Generation of district level rice crop inventory, growth profile and yield estimation in Orissa using spot-vegetation data

Abstract

Spot-vegetation 10 day NDVI composites over Orissa state were analysed to study rice crop inventory and condition assessment. A total of 17 images from July to December during the monsoon (kharif) season of 1998 (S1) and 2001 (S2) a drought and normal year, respectively were analysed. A hierarchical decision rule-based approach that successively eliminated data loss, non vegetated land, forest cover, fallow and other crops was adopted for rice inventory. NDVI temporal profiles of rice could distinguish autumn and winter rice. The total monsoon rice area identified by RS in the state was 4.5 M ha in 1998 and 4.05 M ha in 2001 and was within 7 percent of the state level rice estimate given by Directorate of Economic Survey (DES) i.e., 4.26 and 4.22 M ha, respectively. A new profile fit i.e., a six parameter modified Gaussian approach was adopted.  The spectral profile indicated higher mean NDVI at peak growth profile of lowland winter rice (sown in June-July) in 2001-02 compared to 1998-1999. Thus, 2001-2002 rice was seen to be normal while in 1998-1999 a drought affected year. District-wise NDVI profiles of rice were generated and peak NDVI and date at peak profile were found to be correlated with rice yield at district and agro-climatic zone level. Use of rainfall with spectral profile parameters in yield model group of districts or zonal level improved coefficient of determination. This study demonstrates the utility of 1 km and 10 day NDVI composite data for rice crop assessment during monsoon season.