Comparison between Meterological Drought Indices for Upper Seonath Basin using ANN

Drought forecasting plays a crucial role in drought risk management. The current development of artificial neural networks (ANN) has a remarkable impact on forecasting of drought indices. This paper explains an approach to simulate quantitative values of drought indices using Artificial Neural Network. Drought Indices are continuous function of rainfall, temperature and other hydro-meteorological measurable variables. In this research two meteorological drought indices Effective Drought Index (EDI) and the Standardized Precipitation Index (SPI) are calculated & compared for monitoring drought in Upper Seonath Basin, both indices are the continuous function of rainfall which measures the degree of dryness of any period. A number of different ANN models for both EDI & SPI with the lead time of 12 months and different combinations of past rainfall have been tested at several rainfall stations in the Upper Seonath Basin. The validation of simulate quantitative value done by the using R2 & RMSE. The best models have R2 values of 0.87-0.99 for a lead time of 12 months. The structure of the model inputs (previous year rainfall drought index) does not vary with lead time.