A hybrid statistical downscaling prediction model with timescale decomposition for summer precipitation over Hainan Island, China
Junhu Zhao,Hui Wu,Kaiguo Xiong,Shaobo Qiao,Jie Yang,Guolin Feng,Yingfa Li,Caiying Xing,Jingjing Zhu
DOI: https://doi.org/10.1002/joc.8454
2024-03-30
International Journal of Climatology
Abstract:The multiple timescale variability of summer precipitation over Hainan Island (HNSP) is consistent with the theory of timescale decomposition approach. The interannual and interdecadal variability of the HNSP have respective predictors, such as North Atlantic tripole SST anomalies and Pacific Interdecadal Oscillation, respectively. The hybrid statistical downscaling prediction model effectively improved the forecasting skill of HNSP. By observing well‐known sea surface temperature (SST) indices and outputs from Climate Forecast System version 2 (CFSv2) hindcasts and predictions, in this study, we developed a hybrid statistical downscaling prediction model (HSDPM) based on a timescale decomposition approach, which effectively improved the forecasting ability of summer (June–August) precipitation over Hainan Island (HNSP) in China; this precipitation presented multiple obvious timescale variabilities. We found that the interannual variability of the HNSP is closely related to the observed North Atlantic tripole SST anomalies in the preceding winter and the summer sea level pressure over the South China Sea by CFSv2 predicted in March. On the other hand, the interdecadal variability in the HNSP is linked to the prewinter Pacific Interdecadal Oscillation (PDO) according to the observations, the CFSv2 predictions of the summer surface air temperature over the tropical central‐east Pacific. On this basis, both the interannual and interdecadal components of the HNSP are effectively predicted using the corresponding predictors via multiple regression; thus, the HSDPM is ultimately established by combining the above two components. Compared with the original CFSv2 model, the HSDPM model achieves a considerable improvement in performance in predicting the HNSP. Specifically, the temporal correlation coefficient, ratio of root‐mean‐square error and anomaly sign consistency rate between the observed and HSDPM‐predicted HNSPs are 0.70, 17.4% and 80.5%, respectively, which are significantly greater than the above three forecasting index values of 0.23, 38.3% and 48.8%, respectively, obtained from the original CFSv2 predictions. The application of the HSDPM may be beneficial for drought and flood prevention and mitigation in Hainan Island.
meteorology & atmospheric sciences