ASSESSMENT OF USING LAND SURFACE TEMPERATURE (LST) AND SATELLITE REMOTE SENSING FOR WHEAT YIELD MODELLING IN THE NORTH KAZAKHSTAN REGION

Abstract

The results of the study illustrated that the simple linear regression model using the NDVI index produces better results in comparison to the multiple regression model using NDVI and LST (88% and 71% comparatively). Similarly, the RMSE also showed surpassed results of the simple linear regression model (4,15 to 6,78). Additionally, the fact that the implementation of fertilizers in predicted year for fallow fields has significantly increased wheat yield and influenced to the prediction accuracy. The study illustrates that the accuracy in statistical models with the use of several variables is distorted.The simple linear regression model might be applied to different crop types and territories.