Statistical Modelling for Forecasting of Pearl Millet (Pennisetu glaucum) productivity Based on Weather Variables.Indian

Abstract

A timely and reliable forecast of yield of crop needs little emphasis for monsoon dependent country like India where, the economy is mainly based on agricultural production. Weather is a major factor affecting crop production in advanced agricultural systems. The large variation due to climate change in yield from year to year and place to place is dominated by the weather parameters. In view of fluctuating climate, a timely and reliable forecast of crop productivity could help in deciding the policies. To estimate the effect of weather variables and technological advances, 34 years productivity data of summer pearl millet from 1980 to 2013 were collected. The weekly averages of weather variables viz., bright sunshine hours (BSS), maximum temperature (MAXT), minimum temperature (MINT), morning relative humidity (RH), 1ththafternoon relative humidity (RH) and weekly total rainfall (RF) from 5 to 16 standard meteorological week of the respective year were 22considered in the study. Among the equations fitted under this approach, for 11 and 12 weeks of 27 years, for 11 weeks R was (54.50 percent) and 12 weeks (62.40percent) model and deviations of simulated forecasts ranged from 0.86 to 10.91 percent and from 3.71 to 20.93 percent 2for 11 and 12 weeks models, respectively. In case of 11 and 12 weeks’ models using data for 28 years, R was 67.20 percent for 11 weeks with 2deviation ranging from 2.24 to 10.55 percent from observed data and for 12 weeks, R was 63.00 percent with deviations ranging from 2.19 to 2221.91 percent. For data of 29 years, the R was 57.00 percent (11 weeks), having deviation from 2.75 to 10.19 per cent and R was 65.00 2percent for 12 weeks with deviation ranging from 0.36 to 11.37 percent from observed data. Looking to higher adjusted R (61.60 percent), lower S.E. (170.72), deviations (2.24 to 10.55), RMSE (158.28) and MAE (123.46) in prediction among all models, the model of 11 weeks using 2data of 28 years could be considered as pre harvest forecast model which can predict the productivity at 2 weeks before harvest with R value 67.20 percent.