Monsoon Rain Amount for Marathawada in the Year 2024

Marathawada has problems with rain all the time. Crop failures are very frequent. This poses serious problems to the farmers because they borrow money at a high interest rate from the banks or from private money lenders. Due to the uncertainty of the rainfall, they have difficulty in planning for the planting of crops.

To reduce the uncertainty- this research has been taken up where several methods are used to predict the rainfall. These methods are (a) the Fast Fourier Transform (FFT) method, (b) the Artificial Neural Network (ANN) method, (c) the Time Series method, and (d) the Root Mean Square (RMS) method. The predicted value is equal to the average of these four methods. The advantage of using these four methods is that the calculations can be done about seven months in advance to give sufficient time to the farmers for planning. The methods involve extensive computations detailed in references 46 to 48. Here, the data available or the past rain records are used to make prediction for the coming year. In the Time Series method – each of the monsoon months -June, July, August, and September are considered as separate season of the given year and overall regression line is arrived at. On the other hand, in the RMS – the regression line for each of the methods is found and then the projection is made for the coming year.

In the Fast Fourier Transform method, the Fourier series curve for each of the months is arrived at by computing its coefficients using a faster algorithm which is normally very computation intensive. From this series, a trend is computed and then the projection is made.

In the Neural Network method, which these days are also called Artificial Intelligence method – the network is trained first using the existing data and then based on this training which is essentially to compute the weight matrix due to this training. After computing the weight matrix, basecd on the input of the existing rain record, the rain amount for the coming year is calculated.