauging green leaf area index and canopy chlorophyll density. Both the NIR/R and the NIR/G ratios have moderate correlations to cover quality. For these datasets, the NIR measurements are the main contributors, with measurements in the R and G being similar in magnitude. Energy in the NIR is strongly reflected from green vegetation (Jensen, 2007; Lillesand et al., 2008) and is affected by leaf area, bare soil and standing litter (Asner, 1998). Therefore, these ratios are suitable for cover quality evaluation. However, the cover quality data is ordinal and the ratio data is continuous which may account for the relationship being only moderate.

5.8.4

Cover Quality and Hyperspectral Remote Sensing

Significant relationships are found between the hyperspectral remote sensing data and the cover quality, and are also discernible in the spatial distribution of these datasets. The strongest relationship is between the WBI and cover quality (Table 5 and Table 6). This index has been developed for use as an indicator for leaf and canopy water content (Peñuelas et al., 1993) based on a water absorption feature around 970 nm, and the absence of it at 900 nm. Reflection in the 970 nm and 900 nm wavelengths is also affected by low leaf area, standing litter and bare soil (as observed by Asner, 1998), all of which increase the index value. Peñuelas et al. (1993) also state that the index does not perform well with changes in leaf area and if the vegetation does not completely cover the soil. The presence of standing litter and bare soil strongly affect values for this index and dominate over the subtler changes in absorption feature depth that result from water deficit stress in green vegetation. With the cover quality largely being medium to poor, this implies that substantial amounts of standing litter and bare soil are present. This would explain why, in this case, this index works well for cover quality but not for soil moisture. Table 5. Hyperspectral indices (after Cundill et al., 2013b).

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

55

Table 6. Correlation coefficients for selected datasets (after Cundill et al., 2013b).

The next strongest correlations for hyperspectral data and cover quality are for the two carotenoid reflectance indices, CRI700 and CRI550. Although these indices are developed for assessing carotenoid content in plants, reflectance in the wavelengths 700 nm, 550 nm and 510 nm is strongly affected by the presence of standing litter and bare soil (as observed by Asner, 1998). This supports that these indices are suitable for assessing cover quality, which is to a large degree determined by the presence of standing litter and bare soil. For many of the remote sensing measurements, the signal collected is from a mixture of cover components. These components do not affect the signal linearly. In particular, a small amount of standing litter has an inordinate effect on the overall reflectance from the dike cover (Asner, 1998). This, together with the cover quality data being ordinal, could account for the correlations only being moderate between the remote sensing data and the cover quality data. Using a more detailed classification system for cover quality may improve these correlations. Further, for the hyperspectral data, the collection of end-member spectral signatures for the specific components of the dike cover and the use of these in spectral analysis could also be useful for determining proportions of the these components. This would further improve the characterization of the dike covering.

5.9

DISCUSSION & CONCLUSIONS REEUWIJK LOCATION

The results of the various analyses for the Reeuwijk location are summarized below.

5.9.1

Analyses of relevance of developed methodology

There is clearly useful information to be obtained from remotely sensed data for dike inspection. Information obtained about the vegetation on a dike reveals information not only about the condition of the cover itself (important for dike quality) but also about the condition of the subsurface of the dike, including soil moisture and subsurface structure. This study has revealed which information could be used as well as where further investigation is required. In addition, limitations and considerations have also been identified. Near-infrared remote sensing showed higher values at the bottom of the dike compared to the top. The daytime thermal measurements showed lower values at the bottom of the dike compared to the top. The visible green to red ratio showed greener vegetation at the bottom of the dike compared to the more yellow vegetation at the top of the dike. The brightness values from the visible showed darker values for the bottom of the dike compared to the top. These patterns corresponds well with the soil moisture measurements which showed higher moisture content at the bottom of the dike compared to lower values at the top. Additionally, anomalies could also be detected in the datasets, for example at location 7.

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

56

5.9.2

Identification of properties and characteristics of dikes - key wavelengths

Further, is shown that ratios can reveal additional information and patterns in the data, and are particularly useful for reducing the effects of illumination intensity differences. However, as has been noted, the reflectance intensities themselves often offer much information. The brightness and shape of the spectral signature are affected by standing litter, mean leaf angle, water deficit stress or chlorophyll concentration. Hyperspectral data is invaluable for this application. However, it is not always practical to collect hyperspectral data. It results in large quantities of data. In addition, sensors are often expensive. Thus, key wavelengths are identified. These include specific wavelengths in the blue, green, red, near infrared, and short-wave infrared (see the Hyperspectral section).

5.9.3

Relation remote sensing and covering characteristics dikes

Near-infrared wavelengths are affected by a number of factors (such as vegetation moisture content, leaf area index, standing litter, and total biomass). Data may be required from other wavelengths to clarify which factors are involved. Reflectance in the near infrared is insensitive to small changes in chlorophyll concentration whereas the green wavelengths are strongly affected. Stable green reflectance but decreasing reflectance in the red wavelengths is indicative of increasing leaf area index or decreasing water deficit stress. Increases in reflectance for all wavelengths and especially a strong increase in the red are indicative of standing litter. High reflectance in the short-wave infrared is indicative of dryer vegetation.

5.9.4

Relations dike quality, remote sensing and subsurface data

The shallow information obtained from the higher electromagnetic frequencies reveals information about the subsurface and correlates with the pre-dawn thermal measurements. Therefore, pre-dawn thermal measurements appear to be related to the subsurface structure. This requires further investigation. The daytime thermal measurements are correlated to vegetation condition and soil moisture.

5.9.5

Weather influence

For thermal measurements, local weather conditions are clearly important. As detected in the data collected, a single rain shower completely altered the temperatures. When making thermal observations, special note must be made of local wind conditions, precipitation, and cloud cover. In addition, time of observation is important. In this study, it is found that post-solar noon measurements or pre-dawn measurements provided more information for the thermal measurements. Shading also has a very strong influence on thermal measurements, reducing the apparent temperature significantly. Similarly, for wavelengths from the blue to short-wave infrared illumination has a strong influence on the data obtained. Direct sunlight with clear skies is optimum. In addition to measurement conditions, instrument selection is crucial. Sensor sensitivity, radiometric and spatial resolution, wavelength observed and width of wavelength channel are important considerations.

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

57

6 AMSTERDAM “STOWA DROOGTEPROEF” LOCATION 6.1

INTRODUCTION

University of Twente, Faculty of Geo-Information Science and Earth Observation (ITC), collected data at the Veenderij, “STOWA Droogteproef” location in Amsterdam (Figure 44 & Figure 45) for evaluating the options to identify the state of moisture content in a peat dike. The tests done are part of a research towards establishing the possibilities to determine the quality of secondary dikes, in particular peat dikes (“Boezem kades”) with remote sensing. This research is executed in the context of the RSDYK, Flood Control 2015 project. Only one fieldwork campaign has been executed because the weather during 2012 was such that the conditions of the dike with respect to moisture content in the dike material did not change. This also limited the options to analyze the data because data of a dry dike could not be collected and are thus not available for comparison.

Figure 44. Test site location Amsterdam (STOWA Droogteproef location) (Google Maps, 2012).

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

58

Figure 45. Test site location Amsterdam (STOWA Droogteproef location). The urban area on the left bottom is Ouderkerk aan de Amstel (aerial photo: Google Earth/Aerodata International Surveys; image date 2008, 2013) Grid: UTM (WGS84, zone 31 NH).

6.2

DATA COLLECTION

The data collection and field campaign took place on 17 and 18 August 2012.

