ABSTRACT
348Estimation of crop production in advance plays a key role in the economy of agrarian countries, particularly in developing countries like India in order to obtain cropped area and the average yields. Reliable and timely estimates of crop area coverage are of great importance to planners and policy makers and researchers throughout the world. Sorghum and wheat are dominant crops grown in rabi season in Maharashtra, Central India, and it assumes great importance in estimation of acreage. Traditionally, the estimates of crop acreage are obtained through complete enumeration by revenue agencies (Giradwari system), sample surveys and personal assessment by village patwari. Application of remote sensing and Geographic Information System (GIS) can provide more accurate and timely estimates of crop acreages. The present chapter deals with the methodology adopted for classifying rabi sorghum and wheat crops using time series Normalized Difference Vegetation Index (NDVI) derived from multi-date Indian Remote Sensing (IRS)-P6, Advanced Wide Field Sensor (AWiFS) 56 m resolution data for rabi season of 2012–13. Study area covers the dominant rabi sorghum and wheat growing districts of Maharashtra state namely Pune, Solapur, Ahmednagar, Beed and Osmanabad. Two-stage classification of the datasets has been carried out through unsupervised Iterative Self Organizing Data Analysis Technique (ISODATA) to label classes based on temporal spectral profiles of wheat and rabi sorghum as well as other competing crops, vectorisation of classified image, manual editing and labeling of mixed class polygons. The decision rule based integration was followed to generate final classified image. This hybrid classification technique takes advantage of inherent clustering tendency of land use/land cover classes in feature space with temporal dimension added to it in terms of NDVI time series data. It also makes use of signatures of known crop classes for labeling the clusters. Analysis of rabi sorghum and wheat acreage estimation in the study area shows 16,69,599 ha and 1,89,481 ha, respectively, which is deviating by 8.81% and 9.78%, respectively from the reference data as reported by Government of Maharashtra. It is observed that this technique is simple, time saving, less subjective and requires less expertise compared to hierarchical classification technique.
