Deriving cropping system performance indices using remote sensing data and GIS

International Journal of Remote Sensing Vol. 26, No. 12, 20 June 2005, 2595–2606

Deriving cropping system performance indices using remote sensing data and GIS S. PANIGRAHY*, K. R. MANJUNATH and S. S. RAY Agroecology & Management Division, ARG, RESIPA, Space Applications Centre, ISRO, Ahmedabad 380015, India (Received 1 June 2004; in final form 23 December 2004 ) The cropping system approach is a holistic management of variant and invariant resources to optimize the food production. Various indices are used to assess and evaluate the efficiency and sustainability of the systems. These indices are generally computed from the data collected by traditional survey methods that are time consuming and non-spatial. An attempt has been made to derive such indices using satellite remote sensing data for the state of West Bengal, India. Three indices—Multiple Cropping Index (MCI), Area Diversity Index (ADI) and Cultivated Land Utilization Index (CLUI)—were attempted. Multi-date, multisensor data from Indian Remote Sensing Satellite (IRS) and Radarsat Synthetic Aperture Radar (SAR) were used to derive cropping pattern, crop rotation, and crop calendar. Crop type, acreage, rotation and crop duration were used as inputs to compute the indices at district and state level. The indices were categorized as high, medium and low to evaluate the performance of each of the 16 districts. The average MCI of the state derived was 140. At district level it varied from 104 to 177. The average ADI of state was 2.5 and varied from 1.5 to 5.0.

1.

Introduction

A cropping system is defined as the management of a cropping pattern to optimize benefits from a given resource base under specific environmental conditions (Zandastra 1977). This is now the thrust approach to attain long-term food and nutritional security. Various indices are used to assess system performance (Virmani and Singh 1997). The indices are calculated using data collected by traditional survey methods, so the calculation is time consuming and rarely timely. There is a requirement for timely generation of such indices of large areas, so that effective measures can be planned. Spatial representation of the indices in a map format is desirable, to facilitate analysis of cause–effect relationships. Recent advances in satellite remote sensing (RS) and Geographical Information Systems (GIS) have opened up new dimensions of natural resource analysis. Remotely sensed data have been extensively used to survey various natural resources, such as agriculture, forest and other vegetation. Although the application of remote sensing in agriculture, i.e. in crops and soils, is a complex process, it has many advantages over the traditional methods in agricultural resource surveying. Multi-temporal satellite data are the only feasible source of timely, cost effective and relatively *Corresponding author. Email: [email protected] International Journal of Remote Sensing ISSN 0143-1161 print/ISSN 1366-5901 online # 2005 Taylor & Francis Group Ltd http://www.tandf.co.uk/journals DOI: 10.1080/01431160500114698

2596

S. Panigrahy et al.

accurate information in spatial format (maps) about the existing cropping systems of a country. Many studies have shown the feasibility of mapping cropping pattern, crop rotation, etc, using multi-temporal, multisensor remote sensing data (Panigrahy et al. 1995, Panigrahy and Sharma 1997). Considerable work has been carried out in India to analyse the cropping system using RS and GIS. The current study highlights the methodology used and results obtained in deriving a few important cropping system performance indices using RS data. The methodology was applied over the state of West Bengal, India. 2.

Methodology

2.1

Study area

The state of West Bengal has a geographical area of 8686.63 thousand hectares and lies on the eastern coast of India, between 21u459 and 26u459 N latitude and 86u009 and 90u009 E longitude. Administratively, the state is divided into 17 districts. The state is predominantly agricultural with more than 60% area under agriculture. It has rice based cropping system. There are three major crop seasons in the state. The main season coincides with the south-west monsoon (known as kharif) that extends from July to October. This is followed by the winter season, called rabi (November– February) and summer (March–June). Mustard, potato, tobacco, gram and vegetables are the dominant crops during rabi season in the state, while rice, sesamum and vegetables are grown in kharif season. Table 1 shows the growing calendar of major crops in the three seasons in the state. 2.2

Data

The Indian Remote Sensing Satellite (IRS) Wide Field Sensor (WiFS) that provides 188 m spatial resolution data in red and near infrared (NIR) bands with five-day revisit capability was used to analyse the rabi and summer season cropping patterns. Temporal data of 14 dates acquired from October 2000 to June 2001 were used for this purpose. Three-date Synthetic Aperture Radar (SAR) data from RADARSAT of ScanSAR narrow beam mode acquired during July–September were used to derive kharif season cropping pattern, as cloud-free data from optical sensors is rarely available during this period. ScanSAR narrow beam data has 50-m resolution and a swath of 300 km, and repeat cycles of 24 days. These data have been found suitable for large area rice crop monitoring and are used operationally in India for kharif rice crop acreage estimation (Panigrahy et al. 2000). Table 1.