6.3

DATA COLLECTION REQUIREMENTS

Since the data collection is part of an initial research, the measurements are ground based and thus the collection time is longer than it would be at an operational stage. Therefore, the requirements for collection of the data are clear skies. The first opportunity on which the skies were relatively clear, and that equipment and staff were both available was on 17 and 18 August 2012. This data set is used as a base data set for normal conditions. The intended data collection could not be performed during a relatively dry period because the weather during 2012 was fairly wet, and the test site did not become very dry.

6.4

DATA SETS AND QUALITY

Data has been collected for 110 points in the dike study area of the Droogteproef site, but only 109 locations are used as location 25-02 is excluded because of cables present at this test location. 10 collection lines running from the bottom of the dike to the crest have been measured. The lines are 5 m apart. Each line consists of 11 points with a 2.5 m spacing (Figure 46 and Appendix 2 : Table 7 for layout and coordinates).

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

59

Figure 46. Measurement Amsterdam Location.

6.5

TYPE OF DATA

Data collected per location: 

Surface soil moisture measured with a soil moisture probe (Field Scout TDR 100, 2013).



Hyperspectral reflectance using an ASD Fieldspec Pro (Fieldspec Pro, 2013) at about 12h30 on 18th.



Visible light using a Canon EOS camera at about 14h00 (Canon EOS 400D, 2013).

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

60



Temperatures using a NEC thermo tracer (NEC TH9100, 2013) at 13h00 and 16h00 on the 17th and 05h00, 10h00 and 14h00 on the 18th

For the equipment used to obtain surface measurements is referred to Appendix 1 : .

6.6

EVALUATION

Only one (the “base”) dataset could be collected. A data set under dry condition could not be collected because of the weather during 2012. Comparison of data sets of wet and dry conditions could thus not be done. Data sets of other parties only were received in November 2012. The time frame until this report was too short to make an evaluation and comparison with the data from other parties.

6.6.1

Canon visible-light images

Figure 47 and Figure 48 are examples of images in visible light taken with a normal camera (Canon EOS 400D, 2013).

Figure 47. Canon visible-light images of location 00-01.

Figure 48. Canon visible-light images of location 00-09.

6.6.2

NEC Thermo Tracer images

Figure 49 and Figure 50 are examples of images made with the NEC Thermo Tracer (NEC TH9100, 2013). The color differences indicate the temperature, with red a relative higher and blue lower temperature.

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

61

Figure 49. NEC thermo tracer image of location 00-01.

Figure 50. NEC thermo tracer image of location 00-09.

6.6.3

ASD Fieldspec Pro spectrometer

Figure 51 is an example of data obtained by the ASD Fieldspec Pro spectrometer (Fieldspec Pro, 2013) for locations 00-01 and 00-09.

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

62

Figure 51. ASD Fieldspec Pro spectrometer for locations 00-01 and 00-09.

6.7

DISCUSSION

Only one data set could be obtained and no data under dry conditions because of the weather during 2012. Comparison of data sets under wet and dry conditions could therefore not be done.

6.8

CONCLUSIONS AMSTERDAM LOCATION

The purpose of the data collection is a comparison between remote sensing and ground data collected during wet and dry conditions of the site. That has not been achieved because of the weather conditions. In most years, the weather is such that long dry periods occur during the year and therefore it is expected that during 2013 weather conditions are such that data under dry conditions can be collected as it is unlikely that 2 successive years without dry periods occur. Speculatively, based on experience with Reeuwijk Tempeldijk peat dike (see section 5), it is likely that the remote sensing data can be correlated with the moisture content in the material at the dike surface and probably indicates subsurface inhomogeneity. However, the subsurface and processes in the Tempeldijk with respect to moisture content are different (Cundill et al., 2013a; Hack et al., 2008).

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

63

7 COMMUNICATION; PRESENTATIONS AND ARTICLES In the context of this project various presentations haven been given and articles and book chapters published. The published articles could not be incorporated in this report because of copyright issues. For the articles refer to the original publication media.

Articles published: Cundill, S.L., Hack, H.R.G.K., Van der Meijde, M., Van der Schrier, J.S., Ngan-Tillard, D.J.M., 2012a. Quality of Peat Dikes Evaluated by Remote Sensing. In: Klijn, F., Schweckendiek, T. (Eds), Comprehensive Flood Risk Management: Research for Policy and Practice. CRC Press, Leiden. ISBN: 978-0415621441. pp. CDRom. (Cundill et al., 2012b) Cundill, S.L., Hack, H.R.G.K., Van der Meijde, M., Van der Schrier, J.S., Ngan-Tillard, D.J.M., 2013. Potential of Using Remote Sensing Data for Dike Inspection. In: Huang, Y., Wu, F., Shi, Z., Ye, B. (Eds), New Frontiers in Engineering Geology and the Environment. Springer, Berlin, Heidelberg. 9. ISBN: 978-3-642-31671-5. pp. 203-206. (Cundill et al., 2013a)

Article in review: A further article is in the process of being reviewed: Cundill, S.L., Van der Meijde, M., Hack, H.R.G.K., 2013b. Investigation of Remote Sensing for Potential Use in Dike Inspection. Journal of Selected Topics in Applied Earth Observations and Remote Sensing (IEEE). (in review). (Cundill et al., 2013b)

Presentations & extended abstract/poster: Cundill, S.L., Hack, H.R.G.K., Van der Meijde, M., Van der Schrier, J.S., Ngan-Tillard, D.J.M., 2012b. Remote Sensing Data for Use in Regional Dike Inspection. In: Nederlands Aardwetenschappelijk Congres 11, Koningshof, Veldhoven, The Netherlands, 29-30 March 2012, p. 21 slides. (Cundill et al., 2012c) Cundill, S.L., Hack, H.R.G.K., Van der Meijde, M., 2012. Investigation of Remote Sensing for Dike Inspection. In: IEEE International Geoscience and Remote Sensing Symposium (IGARSS 2012) : Remote sensing for a dynamic earth, Munich, Germany, 22-27 July 2012. IEEE, p. 1. (Cundill et al., 2012a)

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

64

8 CONCLUSIONS The lateral and vertical variations as well as the heterogeneity of dike material are obvious from geological and geophysical investigations. Many of the remote sensing data sets, such as daytime thermal infrared, multispectral NIR and visible G/R ratio, possibly show some evidence of a layered structure that maybe reflects the subsurface structure of the dike. The layered structure is detectible likely because the water content varies from layer to layer and this influences the reflection of both the soil and the vegetation. This is further confirmed by the soil moisture data that shows the same evidence of a layered structure. However, not all the variation in these data sets can be explained by the subsurface structure. Many anomalies in the remote sensing data seem to be directly related to anomalies in the soil moisture. These may be the result of a subsurface process (not related to structure), localized soil compositional differences, or some surface influence. Further, the relationship between the pre-dawn thermal infrared images and the high frequency electromagnetic measurements would indicate that these remote sensing measurements also reflect the subsurface. Certain remote sensing data sets also proved to be useful for characterizing the condition of the grass covering the dike. These included the visible G/R ration, the hyperspectral signatures, and the brightness values from the visible imagery. Relationships between various remote sensing data sets, the subsurface and weak areas have been established. The relationships, although not strong, do highlight potentially problematic areas where more in-depth investigation should be focused. It is likely that by combining the different remote sensing techniques that the relationship to a subsurface feature will be strengthened. The research in this project is not yet complete. The PhD research continues for another 12 months until spring 2014. It is likely that results that are more conclusive can be presented in the PhD thesis.

