Application of Informer Model Based on SPEI for Drought Forecasting

To increase the accuracy of drought prediction, this study proposes a forecasting method based on Informer model. Taking Yellow River Basin as an example, accuracies Autoregressive Integrated Moving Average (ARIMA), Long Short-Term Memory (LSTM), and models multiple timescales Standardized Precipitation Evapotranspiration Index (SPEI) were compared analyzed. The results indicate that, with increasing timescale, ARIMA, LSTM, improved gradually, reaching best 24-month timescale. However, predicted values well those significantly different from true SPEI 1-month model was more accurate than ARIMA LSTM all timescales, indicating that can widely capture information input series over time is effective in long-term prediction problems. Furthermore, enhance precision prediction. closer to values, forecasted trends complied actual trends. adaptively and, therefore, better relevance timecales. NSE for four meteorological stations SPEI24 0.968, 0.974, 0.972, 0.986.