基于机器学习的热带气旋快速增强预报

doi:10.3969/j.issn.1001-909X.2024.03.008

摘要/Abstract

摘要：

极深对流云是热带气旋(tropical cyclone,TC)快速增强的前兆,为预报西北太平洋TC快速增强,该研究开发了一种使用极深对流云相关数据的机器学习模型。该机器学习模型整合了飓风强度统计预报快速增强指数(Statistical Hurricane Intensity Prediction Scheme-Rapid Intensification Index, SHIPS-RII)数据与TC中心300 km半径范围内极深对流云的覆盖面积。基于2011—2019年的数据,对24 h内TC增强超过30 kn和35 kn的快速增强事件分别进行了预报,相较于仅使用SHIPS-RII数据的模型,该机器学习模型在皮尔斯技能得分(PSS)方面分别提升了5.66%和9.58%,在检测概率指标(POD)方面分别提升了8.41%和8.55%。用该模型对典型台风杜鹃(Dujuan,2015)进行预报,其结果证明整合了极深对流云覆盖面积的模型在快速增强预报中具有优势,主要体现在TC初始强度较强时发生的快速增强预报。该模型对于强台风的预报具有较大的应用潜力。

关键词: 西北太平洋, 热带气旋, 快速增强, 云顶红外亮温(IR BT), 极深对流云, 机器学习, 台风杜鹃(Dujuan,2015), TC初始强度

Abstract:

Extremely deep convective clouds are a precursor to the rapid intensification (RI) of tropical cyclones (TCs). To predict the RI of TCs in the western North Pacific (WNP), a machine learning (ML) model using data related to extremely deep convective clouds was developed. The ML model integrates the Statistical Hurricane Intensity Prediction Scheme-Rapid Intensification Index (SHIPS-RII) data and the coverage area of extremely deep convective clouds within a 300 km radius of the center of the TC. Based on data from 2011 to 2019, the model forecasted RI events that increased by more than 30 kn and 35 kn within 24 h. Compared to models using only SHIPS-RII data, this ML model showed an improvement of 5.66% and 9.58% in the Peirce Skill Score (PSS), and a relative increase of 8.41% and 8.55% in the probability of detection (POD). This model is used to forecast typical typhoon Dujuan (2015), and the results show that the model integrating the coverage area of extremely deep convective clouds has advantages in RI predictions, which is mainly reflected in RI forecasts when the initial intensity is strong. The model has great application potential for forecasting strong typhoons.

Key words: western North Pacific, tropical cyclones, rapid intensification, infrared brightness temperature (IR BT), extremely deep convective clouds, machine learning, typhoon Dujuan(2015), initeal intensity of TC

中图分类号:

P732.4

罗通, 洪加诚. 基于机器学习的热带气旋快速增强预报[J]. 海洋学研究, 2024, 42(3): 99-107.

LUO Tong, HONG Jiacheng. Rapid intensification forecast of tropical cyclones based on machine learning[J]. Journal of Marine Sciences, 2024, 42(3): 99-107.

图/表 9

表1 SHIPS-RII预报因子

Tab.1 Predictive factors utilized in SHIPS-RII

序号	预报因子	定义
1	PER	前12 h TC强度变化
2	VMX0	最大持续风速(t=0 h)
3	POT	TC潜在强度(TC当前强度-最大可能强度)
4	OHC	海洋热含量
5	SHRD	滤波后500 km半径范围内850 hPa与200 hPa之间的水平风垂直切变
6	D200	1 000 km半径范围内200 hPa散度
7	TPW	500 km半径范围内以上切方向为中心的90°方位象限内总降水量小于45 mm的区域所占百分比
8	PC2	440 km半径范围内GOES-IR图像的第二主成分(t=0 h)
9	SDBT	50~200 km半径范围内GOES BT的标准差