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

65

APPENDIX 1 : EQUIPMENT 1.1

GROUND SURFACE DATA SENSOR EQUIPMENT

1.1.1

Fieldscout TDR 100

FieldScout TDR 100 Soil Moisture Sensor (Field Scout TDR 100, 2013) is used to determine surface moisture content based on time-domain measurement technology. The portable TDR 100 accurately measures soil moisture across the full range of soil moisture conditions. The LCD interface provides two modes: volumetric water content and relative water content (irrigation management) mode.

Figure 52. FieldScout TDR 100 Soil Moisture Sensor (photo: Field Scout TDR 100, 2013)

1.1.2

ThetaProbe Soil Moisture Sensor - ML2x

ThetaProbe Soil Moisture Sensor - ML2x (Figure 53) (ThetaProbe, 2013) is used to determine the soil moisture volume percentage by applying frequency-domain techniques. For details on the operation refer to ThetaProbe (2013).

Figure 53. ThetaProbe Soil Moisture Sensor - ML2x (photo: ThetaProbe, 2013).

1.1.3

Tetracam Agricultural Digital Camera (ADC)

Tetracam Agricultural Digital Camera (ADC) (Tetracam, 2013) contains a Motorola CMOS image sensor (storing 1280 x 1024 pixels). The camera records images in the JPG format for overlapping broad bands in the green (G), red (R) and near infrared (NIR) wavelengths.

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

66

Figure 54. Tetracam Agricultural Digital Camera (ADC) (photo: Tetracam, 2013)

1.1.4

ASD Fieldspec Pro spectrometer

Hyperspectral reflection coefficient spectra for each location has been determined with the (mobile) ASD Fieldspec Pro spectrometer (Figure 55, Figure 56) (Fieldspec Pro, 2013). This spectrometer has a spectral range of 350-2500 nm. An 8° fore-optic is used at a height of about 75 cm above ground surface to obtain data of about the same area as the other image-based remote sensing instruments used.

Figure 55. ASD Fieldspec Pro spectrometer (photo: Fieldspec Pro, 2013).

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

67

Figure 56. ASD Fieldspec Pro spectrometer in the field.

1.1.5

Canon EOS 400D

Photos of each location are taken with a (standard) visible-light digital Canon EOS 400D camera (Figure 57) (Canon EOS 400D, 2013). The camera stores the images in jpeg and RAW formats (12 bits, Canon original RAW 2nd version). Image size: (RAW) 3888x2592 pixels.

Figure 57. Canon EOS 400 D Camera (photo: Canon EOS 400D, 2013).

1.1.6

Thermo Tracer TH9100pro

The thermal-infrared camera that has been used for determining the apparent temperature is a NEC Thermo Tracer TH9100Pro (Figure 58) (NEC TH9100, 2013). This camera has a spectral range of 8 through 14 µm, a resolution of about 0.04°C, and a focus distance from 30 cm to infinity. The pixel size on a distance of 1 m is about 1.2 mm and about 1,2 cm on a distance of 10 m, and the size of the image is 320 x 240 pixels. The emitted radiation intensity of the surface is received by a detector and converted into an image of pixel values of radiometric temperatures.

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

68

Figure 58. Thermo Tracer TH9100pro TIR camera for obtaining thermal images.

1.2

GEOPHYSICAL SUBSURFACE INVESTIGATION EQUIPMENT

1.1.7

Electrical resistivity

The purpose of the electrical imaging survey is to determine the subsurface resistivity distribution of the sites. The resistivity of the subsurface materials is determined largely by the water content the resistivity of the water, and secondary by the resistivity of the subsurface materials. 2D and 3D resistivity surveys have been done. The 2D survey has mainly been used for determining the best array setup (see for details Hack et al., 2008). 2D and 3D electrical imaging survey are usually carried out using a large number of electrodes connected to a multi-core cable. The typical setup for a 2-D survey with a number of electrodes along a straight line attached to a multi-core cable is illustrated in Figure 59. A computer operated “Sting R1/IP” has been used as measuring device. It is a single channel automatic resistivity-imaging device with a multi-electrode system. It has a built-in set of command files for different electrode arrays. Typically, 28 electrodes are laid out in two strings of 14 electrodes, with electrodes connected by a multi core cable to a switching box and resistance meter (Figure 60). The electrode spacing has been 1 m.

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

69

Figure 59. The electrode arrangement for a 2-D electrical imaging survey and the sequence of measurements used to build up a pseudo-section (after Loke, 2000).

Figure 60. A 2-D electrical imaging survey on the Tempeldijk-North. The equipment consists of a Sting R1/IP and 28 electrodes having a 1 m spacing laid out in two strings of 14 electrodes (photo direction about South).

1.1.8

Advantages and disadvantages of the three arrays

In 2-D imaging surveys, the electrode setups “Schlumberger”, “Wenner” and “dipole-dipole” are the electrode arrays that are the most commonly used. The choice of the “best” array for a field survey depends on the type of structure to be mapped, the sensitivity of the resistivity meter and the background noise level. The Wenner array is relatively sensitive to vertical changes (i.e. horizontal structures) in the subsurface resistivity below the center of the array. However, it is less sensitive to horizontal changes (i.e. narrow vertical structures) in the subsurface resistivity. The dipole-dipole array is most sensitive to resistivity changes between the electrodes in each dipole pair and the sensitivity contour pattern is almost vertical. This array is therefore very sensitive to horizontal changes in resistivity, but relatively insensitive to vertical changes in the resistivity. Unlike the above arrays, the Schlumberger array is moderately sensitive to both horizontal and vertical structures. In areas REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

70

where both types of geological structures are expected, this array might be a good compromise between the Wenner and the dipole-dipole array. The Schlumberger setup is used for surveys in this report.

1.1.9

Electromagnetic

Electromagnetic data is collected using a Geophex GEM2 sensor (Geophex GEM2, 2013). The GEM-2 is a handheld, digital, programmable, broadband electromagnetic sensor. The GEM-2 package consists of the sensor boom (ski), the console, a battery charger, and a shoulder strap. It is combined with a real-time GPS input for navigation (Figure 61).

Figure 61. Geophex GEM2 sensor (photo: Geophex GEM2, 2013).

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

71

APPENDIX 2 : DROOGTEPROEF EXPERIMENT 2012, VEENDERIJ, AMSTERDAM Table 7 lists the measurement locations for the “Droogteproef” experiment 2012, Veenderij, Amsterdam. Table 7. Measurement locations for the “Droogteproef” experiment 2012, Veenderij, Amsterdam; Grid: UTM (WGS84, zone 31 NH). X

