A Hybrid Dynamical‐Statistical Model for Advancing Subseasonal Tropical Cyclone Prediction Over the Western North Pacific

Abstract

Tropical cyclone (TC) genesis prediction at the extended-range to subseasonal timescale (a week to several weeks) is a gap between weather and climate predictions. The current dynamical prediction systems and statistical models show limited skills in TC genesis forecasting at the lead time of 1-3 weeks. A hybrid dynamical-statistical model is developed that reveals capability in predicting basin-wide TC frequency in every 10-day period over the western North Pacific at a 25-day forecast lead, which is superior to the statistical and dynamical model-based predictions examined in this study. In this hybrid model, the cyclogenesis counts for different TC clusters are predicted, respectively, using the statistical models in which the large-scale predictors associated with intraseasonal oscillation evolutions are provided by a dynamical model. A probabilistic map of TC tracks at the subseasonal timescale is further predicted by incorporating the climatological probability of track distributions of these TC clusters. Plain Language Summary Tropical cyclone (TC) is a highly destructive type of natural disaster. Extending forecast lead times and increasing the forecast accuracy of TC genesis and movements are the keys for disaster prevention and mitigation. However, TC predictions at the subseasonal timescale (10 days to several weeks in advance) have not reached a satisfactory level. Most dynamical prediction systems and statistical models show skills of 1-3 weeks for subseasonal TC genesis prediction. In this study, we developed a hybrid dynamical-statistical prediction approach for advancing the capability to predict TC frequency over the western North Pacific (WNP). Considering the close linkage between intraseasonal oscillation and WNP TC genesis, multiple linear regression models in which the intraseasonal dynamic and thermodynamic conditions serve as the predictors were constructed for different TC clusters over the WNP. We find that future TC genesis counts can be predicted once we obtain the information on intraseasonal predictors from a dynamical prediction system. This hybrid model shows good skills for basin-wide TC genesis prediction at the forecast lead time of 25 days, and is superior to the statistical and dynamical model-based predictions examined in this study. In addition, a probabilistic map of WNP TC trajectories is also skillfully predicted at the subseasonal timescale.

Extended-Range Forecast of Regional Persistent Extreme Cold Events Based on Deep Learning

“…The result was substituted into the trained model to obtain the predicted precipitation field. In addition, many scholars have proposed methods using dynamical statistics to predict meteorological elements for an extended range [18][19][20].…”

A New Approach to Skillful Seasonal Prediction of Southeast Asia Tropical Cyclone Occurrence

“…Statistical seasonal prediction of WNP TCs is often made through various predictors, which in principle represent the year‐to‐year variability of these environmental factors (e.g., Chen et al., 1998; Fan & Wang, 2009; Goh & Chan, 2010; Kim et al., 2017; Wang, Xiang, & Lee, 2013; Wang et al., 2019). Statistical methods can be further developed into hybrid statistical‐dynamical methods, in which predictors are extracted from the environment variables that are dynamically predicted by GCMs (e.g., Camp et al., 2019; Kim et al., 2017; Li et al., 2017; Qian et al., 2020; Zhan & Wang, 2016).…”

Enhancing Subseasonal Temperature Prediction by Bridging a Statistical Model With Dynamical Arctic Oscillation Forecasting

“…Therefore, in the present study, we propose a hybrid approach that combines the advantages of both statistical and dynamical models by employing dynamically forecasted phase information as input for the phase model. Hybrid approaches have been previously developed for regression models to generate predictions, such as genesis of tropical cyclones (e.g., Kim & Webster, 2010; Kim et al., 2017; Li et al., 2013; Qian et al., 2020) and Asian summer monsoon rainfall (Wang et al., 2015).…”