Mathematical model for forecasting corn yield depending on the value of the normalized differential vegetation index under irrigation conditions

Abstract

Goal. To develop and scientifically substantiate mathematical models for forecasting the yield of crops on irrigated lands depending on the value of the normalized differential vegetation index (NDVI). Methods. Scientific work on the development of models of corn productivity was carried out based on the results of field studies to determine the influence of stand density and the genetic potential of crop hybrids on grain yield during 2019–2021 at the experimental fields of the Institute of Climate-oriented Agriculture of NAAS. Field experiments were laid out in 4-time replication by the method of randomized split blocks. Biological material — corn hybrids Stepovyi (FAO 190) and Tronka (FAO 380). The normalized differential vegetation index was calculated using the traditional method in QGIS 3.22 software based on Sentine-l2 satellite images. Mathematical modeling was performed in the spreadsheet processor Microsoft Excel 365 and the statistical package BioStat v.7 using correlation-regression data analysis methods. Results. It was established that the maximum closeness of the relationship between the value of satellite NDVI and the yield of corn per grain was observed in the phase of VVSN 82-87, the correlation coefficient for the years of the study was 0.93–0.96. The mathematical polynomial model (polynomial of the 2nd degree) provided a high level of accuracy (relative error 7.62%) and an acceptable level of adequacy of the predictive model (coefficient of determination 0.50). Conclusions. The possibility of highly accurate forecasting of grain corn productivity in the conditions of Southern Ukraine based on satellite NDVI data calculated in the VBSN 82-87 phase was proven. The proposed polynomial model can be used for scientific and practical purposes for forecasting the yield of crops on irrigated lands.