INTER-COMPARISON OF SATELLITE BASED VEGETATION INDICES TO ESTIMATE CROP YIELD IN IRRIGATED AREAS OF INDUS BASIN

摘要

Crop yield prediction is of significant importance to ensure food security and making efficient food management plans. The information on crop yield well before harvesting is required to prepare future plans. Traditionally in Pakistan crop yield estimation is being carried out by manual surveys and Village Master Sampling (VMS) which is laborious and time consuming. Initially, a comparison was performed between various satellite based indices e.g. Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EVI), Soil Adjusted Vegetation Index (SAVI), Modified Soil Adjusted Vegetation Index (MSAVI) to evaluate most appropriate vegetation index that performs better in irrigated areas of Indus Basin. Two Canal Commands i.e. Lower Chenab (LCC) and Muzaffargarh Canal (MZG) were selected for this purpose. A stepwise regression based model was developed for Wheat crop yield using MODIS and Landsat 8 vegetation products. Landuse and Land cover map was prepared by Supervised Classification and expert knowledge. Landsat 8 SAVI performed better than other indices of both Landsat 8 and MODIS with R~2 of 0.6 between SAVI and insitu crop yield. The relation between observed yield and predicted for Landsat 8 using SAVI based regression model was better than other indices with R2, Pearson correlation and Nash Scuffle Efficiency estimated at 0.70, 0.83 and 0.617 respectively. The maps thus generated will help concerned authorities to prepare security plan and reciprocate measures will be suggested for improvement.