STUDY ON METEOROLOGICAL DROUGHT FORECASTING IN THE YELLOW RIVER BASIN BASED ON MULTIVARIATE LSTM MODEL
Drought is one of the natural disasters with the most serious impact on human social development,and meteorological drought forecasting is an important direction of drought research.To improve the forecasting accuracy of meteorological drought index,a multivariate approach was applied to the process of forecasting the standardized precipitation evapotranspiration index(SPEI)in the Yellow River Basin by the long short-term memory(LSTM)model and compared with the results of the univariate LSTM model.Root mean square error,mean absolute error,and Nash efficiency index were used as evaluation indicators.The results show that in the forecasting of SPEI-1,SPEI-3,SPEI-6,SPEI-9,and SPEI-12 at Linxia,Taole,and Tongchuan stations in the Yellow River Basin,the values of the three evaluation indicators of the multivariate LSTM forecasting results are clearly better than those of the univariate LSTM forecasting results;and the visualization results also show that the forecasting curves of the multivariate LSTM method are closer to the observed value curves.The study proves the effectiveness and applicability of the multivariate LSTM model for improving the forecast accuracy of the meteorological drought index in the Yellow River Basin.
Yellow River Basinmeteorological droughtmultivariate forecastingLSTMstandardized precipitation evapotranspiration index
NETL
NSTL
万方数据