Y

Location ID

123037.237668

480164.521099

00-01

123035.253528

480163.000182

00-02

123033.269387

480161.479265

00-03

123031.285247

480159.958349

00-04

123029.301107

480158.437432

00-05

123027.316967

480156.916515

00-06

123025.332827

480155.395598

00-07

123023.348687

480153.874682

00-08

123021.364546

480152.353765

00-09

123019.380406

480150.832848

00-10

123017.396266

480149.311931

00-11

123034.320227

480168.581707

05-01

123032.336092

480167.060784

05-02

123030.351952

480165.539867

05-03

123028.367812

480164.018950

05-04

123026.383672

480162.498033

05-05

123024.399531

480160.977116

05-06

123022.415391

480159.456200

05-07

123020.431251

480157.935283

05-08

123018.447111

480156.414366

05-09

123016.462971

480154.893449

05-10

123014.478831

480153.372532

05-11

123031.402787

480172.642316

10-01

123029.418613

480171.121445

10-02

123027.434473

480169.600529

10-03

123025.450333

480168.079612

10-04

123023.466193

480166.558695

10-05

123021.482052

480165.037778

10-06

123019.497912

480163.516861

10-07

123017.513772

480161.995945

10-08

123015.529632

480160.475028

10-09

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

72

123013.545492

480158.954111

10-10

123011.561352

480157.433194

10-11

123028.485346

480176.702924

15-01

123026.501164

480175.182065

15-02

123024.517024

480173.661149

15-03

123022.532884

480172.140232

15-04

123020.548744

480170.619315

15-05

123018.564603

480169.098398

15-06

123016.580463

480167.577482

15-07

123014.596323

480166.056565

15-08

123012.612183

480164.535648

15-09

123010.628043

480163.014731

15-10

123008.643903

480161.493815

15-11

123025.567905

480180.763533

20-01

123023.583767

480179.242613

20-02

123021.599627

480177.721696

20-03

123019.615487

480176.200779

20-04

123017.631347

480174.679863

20-05

123015.647207

480173.158946

20-06

123013.663066

480171.638029

20-07

123011.678926

480170.117112

20-08

123009.694786

480168.596196

20-09

123007.710646

480167.075279

20-10

123005.726506

480165.554362

20-11

123022.650465

480184.824141

25-01

123020.666312

480183.303241

25-02

123018.682172

480181.782324

25-03

123016.698032

480180.261408

25-04

123014.713892

480178.740491

25-05

123012.729752

480177.219574

25-06

123010.745612

480175.698657

25-07

123008.761472

480174.177740

25-08

123006.777331

480172.656824

25-09

123004.793191

480171.135907

25-10

123002.809051

480169.614990

25-11

123019.733024

480188.884749

30-01

123017.748865

480187.363860

30-02

123015.764724

480185.842943

30-03

123013.780584

480184.322026

30-04

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

73

123011.796444

480182.801109

30-05

123009.812304

480181.280193

30-06

123007.828164

480179.759276

30-07

123005.844024

480178.238359

30-08

123003.859883

480176.717442

30-09

123001.875743

480175.196526

30-10

122999.891603

480173.675609

30-11

123016.815584

480192.945358

35-01

123014.831454

480191.424427

35-02

123012.847313

480189.903510

35-03

123010.863173

480188.382593

35-04

123008.879033

480186.861677

35-05

123006.894893

480185.340760

35-06

123004.910753

480183.819843

35-07

123002.926613

480182.298926

35-08

123000.942472

480180.778010

35-09

122998.958332

480179.257093

35-10

122996.974192

480177.736176

35-11

123013.898143

480197.005966

40-01

123011.914001

480195.485052

40-02

123009.929861

480193.964135

40-03

123007.945721

480192.443218

40-04

123005.961581

480190.922302

40-05

123003.977440

480189.401385

40-06

123001.993300

480187.880468

40-07

123000.009160

480186.359551

40-08

122998.025020

480184.838634

40-09

122996.040880

480183.317718

40-10

122994.056740

480181.796801

40-11

123010.980702

480201.066574

45-01

123008.996559

480199.545662

45-02

123007.012419

480198.024746

45-03

123005.028279

480196.503829

45-04

123003.044139

480194.982912

45-05

123001.059998

480193.461995

45-06

122999.075858

480191.941079

45-07

122997.091718

480190.420162

45-08

122995.107578

480188.899245

45-09

122993.123438

480187.378328

45-10

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

74

122991.139298

480185.857411

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

45-11

75

9 REFERENCES Abu-Farsakh, M.Y., Zhang, Z., Tumay, M., Morvant, M., 2008. Computerized Cone Penetration Test for Soil Classification: Development of MS-Windows Software. Transportation Research Record: Journal of the Transportation Research Board. 2053. DOI: 10.3141/2053-07. ISSN: 0361-1981. pp. 47-64. Adams, M.L., Philpot, W.D., Norvell, W.A., 1999. Yellowness index: An application of spectral second derivatives to estimate chlorosis of leaves in stressed vegetation. International Journal of Remote Sensing. 20 (18). DOI: 10.1080/014311699211264. ISSN: 0143-1161, 1366-5901. pp. 3663-3675. AHN, 2013. Actueel Hoogtebestand Nederland. http://www.ahn.nl/ [Accessed: 6 March 2013] (in Dutch) Asner, G.P., 1998. Biophysical and Biochemical Sources of Variability in Canopy Reflectance. Remote Sensing of Environment. 64 (3). DOI: 10.1016/s0034-4257(98)00014-5. ISSN: 00344257. pp. 234253. Bakkenist, S., Van Dam, O., Van der Nat, A., Thijs, F., De Vries, W., 2012a. Bouwstenen professionele inspecties; Handreiking voor organiseren van inspecties; Rapport 2012-13. Stichting Toegepast Onderzoek Waterbeheer (STOWA), Amersfoort, The Netherlands. ISBN: 9789057735653. p. 62. http://www.stowa.nl [Accessed: 3 June 2013] (in Dutch) Bakkenist, S., Van Dam, O., Van der Nat, A., Thijs, F., De Vries, W., 2012b. Inspectiewijzers; Rapport 201214. Stichting Toegepast Onderzoek Waterbeheer (STOWA), Amersfoort, The Netherlands. p. 143. http://www.stowa.nl [Accessed: 3 June 2013] (in Dutch) Béhaegel, M., Sailhac, P., Marquis, G., 2007. On the use of surface and ground temperature data to recover soil water content information. Journal of Applied Geophysics. 62 (3). DOI: 10.1016/j.jappgeo.2006.11.005. ISSN: 09269851. pp. 234-243. Bermon, T.C.K., 2008. The consequences of adjoining pixel processing in persistent scatterer interferometry; A case study on the Hondsbossche zeewering. Hanssen, R.F., Van Leijen, F.J. (Advs). MSc thesis, Delft University of Technology, Delft. p. 90. Bijlsma, S., 1982. Geology of Holocene in the Western part of the Netherlands. In: De Bakker, H., Van Der Berg, M.W. (Eds) Proceedings of the Symposium on Peatlands below the Sea Level, Wageningen, The Netherlands, 24-28 August 1981. Commission III of the Int. Peat Society & the Netherlands Soil Science Society. International Institute for Land Reclamation and Improvement ILRI, Wageningen, The Netherlands. DOI: 464565. ISBN: 90 70260, pp. 11-30. Blackburn, G.A., 2007. Hyperspectral remote sensing of plant pigments. Journal of Experimental Botany. 58 (4). DOI: 10.1093/jxb/erl123. pp. 855-867. Blum, A., 2010. Plant Breeding for Water-Limited Environments. Springer-Verlag GmbH, Heidelberg. ISBN: 978-1-4419-7491-4. p. 258. Blumberg, D.G., Freilikher, V., Kaganovskii, Y., Maradudin, A.A., 2002. Subsurface microwave remote sensing of soil-water content: Field studies in the Negev Desert and optical modelling. International Journal of Remote Sensing. 23 (19). DOI: 10.1080/01431160110116329. ISSN: 0143-1161; 1366-5901. pp. 4039-4054. Blumberg, D.G., Freilikher, V., Lyalko, I.V., Vulfson, L.D., Kotlyar, A.L., Shevchenko, V.N., Ryabokonenko, A.D., 2000. Soil Moisture (Water-Content) Assessment by an Airborne Scatterometer. Remote Sensing of Environment. 71 (3). DOI: 10.1016/s0034-4257(99)00087-5. ISSN: 00344257. pp. 309319. Bosch, J.H.A., Kok, H., 1994. Toelichtingen bij de Geologische kaart van Nederland 1:50.000. Blad Gorinhem (Gorkum) West (38W). Rijks Geologische Dienst, Haarlem, The Netherlands (in Dutch) Boylan, N., Jennings, P., Long, M., 2008. Peat slope failure in Ireland. Quarterly Journal of Engineering Geology and Hydrogeology. 41 (1). DOI: 10.1144/1470-9236/06-028. ISSN: 1470-9236. pp. 93-108. Broge, N.H., Leblanc, E., 2001. Comparing prediction power and stability of broadband and hyperspectral vegetation indices for estimation of green leaf area index and canopy chlorophyll density. Remote Sensing of Environment. 76 (2). DOI: 10.1016/s0034-4257(00)00197-8. ISSN: 00344257. pp. 156172. Caccamo, G., Chisholm, L.A., Bradstock, R.A., Puotinen, M.L., 2011. Assessing the sensitivity of MODIS to monitor drought in high biomass ecosystems. Remote Sensing of Environment. 115 (10). DOI: 10.1016/j.rse.2011.05.018. ISSN: 00344257. pp. 2626-2639. Canon EOS 400D, 2013. Canon EOS 400D, Canon Inc., Tokyo, Japan. http://www.canon.com/ [Accessed: 5 March 2013]

