Statistical Prediction of Typhoon-Induced Rainfall over China Using Historical Rainfall, Tracks, and Intensity of Typhoon in the Western North Pacific

Abstract

Typhoons or mature tropical cyclones (TCs) can affect inland areas of up to hundreds of kilometers with heavy rains and strong winds, along with landslides causing numerous casualties and property damage due to concentrated precipitation over short time periods. To reduce these damages, it is necessary to accurately predict the rainfall induced by TCs in the western North Pacific Region. However, despite dramatic advances in observation and numerical modeling, the accuracy of prediction of typhoon-induced rainfall and spatial distribution remains limited. The present study offers a statistical approach to predicting the accumulated rainfall associated with typhoons based on a historical storm track and intensity data along with observed rainfall data for 55 typhoons affecting the southeastern coastal areas of China from 1961 to 2017. This approach is shown to provide an average root mean square error of 51.2 mm across 75 meteorological stations in the southeast coastal area of China (ranging from 15.8 to 87.3 mm). Moreover, the error is less than 70 mm for most stations, and significantly lower in the three verification cases, thus demonstrating the feasibility of this approach. Furthermore, the use of fuzzy C-means clustering, ensemble averaging, and corrections to typhoon intensities, can provide more accurate rainfall predictions from the method applied herein, thus allowing for improvements to disaster preparedness and emergency response.

Effects of Learning Rates and Optimization Algorithms on Forecasting Accuracy of Hourly Typhoon Rainfall: Experiments With Convolutional Neural Network

“…Similarly, to forecast typhoon rainfall, some studies applied this distance as input for statistical models (Kim et al., 2020; Lonfat et al., 2007; Tuleya et al., 2007). Moreover, several studies have defined typhoon rainfall events that occurred within a 500 km radius from the typhoon center to show the contribution of typhoon rainfall to total rainfall (F. J. Chen & Fu, 2015; Dare et al., 2012; Hernández Ayala & Matyas, 2018; Jiang & Zipser, 2010; Larson et al., 2005; Lau et al., 2008; Lavender & Abbs, 2013; Lee et al., 2010; Nogueira & Keim, 2011; Prat & Nelson, 2013a, 2013b; Schreck et al., 2011; Uddin et al., 2019).…”

“…Similarly, to forecast typhoon rainfall, some studies applied this distance as input for statistical models (Kim et al, 2020;Lonfat et al, 2007;Tuleya et al, 2007). Moreover, several studies have defined typhoon rainfall events that occurred within a 500 km radius from the typhoon center to show the contribution of typhoon rainfall to total rainfall (F. J.…”

An Improved Cluster‐Wise Typhoon Rainfall Forecasting Model Based on Machine Learning and Deep Learning Models Over the Northwestern Pacific Ocean

“…There are mainly two reasons for the best performance of the current study. First, the input variable-typhoon rainfall within a 500 km radius-helps this study to yield more accurate forecasts than existing models, as earlier studies (e.g., Kim et al, 2020;Lonfat et al, 2007;Tuleya et al, 2007) confirmed the advantage of this input in their statistical models. Second, the parameterization technique (grid-search cross-validation) helps this study to find the best parameters for SVM, RF, AdaBoost, CNN, and LSTM models, as in previous studies (e.g., Noori et al, 2011;Wu et al, 2008Wu et al, , 2019Yu et al, 2006) reported in their flood and streamflow forecasting models that this method was simple, robust, and time-saving.…”

“…The reason to use a 500 km radius from the typhoon center as input is that the range of the typhoon's primary wind circulation region (80-400 km radius) and the curved typhoon cloud shield (550-600 km radius) is consistent with a 500 km radius from the typhoon center (Englehart & Douglas, 2001). The importance of using typhoon rainfall within this distance as input to forecast typhoon rainfall was confirmed by statistical models (Kim et al, 2020;Lonfat et al, 2007;Tuleya et al, 2007). Moreover, previous studies related to the contribution of typhoon rainfall to total rainfall also recognized typhoon rainfall events that occurred within a 500 km radius from the typhoon center (F. J.…”

Heavy Rainfall Forecast of Landfalling Tropical Cyclone Over China With an Upgraded DSAEF_LTP Model

“…This problem can be largely attributed to the lack of sufficient high-performance computing resources (Bachmann & Torn, 2021;Judt, 2018;Selz, 2019) and the limited ability to parameterize sophisticated TC rainfall physical processes in NWP models; significant progress in both cannot be achieved in a short period. Therefore, some scientists have aimed to fill in this gap by exploring TC rainfall forecasts with statistic-or analog-based techniques (Bagtasa, 2021;Li et al, 2015;Kim et al, 2019Kim et al, , 2020Ren et al, 2018;Wei, 2012aWei, , 2012bZhong et al, 2009).…”