Modeling drip-irrigated rice yield using normalized difference vegetation index: A preliminary study.

Abstract

The paper presents the results of the study devoted to the determination of the nterrelationship and interchangeability of five widely used indices of agricultural science and practise, namely NDVI, EVI, EVI2, NDWI and DSWI. The study was carried out for the fixed polygon, located in the cold arid steppe zone of Kherson region of Ukraine, during the period 2019-2022 using the cloud-free satellite imagery from Landsat-8, obtained at AgroMonitoring API. The strength and direction of the interrelationship were determined by a standard correlation analysis procedure for all pairs of vegetation indices. As a result, high discrepancies and inequality in distribution of the values of the vegetation indices studied by the years of the study were established. Analyzing the generalized data for 2019-2022, it was established that the strongest relationship and the highest possible interchangeability is attributed to the NDVI, EVI, and EVI2, where Pearson’s correlation coefficient averaged to 0.8483-0.9090. The DSWI and EVI indices are moderately correlated (the correlation coefficient is 0.7698), therefore, their interchangeability is questionable. The NDWI index has the weakest correlation with the other indices studied (the correlation coefficient does not exceed 0.6269) and is a unique measure for the determination of water stress in plants. Further studies are required to determine whether the DSWI, NDVI and EVI could be interchangeable in crops monitoring.