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

76

Carling, P.A., 1986. Peat slides in Teesdale and Weardale, northern pennines, july 1983: Description and failure mechanisms. Earth Surface Processes and Landforms. 11 (2). DOI: 10.1002/esp.3290110209. ISSN: 01979337. pp. 193-206. Carter, G.A., Knapp, A.K., 2001. Leaf optical properties in higher plants: linking spectral characteristics to stress and chlorophyll concentration. American Journal of Botany. 88 (4). ISSN: 0002-9122. pp. 677684. http://www.amjbot.org/content/88/4/677.abstract Carter, G.A., Spiering, B.A., 2002. Optical Properties of Intact Leaves for Estimating Chlorophyll Concentration. Journal of Environment Quality. 31 (5). DOI: 10.2134/jeq2002.1424. ISSN: 1537-2537. pp. 1424-1432. Carter, G.A.C., 1991. Primary and secondary effects of the water content on the spectral reflectance of leaves. American Journal of Botany. 78 (7). pp. 916-924. Chen, D., Huang, J., Jackson, T.J., 2005. Vegetation water content estimation for corn and soybeans using spectral indices derived from MODIS near- and short-wave infrared bands. Remote Sensing of Environment. 98 (2-3). DOI: 10.1016/j.rse.2005.07.008. ISSN: 00344257. pp. 225-236. Chunli, F., Fengrui, S., Li, Y., 2005. Investigation on nondestructive evaluation of pipelines using infrared thermography. In: The Joint 30th International Conference on Infrared and Millimeter Waves and 13th International Conference on Terahertz Electronics. IRMMW-THz 2005. , Williamsburg, Virginia, USA, 19-23 September 2005. IEEE, Piscataway, NJ, USA. DOI: 10.1109/icimw.2005.1572551. 2, pp. 339340. Cooksley, G., Arnaud, A., Banwell, M.-J., 2013. Geohazard monitoring and modelling using Persistent Scatterer; Interferometry in the framework of the European project Terrafirma (EGU2013-9633). In: Geophysical Research Abstracts; European Geosciences Union (EGU); General Assembly 2013, Vienna, 7–12 April 2013. 15, p. 1. http://meetingorganizer.copernicus.org/EGU2013/orals/11854 [Accessed: 20 April 2013] Cundill, S.L., Hack, H.R.G.K., Van der Meijde, M., 2012a. Investigation of remote sensing for dike inspection. In: IEEE International Geoscience and Remote Sensing Symposium (IGARSS 2012) : Remote sensing for a dynamic earth, Munich, Germany, 22-27 July 2012. IEEE, p. 1. Cundill, S.L., Hack, H.R.G.K., Van der Meijde, M., Van der Schrier, J.S., Ngan-Tillard, D.J.M., 2012b. Quality of peat dykes evaluated by remote sensing. In: Klijn, F., Schweckendiek, T. (Eds), Comprehensive Flood Risk Management: Research for Policy and Practice. CRC Press, Leiden. ISBN: 9780415621441. pp. CD-Rom. Cundill, S.L., Hack, H.R.G.K., Van der Meijde, M., Van der Schrier, J.S., Ngan-Tillard, D.J.M., 2012c. Remote sensing data for use in regional dike inspection. In: Nederlands Aardwetenschappelijk Congres 11, Koningshof, Veldhoven, The Netherlands, 29-30 March 2012, p. 21 slides. http://intranet.itc.nl/papers/2012/pres/cundill_rem_ppt.pdf Cundill, S.L., Hack, H.R.G.K., Van der Meijde, M., Van der Schrier, J.S., Ngan-Tillard, D.J.M., 2013a. Potential of using remote sensing data for dike inspection. In: Huang, Y., Wu, F., Shi, Z., Ye, B. (Eds), New Frontiers in Engineering Geology and the Environment. Springer, Berlin, Heidelberg. 9. ISBN: 978-3642-31671-5. pp. 203-206. Cundill, S.L., Van der Meijde, M., Hack, H.R.G.K., 2013b. Investigation of Remote Sensing for Potential Use in Dike Inspection. Journal of Selected Topics in Applied Earth Observations and Remote Sensing (IEEE). p. (in review). Dalton, J., 1954. Air Drying of Peat (section C3). In: Sundness, O.A. (Ed.) International Peat Symposium, Dublin, 12-17 July 1954. The Irish National Peat Committee Darvishzadeh, R., Atzberger, C., Skidmore, A.K., Abkar, A.A., 2009. Leaf Area Index derivation from hyperspectral vegetation indicesand the red edge position. International Journal of Remote Sensing. 30 (23). DOI: 10.1080/01431160902842342. ISSN: 0143-1161; 1366-5901. pp. 6199-6218. De Mulder, E.F.J., Geluk, M.C., Ritsema, I., Westerhoff, W.E., Wong, T.E., 2003. De ondergond van Nederland. Geologie van Nederland 7. Nederlands Instituut voor Toegepaste Geowetenschappen TNO, Utrecht. ISBN: 90-5986-007-1. p. 379. (in Dutch) Dykes, A.P., Kirk, K.J., 2001. Initiation of a multiple peat slide on Cuilcagh Mountain, Northern Ireland. Earth Surface Processes and Landforms. 26 (4). DOI: 10.1002/esp.188. ISSN: 0197-9337. pp. 395-408. Earth System Research Laboratory, 2013. Earth System Research Laboratory; NOAA Solar Calulator. http://www.esrl.noaa.gov/gmd/grad/solcalc/ [Accessed: 20 April 2013] ESA; Satellites keep an eye on Dutch dikes, 2012. Satellites keep an eye on Dutch dikes; Observing the Earth; European Space Agency. http://www.esa.int/Our_Activities/Observing_the_Earth/Satellites_keep_an_eye_on_Dutch_dikes [Accessed: 20 April 2013]

