Improving TIGGE Precipitation Forecasts Using an SVR Ensemble Approach in the Huaihe River Basin

Abstract

Recently, the use of the numerical rainfall forecast has become a common approach to improve the lead time of streamflow forecasts for flood control and reservoir regulation. The control forecasts of five operational global prediction systems from different centers were evaluated against the observed data by a series of area-weighted verification and classification metrics during May to September 2015–2017 in six subcatchments of the Xixian Catchment in the Huaihe River Basin. According to the demand of flood control safety, four different ensemble methods were adopted to reduce the forecast errors of the datasets, especially the errors of missing alarm (MA), which may be detrimental to reservoir regulation and flood control. The results indicate that the raw forecast datasets have large missing alarm errors (MEs) and cannot be directly applied to the extension of flood forecasting lead time. Although the ensemble methods can improve the performance of rainfall forecasts, the missing alarm error is still large, leading to a huge hazard in flood control. To improve the lead time of the flood forecast, as well as avert the risk from rainfall prediction, a new ensemble method was proposed on the basis of support vector regression (SVR). Compared to the other methods, the new method has a better ability in reducing the ME of the forecasts. More specifically, with the use of the new method, the lead time of flood forecasts can be prolonged to at least 3 d without great risk in flood control, which corresponds to the aim of flood prevention and disaster reduction.

Analysis and Prediction of Terrestrial Water Storage Anomalies in the Lower Yangtze River Basin Based on an Improved Grey-Markov Model

“…Specifically, the training sample is randomly divided into five categories, and each category is used as the validation set, whereas the rest is used as the training set to search the best parameters through GA. All the SVR models used in this research are based on an open-source software LIBSVM developed by Lin Chih-Jen (LIBSVM, 2022). For more information about SVR, see Cai, Wang, and Li (2018) and Yu, Chen, and Chang (2006) .…”

Improved Rainfall Prediction through Nonlinear Autoregressive Network with Exogenous Variables: A Case Study in Andes High Mountain Region

“…e resulting model was used for predicting the rainfall in 2015 in the two study stations. Many studies in climate and meteorology have used SVM [12,13,24,25] and RF in prediction, so we use them here as a baseline to compare the performance of NARX.…”

“…Other approaches have used the outputs of global climate models to improve the seasonal and subseasonal forecasting [14][15][16] and the probabilistic forecasts for uncertainty quantification [17][18][19]. Other studies have applied artificial intelligence approach for rainfall prediction such as artificial neural networks (ANNs) [5,[10][11][12][20][21][22][23][24], support vector machine (SVM) [12,13,24,25], logistic regression [26], and random forest [27,28]. Also, other studies have investigated the optimum selection of predictors to improve forecast accuracy [29].…”

Floodwater Utilization Based on Reservoir Pre-Release Strategy Considering the Worst-Case Scenario

“…Zhong et al [28] established a probabilistic inflow forecasting scheme based on the numerical weather prediction (NWP) data by applying the Bayesian model averaging (BMA) method, which provided the basis for its application in reservoir operation. Cai et al [29,30] established an uncertainty analysis model under the Bayesian theory with a generalized probability density function and adopted the fuzzy probability to describe the fuzziness of the system. The new model can obtain the conditional probability distribution of the precipitation with high accuracy and reliability by the NWP as well as account for the fuzzy events in real-world flood control.…”