Wheat Yield Prediction Based on Regression Model for Prayagraj Region, Uttar Pradesh, India

Abstract

The world's agricultural landscape is dynamic and complex, with interactions between various meteorological, geographical, and agronomic factors that collectively influence crop yields. Fluctuations in weather conditions can significantly impact agricultural productivity, often leading to substantial yield losses. Accurate weather forecasting plays a crucial role in supporting the development and growth of crops. The present study aims to analyse the relationship between weather factors and wheat yield to develop accurate forecasts for better agricultural planning. Wheat yield data for the period 1997 to 2022 have been obtained from DACNET, specifically focusing on the Prayagraj district in Uttar Pradesh, India. Corresponding weather data for the same time frame have been sourced from NASA POWER. Crop yield prediction using regression is developed through the SPSS software package to assist agriculturists and farmers in obtaining accurate reports on crop production from various agricultural sources. In many cases, it becomes necessary to analyse multiple variables or entities simultaneously to enable effective decision-making. Analysis has been carried out using a dataset spanning 26 years for calibration (78%) and the remaining dataset for validation (22%). In this study, the emphasis was on constructing multivariate meteorological yield models using the stepwise linear regression method, incorporating weather parameters and historical crop production data. The model utilises variables such as maximum and minimum temperatures, precipitation, relative humidity, and wind velocity. The accuracy of these models was tested with the coefficient of determination (R2) and RMSE.