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

77

Evans, M.G., Burt, T.P., Holden, J., Adamson, J.K., 1999. Runoff generation and water table fluctuations in blanket peat: evidence from UK data spanning the dry summer of 1995. Journal of Hydrology. 221 (34). DOI: 10.1016/s0022-1694(99)00085-2. ISSN: 00221694. pp. 141-160. Famiglietti, J.S., Devereaux, J.A., Laymon, C.A., Tsegaye, T., Houser, P.R., Jackson, T.J., Graham, S.T., Rodell, M., van Oevelen, P.J., 1999. Ground-based investigation of soil moisture variability within remote sensing footprints During the Southern Great Plains 1997 (SGP97) Hydrology Experiment. Water Resources Research. 35 (6). DOI: 10.1029/1999wr900047. ISSN: 0043-1397. pp. 1839-1851. FC2015, 2013. Flood Control 2015. http://www.floodcontrol2015.com/ [Accessed: 20 April 2013] Fensholt, R., Sandholt, I., 2003. Derivation of a shortwave infrared water stress index from MODIS near- and shortwave infrared data in a semiarid environment. Remote Sensing of Environment. 87 (1). DOI: 10.1016/j.rse.2003.07.002. ISSN: 00344257. pp. 111-121. Field Scout TDR 100, 2013. Field Scout TDR 100 Soil Moisture Meter, Spectrum Technologies, Inc, 12360 S. Industrial Dr. East Plainfield, IL, USA. http://www.specmeters.com/ [Accessed: 5 March 2013] Fieldspec Pro, 2013. Fieldspec Pro spectrometer, ASD, Inc., Boulder, Colorado, USA. http://www.asdi.com/ [Accessed: 5 March 2013] Fuchs, M., Tanner, C.B., 1966. Infrared Thermometry of Vegetation1. Agronomy Journal. 58 (6). DOI: 10.2134/agronj1966.00021962005800060014x. ISSN: 0002-1962. pp. 597-601. Gausman, H.W., 1974. Leaf Reflectance of Near-Infrared. Photogrammetric Engineering. 40 (2). pp. 183-192. Geophex GEM2, 2013. GEM2, Geophex, Ltd. Raleigh, NC, USA. http://www.geophex.com/ [Accessed: 7 March 2013] Gernhardt, S., Adam, N., Eineder, M., Bamler, R., 2010. Potential of very high resolution SAR for persistent scatterer interferometry in urban areas. Annals of GIS. 16 (2). DOI: 10.1080/19475683.2010.492126. ISSN: 1947-5683; 1947-5691. pp. 103-111. Givehchi, M., Vrijling, J.K., Hartmann, A., Van Gelder, P.H.A.J.M., Van Baars, S., 2002. Application of remotely sensed data for detection of seepage in dikes. In: International Symposium on Resource and Environmental Monitoring (ISPRS), Hyderabad, India, 3-6 December 2002. Indian Society of Remote Sensing (ISRS), p. 11. Google Earth/Aerodata International Surveys, 2009. Google Earth/Aerodata International Surveys. http://www.google.com/earth/index.html [Accessed: 13 March 2009] Google Earth/Aerodata International Surveys; image date 2005, 2013. Google Earth/Aerodata International Surveys. http://www.google.com/earth/index.html [Accessed: 8 May 2013] Google Earth/Aerodata International Surveys; image date 2008, 2013. Google Earth/Aerodata International Surveys. http://www.google.com/earth/index.html [Accessed: 8 May 2013] Google Maps, 2012. Google Maps. http://maps.google.nl [Accessed: 30 July 2012] Haarbrink, R.B., Shutko, A.M., 2008. Simultaneous multi-sensor data for global information management system. In: 2008 Microwave Radiometry and Remote Sensing of the Environment; 10th Specialist Meeting; MICRORAD 2008, Florence, Italy, 11-14 March 2008. Institute of Electrical and Electronics Engineers (IEEE)/Curran Associates Inc., Red Hook, NY, USA. DOI: 10.1109/micrad.2008.4579460. ISBN: 978-1-4244-1986-9, pp. 17-20. Haboudane, D., Miller, J.R., Pattey, E., Zarco-Tejada, P.J., Strachan, I.B., 2004. Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop canopies: Modeling and validation in the context of precision agriculture. Remote Sensing of Environment. 90 (3). DOI: 10.1016/j.rse.2003.12.013. ISSN: 00344257. pp. 337-352. Hack, H.R.G.K., Van der Meijde, M., Van der Schrier, J.S., Awaju, J.H., Rupke, J., Barritt, S., Van 'T Hof, J., Maccabiani, J., Maresch, S., Calero, D.P., Reymer, A., Schweckendiek, T., Stoop, J., Wilbrinck, H., Zomer, W., 2008. Strength of peat dykes evaluated by remote sensing (Gebiedsdekkende dijksterkte bepaling met remote sensing), Pilot project: RSDYK2008, Program Flood Control 2015. Flood Control 2015, Delft. p. 135 Hanssen, R., Van Leijen, F., Dentz, F., Van Halderen, L., Possel, B., Esfahany, S., Slobbe, C., Wortel, T., 2007. Satellite dike monitoring; Safety checks from space. Leonardo Times; Periodiek van de Vliegtuigbouwkundige Studievereniging 'Leonardo Da Vinci'. 11 (1). pp. 18-19. http://www.vsv.tudelft.nl/en/node/398 [Accessed: 20 April 2013] Hanssen, R.F., Van Leijen, F.J., 2008. Monitoring water defense structures using radar interferometry. In: 2008 IEEE Radar Conference, Rome, 26-30 May 2008. Institute of Electrical and Electronics Engineers ( IEEE )/Curran Associates Inc., Red Hook, NY, USA. DOI: 10.1109/RADAR.2008.4720874. ISBN: 1097-5659; E-ISBN : 978-1-4244-1539-7; P-ISBN: 978-14244-1538-0, pp. 1594-1597.

