APPLICATION OF VHI VEGETATION INDICES TO CROP YIELD FORECASTING

Abstract

The aim of this work is to highlight the algorithm and results of modeling the average regional levels of cereals and legumes yields in some regions of Ukraine (Odessa region for example) using remote data, which used the vegetation index VHI. Methods. Model calculations were performed according to the productivity of cereals and legumes in Odessa region for 2011-2020 and the vegetation index VHI for the same period. VHI products received from NOAA STAR - Global Vegetation Health Products system (4 km resolution, 7-day composite). The relationship between VHI and cereals and legumes yields was assessed by correlation-regression analysis. Results. Statistically significant relationships between VHI and cereals and legumes yields levels in Odessa region with a correlation coefficient of 0.8- 0.9 in the period from April to July were establish. Regression dependences for early forecast of сereals and legumes yields (as of the end of April and May) were established using VHI for 16 and 20 weeks (from the beginning of the year). The correlation coefficient between the actual yield Ufact and the model values is 0.93 for Ufor(16) and 0.89 for Ufor(20). The forecast error did not exceed 10 % for Ufor(16) in 70 % of cases, and for Ufor(20) – in 80 % of cases. Conclusions. The authors established regression dependences for the early forecast (as of the end of April and May) of cereals and legumes yields in Odesa region using the region-averaged vegetation indices VHI for 16 and 20 weeks from the beginning of the year. This algorithm can be used to build model ratios for calculating crop yields for different regions of Ukraine and separately for different crops.