A field study was conducted at G.B.P.U.A.&T., Pantnagar to investigate the feasibility of estimating the yield of sugarcane crop based on weather variables. Five years (2004 to 2009) crop management data (sowing/harvest/irrigation etc.) for sugarcane were collected from Agricultural farm, Pantnagar. The development of multiple variable regression models employs that the dependent variable (yield) of multiyear is related with independent weather variables. SPSS (Statistical Package for the Social Sciences) software was used for the statistical analysis and development of multiple regression models based on fortnightly meteorological parameters. A total number of 8 models were developed using different combinations of fortnightly weather variables at different crop growth stages. Among all models, the performance of the model 8 was superior as compared to other models. The predicted yield by this model ranged between 349.39q ha-1 to 803.76q ha-1 with the value of R2 as 0.668, while the observed values ranged from 221.80q ha-1 to 824.16q ha-1. The RMSE between observed and predicted yield of sugarcane by model 8 was 15.49%, while the value of F test was 7.32 which is significant at 1% probability level. Hence, it can be concluded that the observed and predicted values were close enough in model 8 as compared to other meteorological models. The reason is obvious because model 8 used more number of weather variables.
Keywords: Weather parameters; Coefficient of correlation and Yield prediction models.
: Rajeev Ranjan