Crop seasons and growing calendar of major crops in the study area.

Cropping system indices using remote sensing 2.3

2597

Analysis approach

2.3.1 Cropping pattern map. Optical datasets were used to generate rabi and summer season cropping pattern maps. The broad procedure includes creating a geo-referenced, registered multi-date optical database, overlaying administrative boundaries (state, district, block), generating an agricultural area mask, and transferring the location of ground truth information of major crops. The temporal Normalised Difference Vegetation Index (NDVI) of all the dates was created to analyse the growth pattern of various crops in a time domain for a crop year. This was used to select the optimum period for each season that could be used to classify the crops. Two- and three-date cloud-free data were selected for rabi and summer season respectively. Supervised classification was carried out to generate cropping pattern maps. Three-date ScanSAR data were analysed to derive the kharif cropping pattern. Rice is the single most dominant crop during kharif, occupying 95% of crop area. Lowland rice (ponding of water and puddling for planting) is the major practice. Mapping of rice was carried out using calibrated SAR images in backscatter domain. The classification algorithm uses a decision rule based on the unique temporal backscatter of rice crop (LeToan et al. 1997, Chakraborty and Panigrahy 2000). Cropping pattern maps were used to compute acreage of different crops. 2.3.2 Crop rotation map. The three cropping pattern maps (kharif, rabi and summer) were used to generate crop rotation map. A logical class-code combination algorithm was used to derive the crop rotation map (Panigrahy and Sharma 1997). Crop rotation was used to compute how many crops were grown in a unit area in a sequence within a year. 2.3.3 Crop calendar. The NDVI profile of the crop rotation classes was used to derive the crop phenology (total duration, emergence and maturity). This was used to derive the activity period for each crop, accounting for sowing and harvesting activity. Ancillary data on crop type, management practice and survey data were used to derive this. For example, 20 days were added to the beginning and end of the total duration of NDVI profile to account for field preparation/sowing activity in the beginning and harvesting activity (uprooting, tuber extraction, piling). For the kharif season rice, the crop duration was derived by computing the transplanting period from the characteristic dip in backscatter and adding the total duration as per the crop variety. The advantage of using spectral profile to derive the crop duration is that it represented the spatial variation of crop calendar within an area that exists due to climate, variety and socio-economic reasons. 2.3.4 Cropping system indices. Several indices have been proposed to evaluate and compare the efficiencies of different cropping systems (Palaniappan 1985). In the current study we have tried to derive three of these indices using remote sensing data. These are: Multiple Cropping Index (MCI), Area Diversity Index (ADI), and Cultivated Land Utilization Index (CLUI). The first one is indicator of crop intensification in time domain and quantifies vertical expansion of agricultural area. It is the most common index used to quantify intensity as single, double or triple cropping practice. The ADI represents diversity of crops grown and indicates sustainability. For example a unit land area practicing double cropping pattern will have the same MCI but will differ in ADI, depending on whether the second crop is the same as or different to the first crop. For example, an area where rice is followed

2598

S. Panigrahy et al.

by rice will have the same MCI as an area where rice is followed by potato, but diversity is low in the former and high in the latter. CLUI is indicator of land utilization efficiency, and indicates when and where the land lies unutilized. For example, cotton-wheat double cropping pattern (long duration crop rotation) and groundnut-potato cropping pattern (short duration crop rotation) will result in the same MCI but CLUI will be higher in the former. This index is useful in fine-tuning the cropping pattern to increase the total productivity. These three indices are useful to identify areas that need crop intensification, crop diversification or both. A system that has high MCI and high ADI is a desirable system. The mathematical derivation of the three selected indices is given below: 2.3.4.1 Multiple cropping index (MCI). This index measures the cropping intensity. It is calculated by dividing the sum of the areas planted with different crops and harvested in a single year by the total cultivated area, times 100. n P