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

78

http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=4720874&isnumber=4720717 [Accessed: 20 April 2013] Heasler, H.P., Jaworowski, C., Foley, D., 2009. Geothermal systems and monitoring hydrothermal features. In: Young, R., Norby, L. (Eds) Geological Monitoring, Boulder, Colorado. The Geological Society of America. DOI: 10.1130/2009.monitoring(05), pp. 105–140. Heitger, R., 2006. Thermal Infrared Imaging for the Charity Hospital Cemetery Archaeological Survey: Implications for Further Geological Applications. Serpa, L., Sarwar, A.K.M., Easley, D. (Advs). MSc thesis, University of New Orleans, New Orleans. (318). p. 53. http://scholarworks.uno.edu/td/318 [Accessed: 22 May 2013] Hobbs, N.B., 1986. Mire morphology and the properties and behaviour of some British and foreign peats. Quarterly Journal of Engineering Geology and Hydrogeology. 19 (1). DOI: 10.1144/gsl.qjeg.1986.019.01.02. ISSN: 1470-9236. pp. 7-80. Hopkins, W.G., Hüner, N.P.A., 2009. Introduction to Plant Physiology, 4 ed. Wiley, New York. ISBN: 978-0470-24766-2. p. 528. Howard, G.B., 2001. Application of airborne thermal infrared imaging for the detection of unexploded ordnance. In: Rozlosnik, A.E., Howard, G.B., Dinwiddie, R.B. (Eds) Thermosense XXIII, Orlando, FL, 23 March 2001. The International Society for Optics and Photonics (SPIE), Bellingham, Washington. DOI: 10.1117/12.420986. ISBN: 9780819440556. 4360, pp. 149-160. Hunt, E.R., Rock, B.N., 1989. Detection of changes in leaf water content using Near- and Middle-Infrared reflectances☆. Remote Sensing of Environment. 30 (1). DOI: 10.1016/0034-4257(89)90046-1. ISSN: 00344257. pp. 43-54. Idso, S.B., Schmugge, T.J., Jackson, R.D., Reginato, R.J., 1975. The utility of surface temperature measurements for the remote sensing of surface soil water status. Journal of Geophysical Research. 80 (21). DOI: 10.1029/JC080i021p03044. ISSN: 0148-0227. pp. 3044-3049. Inagaki, T., Okamoto, Y., 1997. Diagnosis of the leakage point on a structure surface using infrared thermography in near ambient conditions. NDT & E International. 30 (3). DOI: 10.1016/s09638695(96)00040-0. ISSN: 09638695. pp. 135-142. International Peat Society, 2013. website of the International Peat Society. http://www.peatsociety.org/ [Accessed: 27 February 2013] Jackson, R.D., Idso, S.B., Reginato, R.J., Pinter, P.J., 1981. Canopy temperature as a crop water stress indicator. Water Resources Research. 17 (4). DOI: 10.1029/WR017i004p01133. ISSN: 0043-1397. pp. 1133-1138. Jensen, J.R., 2007. Remote sensing of the environment: an earth resource perspective (2nd ed.). Pearson Prentice Hall, Upper Saddle River, NJ. ISBN: 9780131889507. p. 592. Kim, Y., Glenn, D.M., Park, J., Ngugi, H.K., Lehman, B.L., 2010. Hyperspectral image analysis for plant stress detection. In: Proceedings American Society of Agricultural and Biological Engineers Annual International Meeting (ASABE), Pittsburgh, PA, USA, 20-23 June 2010. American Society of Agricultural and Biological Engineers (ASABE)/Curran Associates, Inc., St. Joseph, Michigan, USA. ISBN: 9781617388354, pp. 3512-3524. Kottek, M., Grieser, J., Beck, C., Rudolf, B., Rubel, F., 2006. World Map of the Köppen-Geiger climate classification updated. Meteorologische Zeitschrift. 15 (3). DOI: 10.1127/0941-2948/2006/0130. ISSN: 09412948. pp. 259-263. Kyllo, K.P., 2003. NASA funded research on agricultural remote sensing, Department of Space Studies, University of North Dakota. Li, J., Islam, S., 1999. On the estimation of soil moisture profile and surface fluxes partitioning from sequential assimilation of surface layer soil moisture. Journal of Hydrology. 220 (1-2). DOI: 10.1016/s00221694(99)00066-9. ISSN: 00221694. pp. 86-103. Lillesand, T.M., Kiefer, R.W., Chipman, J., 2008. Remote sensing and image interpretation, 6 ed. Wiley, New York. ISBN: 978-0-470-05245-7. p. 804. Loke, M.H., 2000. Electrical imaging surveys for environmental and engineering studies, A practical guide to 2D and 3-D surveys. Geotomo Software SDN. BHD., 115, Cangkat Minden Jalan 5, Minden Heights, 11700 Gelugor, Penang, Malaysia. p. 67. http://www.geotomosoft.com/index.php [Accessed: 5 March 2013] Long, M., Jennings, P., 2006. Analysis of the peat slide at Pollatomish, County Mayo, Ireland. Landslides. 3 (1). DOI: 10.1007/s10346-005-0006-z. ISSN: 1612-510X. pp. 51-61. http://dx.doi.org/10.1007/s10346005-0006-z (in English) Luyken, J., 1694. Het Menselyk Bedryf ("The Book of Trades" by Luyken, 16 April 1649 - 5 April 1712; Dutch poet, illustrator and engraver). Johannes en Caspaares Luyken, Amsterdam. p. 212. (in Old Dutch)

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

79

MacFarlane, I.C., Radforth, M.W., 1965. A study of the physical behaviour of peat derivatives under compression (article III.5). In: Macfarlane, I.C. (Ed.) 10th Muskeg Research Conference, Technical Memorandum 85, Ottawa, 21-22 May 1964. National Research Council of Canada; Associate Committee on Soil and Snow Mechanics, Ottawa. DOI: irc_id:7530, pp. 159-164. McCahon, C.P., Carling, P.A., Pascoe, D., 1987. Chemical and ecological effects of a pennine peat-slide. Environmental Pollution. 45 (4). DOI: 10.1016/0269-7491(87)90102-3. ISSN: 0269-7491. pp. 275-289. http://www.ncbi.nlm.nih.gov/pubmed/15092746 (in eng) Miramap, 2013. Miramap homepage. Miramap, 2007. Miramap homepage. [Online URL: http://www.miramap.com/ Accessed on11 August 2010]. [Accessed: 20 April 2013] Mora, P., Berti, M., Simoni, A., 2007. Observations of soil moisture from thermal infrared data. Geophysical Research Abstracts; European Geosciences Union (EGU). 9 (04157). DOI: 1607-7962/gra/EGU2007A-04157. ISSN: 1607-7962. p. 2. Moser, G.M., Zomer, W.S., 2006. Inspectie van Waterkeringen; Een overzicht van Meettechnieken; STOWA Rapport 2006-10; DWW Rapport 2006-60. STOWA; Ministerie van Verkeer en Waterstaat; Rijskwaterstaat, Utrecht, The Netherlands. ISBN: 9057733196. p. 106 (in Dutch) NEC TH9100, 2013. NEC TH9100 Series Infrared Thermal Imaging Cameras, Nippon Avionics, Co., Ltd. NEC San-ei Instruments Ltd., Tokio. http://www.avio.co.jp/english/products/infrared/index.html [Accessed: 5 March 2013] Nellis, M.D., 1982. Application of thermal infrared imagery to canal leakage detection. Remote Sensing of Environment. 12 (3). DOI: 10.1016/0034-4257(82)90055-4. ISSN: 00344257. pp. 229-234. Ngan-Tillard, D.J.M., 2011. Electromagnetic survey Reeuwijk Tempeldijk-South (unpublished), Delft Nienhuis, P.H., 2008. Environmental History of the Rhine–Meuse Delta; An ecological story on evolving human–environmental relations coping with climate change and sea-level rise, 1 ed. Springer, Berlin. ISBN: 978-1-4020-8211-5 (Print) 978-1-4020-8213-9 (Online) NOAA-CSC, 2008. Spectral signature. National Oceanic and Atmospheric Administration; Coastal Services Center, Charleston, SC, USA. www.csc.noaa.gov/product/sccoasts/html/rsdetail.htm Nowatzki, J., Andres, R., Kyllo, K., 2004. Agricultural Remote Sensing Basics. NDSU Extension Service, Fargo, North Dakota. http://www.ag.ndsu.edu/pubs/ageng/gis/ae1262.pdf [Accessed: 13 March 2013] Nterekas, D., Oostveen, J.P., Mathijssen, F.A.J.M., Brinkgreve, R.B.J., Hicks, M.A., Molenkamp, F., 2011. Failure of peat dykes by drought. In: Alonso, E., Gens, A. (Eds) Unsaturated Soils; Proceedings of The Fifth International Conference on Unsaturated Soils, Barcelona, Spain, 6–8 September 2010. CRC Press, Taylor & Francis Group, London. DOI: 10.1201/b10526-1. 2, pp. 1325–1330. O’Kelly, B.C., 2008. On the geotechnical design and use of peat bunds in the conservation of bogs. In: Bouassida, M., Romdhane, N.B., Hamdi, E. (Eds) Proceedings Innovative Geotechnical Engineering; International Conference on Geotechnical Engineering (ICGE’08), Hammamet, Tunisia, 24-26 March 2008. Sfax; Nouha Editions, Tunisia, pp. 259–267. Peñuelas, J., Filella, I., Biel, C., Serrano, L., Savé, R., 1993. The reflectance at the 950–970 nm region as an indicator of plant water status. International Journal of Remote Sensing. 14 (10). DOI: 10.1080/01431169308954010. ISSN: 0143-1161; 1366-5901. pp. 1887-1905. Petrone, R.M., Price, J.S., Carey, S.K., Waddington, J.M., 2004. Statistical characterization of the spatial variability of soil moisture in a cutover peatland. Hydrological Processes. 18 (1). DOI: 10.1002/hyp.1309. ISSN: 0885-6087. pp. 41-52. Pion, 2013. Pion. http://www.pion.cz [Accessed: 20 April 2013] Price, J.C., 1980. The potential of remotely sensed thermal infrared data to infer surface soil moisture and evaporation. Water Resources Research. 16 (4). DOI: 10.1029/WR016i004p00787. ISSN: 0043-1397. pp. 787-795. Reginato, R.J., Idso, S.B., Vedder, J.F., Jackson, R.D., Blanchard, M.B., Goettelman, R., 1976. Soil water content and evaporation determined by thermal parameters obtained from ground-based and remote measurements. Journal of Geophysical Research. 81 (9). DOI: 10.1029/JC081i009p01617. ISSN: 0148-0227. pp. 1617-1620. Robertson, P.K., 1990. Soil classification using the cone penetration test. Canadian Geotechnical Journal. 27 (1). DOI: 10.1139/t90-014. ISSN: 0008-3674; 1208-6010. pp. 151-158. Rouse, J.W., Haas, R.H., Schell, J.A., Deering, D.W., 1974. Monitoring vegetation systems in the Great Plains with ERTS; SP-351. In: Freden, S.C., Mercanti, E.P., Becker, M.A. (Eds) 3rd Earth Resources Technology Satellite (ERTS)-1 Symposium, Goddard Space Flight Center, Washington D.C., 10-14 December 1973. NASA, Goddard Space Flight Center, Scientific and Technical Information Office, Washington, D.C. 1, pp. 309-317.

