USE OF VEGETATION CONDITION INDEX FOR RICE YIELD FORECASTING

摘要

Rice is the staple food of country grown over the entire year. There are three main seasons for rice cultivation and they are named according to the seasons of harvest of the crop. i.e Autumn, Kharif or Winter and Boro or summer or Rabi. Major Rabi rice growing states are in the costal part of country i.e. Andhra Pradesh, Karnataka, Telangana, Odisha and West Bengal. They together consist of 9 per cent of total rice area. Accurate and timely forecasting of crop yield is a cornerstone for monitoring crop production and planning purpose, but the efficiency of the current system for near real-time forecasting may be improved by applying the remote sensing based approach. This study explored an approach for predicting the yield of Rabi rice using Vegetation Condition Index (VCI) derived from remote sensing data. Historic data of last 14-years (2003—2016) of NDVI (Normalized Difference Vegetation Index) and NDWI (Normalised Difference Wetness Index) were used to derive the VCI. MOD-13A2 series of MODIS instrument on-board Terra satellite at 16 days interval from first fortnight of November to second fortnight of March (10 fortnights) were used to calculate the NDVI & NDWI. District wise historical yield data was taken from DES. Study was carried out for 32 major Rabi Rice growing districts of Andhra Pradesh (6), Karnataka (4), Telangana (5), Odisha (8) and West Bengal (9). Stepwise regression technique were used to quantifying the relation between district wise VCI and historical yield. Strong relation (R~2) between the VCI and district wise DES yield was observed i.e. 0.45-0.97, 0.39-0.80, 0.70-0.95, 0.33-0.89, 0.25-0.70, for Andhra Pradesh, Karnataka, Odisha, Telangana and West Bengal, respectively. Except for 2 districts, the relationship was found to be statistically significant in all the districts. In 22 districts out of 32 districts, the relative deviation between DES yield and VCI estimated yield was lower than 10 per cent. Thus, this analysis showed that, in absence of weather soil and ground based observations (which are the major factors of crop production) the VCI can be used as a proxy variable for reliable yield estimation on operational basis.