This review summarizes the rapporteur report on tropical cyclone(TC)intensity change from the operational perspective,as presented to the 10th International Workshop on TCs(IWTC-10)held in Bali,Indonesia,from Dec.5–9...This review summarizes the rapporteur report on tropical cyclone(TC)intensity change from the operational perspective,as presented to the 10th International Workshop on TCs(IWTC-10)held in Bali,Indonesia,from Dec.5–9,2022.The accuracy of TC intensity forecasts issued by operational forecast centers depends on three aspects:real-time observations,TC dynamical model forecast guidance,and techniques and methods used by forecasters.The rapporteur report covers the progress made over the past four years(2018–2021)in all three aspects.This review focuses on the progress of dynamical model forecast guidance.The companion paper(Part II)summarizes the advance from operational centers.The dynamical model forecast guidance continues to be the main factor leading to the improvement of operational TC intensity forecasts.Here,we describe recent advances and developments of major operational regional dynamical TC models and their intensity forecast performance,including HWRF,HMON,COAMPS-TC,Met Office Regional Model,CMA-TYM,and newly developed HAFS.The performance of global dynamical models,including NOAA's GFS,Met Office Global Model(MOGM),JMA's GSM,and IFS(ECMWF),has also been improved in recent years due to their increased horizontal and vertical resolution as well as improved data assimilation systems.Recent challenging cases of rapid intensification are presented and discussed.展开更多
The challenges of applying deep learning(DL) to correct deterministic numerical weather prediction(NWP) biases with non-Gaussian distributions are discussed in this paper.It is known that the DL UNet model is incapabl...The challenges of applying deep learning(DL) to correct deterministic numerical weather prediction(NWP) biases with non-Gaussian distributions are discussed in this paper.It is known that the DL UNet model is incapable of correcting the bias of strong winds with the traditional loss functions such as the MSE(mean square error),MAE(mean absolute error),and WMAE(weighted mean absolute error).To solve this,a new loss function embedded with a physical constraint called MAE_MR(miss ratio) is proposed.The performance of the UNet model with MAE_MR is compared to UNet traditional loss functions,and statistical post-processing methods like Kalman filter(KF) and the machine learning methods like random forest(RF) in correcting wind speed biases in gridded forecasts from the ECMWF high-resolution model(HRES) in East China for lead times of 1–7 days.In addition to MAE for full wind speed,wind force scales based on the Beaufort scale are derived and evaluated.Compared to raw HRES winds,the MAE of winds corrected by UNet(MAE_MR) improves by 22.8% on average at 24–168 h,while UNet(MAE),UNet(WMAE),UNet(MSE),RF,and KF improve by 18.9%,18.9%,17.9%,13.8%,and 4.3%,respectively.UNet with MSE,MAE,and WMAE shows good correction for wind forces 1–3 and 4,but negative correction for 6 or higher.UNet(MAE_MR) overcomes this,improving accuracy for forces 1–3,4,5,and 6 or higher by 11.7%,16.9%,11.6%,and 6.4% over HRES.A case study of a strong wind event further shows UNet(MAE_MR) outperforms traditional post-processing in correcting strong wind biases.展开更多
文摘This review summarizes the rapporteur report on tropical cyclone(TC)intensity change from the operational perspective,as presented to the 10th International Workshop on TCs(IWTC-10)held in Bali,Indonesia,from Dec.5–9,2022.The accuracy of TC intensity forecasts issued by operational forecast centers depends on three aspects:real-time observations,TC dynamical model forecast guidance,and techniques and methods used by forecasters.The rapporteur report covers the progress made over the past four years(2018–2021)in all three aspects.This review focuses on the progress of dynamical model forecast guidance.The companion paper(Part II)summarizes the advance from operational centers.The dynamical model forecast guidance continues to be the main factor leading to the improvement of operational TC intensity forecasts.Here,we describe recent advances and developments of major operational regional dynamical TC models and their intensity forecast performance,including HWRF,HMON,COAMPS-TC,Met Office Regional Model,CMA-TYM,and newly developed HAFS.The performance of global dynamical models,including NOAA's GFS,Met Office Global Model(MOGM),JMA's GSM,and IFS(ECMWF),has also been improved in recent years due to their increased horizontal and vertical resolution as well as improved data assimilation systems.Recent challenging cases of rapid intensification are presented and discussed.
基金Supported by the National Key Research and Development Program of China (2021YFC3000905)Key Innovation Team Fund of China Meteorological Administration (CMA2022ZD04)。
文摘The challenges of applying deep learning(DL) to correct deterministic numerical weather prediction(NWP) biases with non-Gaussian distributions are discussed in this paper.It is known that the DL UNet model is incapable of correcting the bias of strong winds with the traditional loss functions such as the MSE(mean square error),MAE(mean absolute error),and WMAE(weighted mean absolute error).To solve this,a new loss function embedded with a physical constraint called MAE_MR(miss ratio) is proposed.The performance of the UNet model with MAE_MR is compared to UNet traditional loss functions,and statistical post-processing methods like Kalman filter(KF) and the machine learning methods like random forest(RF) in correcting wind speed biases in gridded forecasts from the ECMWF high-resolution model(HRES) in East China for lead times of 1–7 days.In addition to MAE for full wind speed,wind force scales based on the Beaufort scale are derived and evaluated.Compared to raw HRES winds,the MAE of winds corrected by UNet(MAE_MR) improves by 22.8% on average at 24–168 h,while UNet(MAE),UNet(WMAE),UNet(MSE),RF,and KF improve by 18.9%,18.9%,17.9%,13.8%,and 4.3%,respectively.UNet with MSE,MAE,and WMAE shows good correction for wind forces 1–3 and 4,but negative correction for 6 or higher.UNet(MAE_MR) overcomes this,improving accuracy for forces 1–3,4,5,and 6 or higher by 11.7%,16.9%,11.6%,and 6.4% over HRES.A case study of a strong wind event further shows UNet(MAE_MR) outperforms traditional post-processing in correcting strong wind biases.