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

80

Rupke, J., 2008. Surface layer of peat dykes often consists of man-made layer of clay and peat made by farmers to improve bearing capacity for cattle (personal communication). Schubert, E.F., 2006. Light-Emitting Diodes (2nd ed.). Cambridge University Press, Cambridge. ISBN: 9780521865388. p. 432. Sikder, S.H., 1994. Variability of geotechnical properties of peat. Zigterman, W. (Adv.). MSc thesis, International Institute for Geo-Information Science and Earth Observation (ITC), Delft. p. 250. Swart, L.M.T., 2007. Remote sensing voor inspectie van waterkeringen (602.29; version 4.3.2 definitief). Swartvast, Nieuw-Vennep, The Netherlands. ISBN: 978-90-79331-01-7. p. 136 (in Dutch) Szatyłowicz, J., Gnatowski, T., Szejba, D., Oleszczuk, R., Brandyk, T., 2007. Moisture content variability in drained fen soil. In: Okruszko, T., Maltby, E., Szatytowicz, J., Swiatek, D., Kotowski, W. (Eds) International conference W3M: Wetlands: modelling, monitoring, and management, Wierzba, Poland, 22-25 September 2005. Taylor & Francis, London. ISBN: 978-0-415-40820-2, pp. 113-120. Tallis, J.H., Meade, R., Hulme, P.D. (Eds), 1997. University of Manchester, 9-11 April 1997. Mires. Research Group, Macaulay Land Use Research Institute, Aberdeen. ISBN: 0708405975, 9780708405970. p. 222. Tansey, K.J., Millington, A.C., Battikhi, A.M., White, K.H., 1999. Monitoring soil moisture dynamics using satellite imaging radar in northeastern Jordan. Applied Geography. 19 (4). DOI: 10.1016/s01436228(99)00009-0. ISSN: 01436228. pp. 325-344. Ten Cate, J.A.M., 1982. The River and Coastal Plains of The Netherlands. Report of the working group from September 13, 1981 till May 31, 1982 - IGU working Group on Geomorphology of River and Coastal Plains. Ten Cate, J.a.M. (Ed.), Wageningen, The Netherlands. pp. 4-11 Terrafirma, 2013. Terrafirma; Pan-European Ground Motion Hazard Information Service; EESA. EESA. http://www.terrafirma.eu.com/ [Accessed: 20 April 2013] Tetracam, 2013. Tetracam Agricultural Digital Camera (ADC). Tetracam Chatsworth, CA, USA. http://www.tetracam.com/ [Accessed: 12 March 2013] ThetaProbe, 2013. ThetaProbe Soil Moisture Sensor - ML2x, Delta-T Devices Ltd, Cambridge, UK. http://www.delta-t.co.uk/ [Accessed: 5 March 2013] Tucker, C.J., 1979. Red and photographic infrared linear combinations for monitoring vegetation. Remote Sensing of Environment. 8 (2). DOI: 10.1016/0034-4257(79)90013-0. ISSN: 00344257. pp. 127-150. Van Baars, S., 2005. The horizontal failure mechanism of the Wilnis peat dyke. Géotechnique. 55 (4). DOI: 10.1680/geot.2005.55.4.319. ISSN: 0016-8505, 1751-7656. pp. 319-323. Van Baars, S., Van Kempen, I.M., 2009. The cause and mechanisms of historical dike failures in The Netherlands. E-Water; European Water Association (EWA). 2009/06. ISSN: 1994-8549. p. 14. http://www.dwa.de/portale/ewa/ewa.nsf/home?readform&objectid=8428F628AB57BECFC125766C00 3024B6 [Accessed: 12 December 2012] Van de Griend, A.A., Owe, M., Groen, M., Stoll, M.P., 1991. Measurement and Spatial Variation of Thermal Infrared Surface Emissivity in a Savanna Environment. Water Resources Research. 27 (3). DOI: 10.1029/90wr02616. ISSN: 0043-1397. pp. 371-379. Van der Meer, F.D., Van der Meijde, M., Kooistra, J.F., Van der Werff, H.M.A., Noomen, M.F., 2006. Detection of hazardous gas leakage from pipelines using anomalous spectral reflectance features of vegetation. In: Marçal, A. (Ed.) Proceedings 25th Symposium European Association Remote Sensing Laboratories (EARSeL); Global Developments in Environmental Earth Observation from Space, Porto, Portugal, 6-11 June 2005. Millpress, Rotterdam. DOI: 1874/142608, pp. 191-198. Van der Meijde, M., Van der Werff, H.M.A., Van der Meer, F.D., 2006. Detection of anomalous vegetation related to pipeline leakages using airborne hyperspectral measurements : final report. University Twente, ITC, Enschede. p. 25 Van der Werff, H.M.A., Noomen, M.F., Van der Meijde, M., Van der Meer, F.D., 2007. Remote sensing of onshore hydrocarbon seepage: problems and solutions. In: Teeuw, R.M. (Ed.), Mapping Hazardous Terrain using Remote Sensing, Special Publications. Geological Society, London. 283 (1). pp. 125133. Warburton, J., Holden, J., Mills, A.J., 2004. Hydrological controls of surficial mass movements in peat. EarthScience Reviews. 67 (1-2). DOI: 10.1016/j.earscirev.2004.03.003. ISSN: 00128252. pp. 139-156. Ward, W.H., Penman, A., Gibson, R.E., 1955. Stability of a Bank on a Thin Peat Layer. Géotechnique. 5 (2). ISSN: 1751-7656. pp. 154 –163. World Maps climate classification, 2013. World Maps of Köppen-Geiger climate classification. http://koeppengeiger.vu-wien.ac.at/ [Accessed: 2 March 2013]

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

81

Yang, H., 1995. Detection of areas of spontaneous combustion of coal by using airborne and TM data in Xinjiang, China. MSc thesis, International Institute for Geo-Information Science and Earth Observation (ITC), Enschede. p. 73. Zwang, L.W.A., Monden, M., 2012. MIP-Sonderingen a/d Noordzijde, Tempeldijk te Reeuwijk; opdrachtnummer: 1212-0114-000. Fugro GeoServices B.V.; Waterbouw, Nieuwegein. p. 35 (in Dutch)

REMOTE SENSING & DIKE QUALITY - 2012.11.01.1, 2, 4 & 6

82