Mapping of sugarcane crop types from multi-date IRS-Resourcesat satellite data by various classification methods and field-level GPS survey

Abstract Sugarcane crop identification and crop distribution information provides an important basis for crop acreage and yield estimation. In this study three methods i.e. object-based classification, multi-date supervised classification and knowledge-based classification methods were used for the analysis of multi-temporal LISS-III and single date LISS-IV image of Resourcesat-2A satellite. The comparison of classification results showed that planted sugarcane area overlapping was ranging from 79.30% to 91.66% and ratoon sugarcane area overlapping was ranging from 61.17% to 79.44%. Further, the classification results for planted and ratoon sugarcane at pixel levels were integrated to increase the reliability of results using the decision tree method. The integration of all the three classification results increased the overall accuracy that was 86.15% with kappa coefficient of 0.73. The use of remote sensing techniques to extract the sugarcane field information is an economically effective method that can be further used for modelling crop production in the region, forecast crop production and management of resources.