Thumbnail Image
Journal Title
Journal ISSN
Volume Title
Acharya N.G. Ranga Agricultural University
Rainfall is one of the important climatic factor that influences crop production in particular and agriculture in general. Here, the distribution of rainfall can be studied by fitting suitable statistical distribution to the monthly rainfall data recorded over 15 years (2000-01 to 2014-15) for Prakasam district in Andhra Pradesh. It revealed that most of the months follow ‘Type I’ (or) Beta distribution while the months of July and September assumed Pearsonian “Type II” (or) “Type VII” i.e., Normal distribution and the month of November shown “Pearsonian Type IV” distribution. Statistically, estimation of a crop yield-weather relationship is fitting a multiple regression equation with yield as the dependent variable and weather parameters during the crop growth period as the independent variables. The analysis has been carried out on the basis of crop yields and weather variables for 15 years of monthly time series data (2000-01 to 2014-15). Weather impact on the crop yields was studied on the basis of rainfall, temperature (maximum and minimum) and relative humidity (AM and PM). In fitting the crop yield-weather relationships, the assumption of a continuous time trend was found to be inappropriate when the impact of new technology may exists in the form of quantum jumps over time which is termed as discrete time effect. For this situation, concept of control charts was applied using one sigma limits and sub-periods were identified. These sub-periods were formed with the year of quantum jump as the cut-off point. Nature of time trend in the sub-periods and overall yields was investigated by fitting time trend regression equation respectively. All the crops revealed ‘differential’ trend effect in the yields of the two sub-periods indicating that there could be a differential weather response of the crop. Hence, it is appropriate to fit the crop yield-weather relationships separately for each of the two sub-periods as well as for overall period. It was observed that an overall relationship may not be appropriate to explain the yield variations as it consisted of certain irrelevant regressors. Considering this behaviour, separate relationships were fitted for the different subperiods existing in the crop yield data and the analysis revealed the existence of a differential response of the yields to weather. The variables identified in these relations suitably explain the weather response with respect to the crop growth stages. Hence, it was concluded that yields under a given technology only could be forecasted (based on weather variables) on the basis of the corresponding subperiod relationship (equation).