MCI~100|

ai

i~1

,

A

ð1Þ

where n5total number of crops, ai5area occupied of the ith crop planted and harvested within a year, and A5total cultivated land area available. MCI was computed at district level. Area under different crops (ai) and net sown area (A) were derived from remote sensing data at state and district level by superimposing the administrative boundaries. MCI at pixel level was also computed by inferring n from the crop rotation map and assuming the same value, i.e. pixel size, for ai and A. This was done to prepare a spatial crop intensity map and state level MCI. 2.3.4.2 Area diversity index (ADI). Area diversity index represents the diversity of crops grown in an area over a crop year, both in time and space. This index is developed by slight modification of the commonly used diversity index (DI), which is computed as the reciprocal of sum of squares of the share of gross revenue received from each individual farm enterprise in a single year. It measures the multiplicity of crops or farm products planted in a single year.

n P

1
n yi P

i~1

i~1

DI~

!2 ,

ð2Þ

yi

where n5total numbers of enterprises (crops or farm products), and yi5gross revenue of the ith enterprise produced within the year. In this study, we were only concerned with the crop, not the farm products, and instead of revenue we computed the area occupied by each crop. Hence, the modified diversity index (which can be termed area diversity index, or ADI) is used, and can be defined as:

n P

1
n ai P

i~1

i~1

ADI~

!2 ,

ð3Þ

ai

where ai is the area under each crop that was derived from district-level crop

Cropping system indices using remote sensing

2599

statistics generated using remote sensing data. If one is interested in comparing the crop diversity in each season, n is used as the number of crops grown in a season. 2.3.4.3 Cultivated land utilization index (CLUI). This index is calculated by summing the products of land area planted to each crop, multiplying by the actual duration of that crop and dividing by the total cultivated land area times 365 days. This index measures how efficiently the available land area has been used over the year. n P

CLUI~

ai di

i~1

365|A

,

ð4Þ

where n5total number of crops, ai5the area occupied by the ith crop, di5days that the ith crop occupied ai, and A5total cultivated land area available during the 365 day period. Here, also, ai and A were computed from the crop maps, as the crop specific area and total agricultural area, respectively. The component di was computed using vegetation index (VI) profile. 3. 3.1

Results and discussion Cropping pattern and crop rotation

As observed from table 1, rice is the single dominant crop during kharif season. It is grown as a vast monoculture in the area. Analysis of the three-date ScanSAR data resulted in better than 95% classification accuracy for lowland rice, as it occupied more than 90% of net sown area in many districts. The characteristic dip in backscatter due to puddling was used to derive the transplanting period. Analysis showed that more than 80% of the crop was sown by the end of July. As the crop cycle is generally 120 days, the crop was harvested by mid-November. Mustard, potato, wheat and gram were the major rabi crops that were possible to map using WiFS data. Classification accuracy of potato was around 90%, as it was grown in large and contiguous areas. Similar accuracy was obtained for summer rice. While cropping pattern is indicative of the crops grown during a given period and helps to analyse the seasonal diversity, it is crop rotation that is key to longterm sustainability. This indicates crop intensification as well as sustainability of land and water resources. Rice grown in the kharif season, followed by a long fallow period, emerged as a major system, accounting for around 56%. The other major crop rotations mapped were rice–rice, rice–mustard, rice–potato, rice– wheat and rice–gram. 3.2

Indices

3.2.1 Multiple cropping index (MCI). The overall MCI computed using remote sensing data for the state of West Bengal was a little more than 137. At district level it varied from 104 to 174 (figure 1). In agricultural terms, MCI is indicative of multiple cropping patterns and as in equation (1) it will be 100, 200 or 300 if the total area practices single, double or triple crop in a year. Double crop is a usual practice in irrigated areas, while triple crop is practiced if short duration crops are taken. Since the study state is a predominantly rain fed area, there is hardly any district that is fully double cropped. Almost all the area is put to crop in the kharif season, and, depending on irrigation facility and residual soil moisture, a portion of land is

2600

S. Panigrahy et al.

Figure 1. Multiple cropping index (MCI) at district and state level, generated using remote sensing data for West Bengal.

