2023, Vol. 8, Special Issue 4

The investigation was undertaken during Kharif season 2019 and 2020 at Research Farm, Indira Gandhi Krishi Vishwavidhyalaya, Raipur (Chhattisgarh) to study the relationship between actual and predicted post-harvest soil test value and generate post-harvest soil test value prediction equation after rice. Multiple regression analysis was performed by taking into account the post-harvest soil test values after rice as a dependent variable, and initial soil test levels, applied fertilizer doses and grain yields of rice as independent variables and predict the post-harvest soil test values for succeeding crops. The equations developed showed high predictability values for KMnO₄-N, Olsen’s P and NH₄+- acetate extractable K, respectively (98%, 99%, and 82%, respectively for rice). Significantly high R² values were recorded. This shows that such regression equations can be used for precisely predicting the availability of N, P, and K after rice for fertilizer recommendations based on targeted yield for succeeding crops. The prediction equations were then used to determine the predicted post-harvest soil test values of N, P and K for rice during 2020-21. The predicted values were then compared with the actual values observed for N, P and K and it was observed that considerably large R2 values were obtained (0.94, 0.86 and 0.85 for rice) for N, P and K, respectively. The generated model for prediction was then calibrated by determining the root mean square error (RMSE) and relative error percentage (RE). RMSE values for N, P and K followed the order: K>N>P, while the RE values followed the order: P>N>K for rice RE values less than 10% are considered that prediction is excellent.