Summer Precipitation Forecast Using an Optimized Artificial Neural Network with a Genetic Algorithm for Yangtze-Huaihe River Basin, China

Owing to the complexity of climate system and limitations numerical dynamical models, machine learning based on big data has been used for forecasting in recent years. In this study, we attempted use an artificial neural network (ANN) summer precipitation forecasts Yangtze-Huaihe River Basin (YHRB), eastern China. The major ANN employed here is standard backpropagation (BPNN), which was modified application YHRB. Using analysis predictors/factors atmospheric circulation sea surface temperature, calculated correlation coefficients between factors. addition, sorted top six factors forecasts. order obtain accurate forecasts, month (factor)-to-month (precipitation) forecast models were applied over training validation periods (i.e., months 1979–2011 2012–2019, respectively). We compared BPNN with using a genetic algorithm-based (GABP), support vector (SVM) multiple linear regression (MLR) after model period, found that GABP method best among above methods forecasting, mean absolute percentage error (MAPE) approximately 20% YHRB, substantially lower than BPNN, SVM MLR values. then selected month-to-month by summing up monthly precipitation, scale forecast, presents very successful performance terms evaluation measures. For example, basin-averaged MAPE anomaly rate reach 4.7% 88.3%, respectively, can be good recommendation future operational services. It appears temperatures (SST) some key areas dominate These results indicate potential applying