again cultivated in the rabi and summer seasons. The reported MCI by traditional survey method is 145 (Anon 1998), which is a little higher than that computed from remote sensing data. This was attributed to the exclusion of some minor crops due to the limited spatial resolution of the data used. The cropping intensity maps showing the spatial pattern of the single, double and triple crop areas within each district are shown in figure 2. Since it was derived using pixel-based classification, each unit represented 200 m. The accuracy derived is directly related to the accuracy of cropping pattern maps. 3.2.2 Area diversity index (ADI). This index highlights the practice of growing different crops per unit area within a season (seasonal diversity) as well as within a year (rotational diversity). The former is encouraged as a security measure against crop failure, for farmers of rain-fed areas. The latter is of great significance for sustainable nutrient, water and pest/disease management of the agriculture system. We report here the rotational diversity. The ADI was little below 3.0 for the state, while considerable variation at district level was observed (from 1.13 to 5.0). This was found very useful particularly in discriminating high MCI districts. For example, Bardhaman and Murshidabad districts had comparative MCI of 158 and 159 respectively (figure 1). However, ADI was around 5.0 in the later and 3.5 in the former. This is mainly due to the diverse crop rotations practiced in Murshidabad, i.e. rice–rice, rice–wheat, rice–mustard, rice–gram, etc., compared to dominance of rice–rice followed by rice–potato rotation in Bardhaman. Analysis of ADI data indicated that command area districts like Bardhaman and Hughly had high MCI due to irrigation water availability, but were low in ADI due to the preference for growing rice in the summer season as a monoculture. On the other hand, districts farmers of Nadia, Murshidabad and Malda, who utilize surface and well irrigation

Cropping system indices using remote sensing

2601

Figure 2. Cropping intensity map showing distribution of single, double and triple cropping pattern areas in the state.

2602

S. Panigrahy et al.

more, opt for a number of crops other than rice, and thus have high diversity (figure 3). 3.2.3 Cultivated land utilization index (CLUI). The CLUI for the entire state was found to be 0.55, i.e. crop activities occupy the land for 180 days of a year. Since the cropping intensity is low, a low CLUI is obvious. CLUI is required to evaluate performance of areas with high cropping intensity. Districts having similar MCI may have different CLUI based on what types of crops are grown in rotation. For example, if a unit area practices rotation of rice–rice or rice–potato, the MCI will be the same but CLUI will be lower for the latter. Highest CLUI (around 230 days) was observed for the Nadia district, as it also had the highest MCI (figure 4). 3.3

System planning

Crop intensification and diversification are the two most important cropping system steps for food security in India. The former exploits the scope of vertical expansion of land utilization, while the later introduces sustainability. The ideal case is to have high MCI, high ADI and medium CLUI. Thus, areas with low and medium MCI need to set crop intensification as a priority, especially those with low MCI. These areas will have low to medium CLUI, but ADI may vary depending upon mono or multi-crop practice. Areas with high MCI generally need crop diversification if ADI is low to medium. The additional information provided by CLUI can be used to analyse what type of crops can be substituted (long or short duration). Thus, these three indices have potential use in system planning. We have tried to use a rating system for each index to categorize them as high, medium and low (table 2), and use them in conjunction to derive system planning (table 3, figure 5). Analysis showed that the districts could be categorized under five

Figure 3. Area diversity index (ADI) at district and state level generated using remote sensing data for West Bengal.

Cropping system indices using remote sensing

2603

Figure 4. Cultivated land utilization index (CLUI) at district and state level, generated using remote sensing data for West Bengal.

Table 2. Threshold values used for rating of different indices. Rating Low Medium High

MCI

ADI

CLUI

,130 130–160 .160

,2.0 2–5 .5.0

,0.5 0.5–0.6 .0.6

Table 3. Categorization of districts based on combination of rating for cropping system planning. Rating

MCI

ADI

CLUI

Cropping system plan

1

High

High

Medium/high

To maintain

2

High

Medium

High

3

Medium

Medium

High

4

Low

Low

Low/medium

Diversification with short duration crops Intensification with short duration crops Intensification and diversification

5

Medium

Medium

Medium

Intensification

District 1. 2. 3. 1.

Murshidabad Nadia Malda Hughly

1. Howrah 2. Bardhaman 1. Purulia 2. Darjeeling 3. Jalpaiguri 4. Bankura Rest of the districts

combinations. Districts having low MCI and low ADI have priority for crop intensification and diversification. This can be achieved by introducing crops other than rice in kharif and other seasons. Districts having high MCI, high CLUI and

2604

Figure 5.

S. Panigrahy et al.

Stratification of districts of West Bengal based on three cropping system indices.

Cropping system indices using remote sensing

2605

medium ADI need crop diversification with short duration crops. The ideal combination indicative of a sustainable cropping system is high MCI, ADI and moderate to high CLUI. Only three districts, Murshidabad, Nadia and Malda, belonged to this category. There is a scope for further use of these indices in efficient management both of water use and residual moisture.

4.

Conclusions

Cropping system indices are essential in the evaluation of the performance of existing agricultural systems in an area, and for carrying out effective measures to achieve desired systems in the long run. Remote sensing and GIS have the potential for large area application of these indices in the spatial domain. The multi-date, multisensor satellite data comprising IRS WiFS and Radarsat SAR were used in this study to derive cropping pattern, crop rotation and crop calendar. These were used as inputs to generate three cropping system performance indices, MCI, ADI and CLUI. These indices were used to prioritize the districts for crop intensification and diversification. The results showed the variation of the rice-based cropping system within the state, with high potential for crop intensification. Selection of optimum spatial resolution of remote sensing data is essential depending on the scale of information and complexities of cropping pattern. IRS WiFS data with 188-m resolution was found adequate for state level analysis. The study shows the possibility of using RS and GIS to create cropping system databases and its monitoring for efficient management. Acknowledgements The study was carried out under the Remote Sensing Applications Mission project of SAC, ISRO. The authors are thankful to Shri J. S. Parihar, Group Director, ARG for his guidance and critical suggestions. Thanks are due to Shri R. K. Panigrahy for his help in generating outputs. References ANON, 1998. Statistical abstracts, 1997–98, BAES, West Bengal, India (Government of West Bengal, Calcutta) p. 765. CHAKRABORTY, M. and PANIGRAHY, S., 2000. A processing and software system for rice crop inventory using multi-date Radarsat ScanSAR data. ISPRS Journal of Photogrammetry and Remote Sensing, 55, pp. 119–128. LE TOAN, T., RIBBES, F., WANG, L.-F., FLUORY, N., DING, K.-H., KUNG, J.A. and FUJITA, M., 1997, Rice crop mapping and monitoring using ERS-1 data based on experiment and modeling results. IEEE Transactions on Geoscience and Remote Sensing, 35, pp. 41–56. PALANIAPPAN, S.P., 1985, Cropping system in the tropics: Principles and management. (New Delhi: Wiley Eastern). PANIGRAHY, S. and SHARMA, S.A., 1997, Crop rotation mapping using multidate IRS digital data. ISPRS Journal of Photogrammetry and Remote Sensing, 52, pp. 85–91. PANIGRAHY, S., SINGH, R.P., SHARMA, S.A. and CHAKRABORTY, M., 1995, Results of potential use of simulated IRS-1C WiFS data for crop monitoring. Journal of the Indian Society of Remote Sensing, 23, pp. 175–183. PANIGRAHY, S., CHAKRABORTY, M., MANJUNATH, K.R., KUNDU, N. and PARIHAR, J.S., 2000, Evaluation of Radarsat ScanSAR data for rice crop inventory and monitoring. Journal of the Indian Society of Remote Sensing, 28, pp. 59–65.

2606

Cropping system indices using remote sensing

VIRMANI, S.M. and SINGH, G.B., 1997, Sustainable agriculture biophysical and agroecological indicators. In Proceedings of the 3rd Agricultural Science Congress, M.S. Bajwa, et al. (eds), Punjab Agricultural University, Ludhiana, 12–15 March, Vol. 1 (New Delhi: National Academy of Agricultural Sciences), pp. 57–62. ZANDASTRA, H.G., 1977, Cropping system research for the Asian farmers. In Proceedings of the Symposium on Cropping Systems Research and Development for the Asian farmer, IRRI, Manila, 21–24 September 1976 (Manila: IRRI), pp. 10–29.