图1 模型训练和测试流程图

Fig.1 Flowchart of model training and testing process

图2 2000—2019年西北太平洋TC中心300 km半径范围内极深对流云的平均亮温随TC强度变化率变化的合成图

Fig.2 Composite image showing the variation in average brightness temperature of extremely deep convective clouds within a 300 km radius of the Western North Pacific (WNP) TC center from 2000 to 2019, as a function of TC intensity change rate

图3 2000—2019年西北太平洋TC中心300 km半径范围内极深对流云覆盖面积比及平均亮温随TC强度变化率的变化趋势

Fig.3 Trends of coverage area ratio and average brightness temperature of extremely deep convective clouds with TC intensity change rate within a 300 km radius of the TC center in the WNP from 2000 to 2019

图4 SHIPS-RII模型和ML.Base模型对于西北太平洋 2011—2019年间TC快速增强的预测能力对比

Fig.4 Comparison of prediction capabilities of the SHIPS-RII model and the ML.Base model for TC rapid intensification in WNP from 2011 to 2019

图5 ML.Base模型和ML.DC200模型对于西北太平洋2011—2019年间TC快速增强的预测能力对比

Fig.5 Comparison of prediction capabilities of the ML.Base model and the ML.DC200 model for TC rapid intensification in WNP from 2011 to 2019

图6 ML.Base模型和ML.DC300模型对于西北太平洋2011—2019年间TC快速增强的预测能力对比

Fig.6 Comparison of prediction capabilities of the ML.Base model and the ML.DC300 model for TC rapid intensification in WNP from 2011 to 2019

图7 2015年台风杜鹃增强期间的轨迹 (图中RI代表快速增强开始的位置,蓝色线表示快速增强,红色线表示非快速增强。)

Fig.7 The track during typhoon Dujuan’s intensification in 2015 (RI represents where rapid intensification begins, blue line indicates rapid intensification and red line indicates non-rapid intensification.)

图8 典型台风杜鹃生命周期内TC中心300 km半径范围内极深对流云覆盖面积和台风强度随时间的变化 (图中RI代表快速增强开始的位置。)

Fig.8 Variation of coverage area of extremely deep convective clouds within a 300 km radius of the TC center and TC intensity during the life cycle of typical typhoon Dujuan (RI represents where rapid intensification begins.)

参考文献 23

[1]	张娇艳, 吴立广, 张强. 全球变暖背景下我国热带气旋灾害趋势分析[J]. 热带气象学报, 2011, 27(4):442-454.
	ZHANG J Y, WU L G, ZHANG Q. Tropical cyclone damages in China under the background of global warming[J]. Journal of Tropical Meteorology, 2011, 27(4): 442-454.
[2]	黄昌兴, 周国良, 郑磊, 等. 登陆我国台风的时空分布特征及其影响[J]. 水文, 2014, 34(6):81-85.
	HUANG C X, ZHOU G L, ZHENG L, et al. Spatial and temporal distribution characteristics of landing typhoons in China and their influence[J]. Journal of China Hydrology, 2014, 34(6): 81-85.
[3]	CANGIALOSI J P, BLAKE E, DEMARIA M, et al. Recent progress in tropical cyclone intensity forecasting at the national hurricane center[J]. Weather and Forecasting, 2020, 35(5): 1913-1922.
[4]	TRABING B C, BELL M M. Understanding error distributions of hurricane intensity forecasts during rapid intensity changes[J]. Weather and Forecasting, 2020, 35(6): 2219-2234.
[5]	KAPLAN J, DEMARIA M. Large-scale characteristics of rapidly intensifying tropical cyclones in the North Atlantic Basin[J]. Weather and Forecasting, 2003, 18(6): 1093-1108.
[6]	HENDRICKS E A, PENG M S, FU B, et al. Quantifying environmental control on tropical cyclone intensity change[J]. Monthly Weather Review, 2010, 138(8): 3243-3271.
[7]	CHEN H, ZHANG D L. On the rapid intensification of hurricane Wilma (2005). part II: Convective bursts and the upper-level warm core[J]. Journal of the Atmospheric Sciences, 2013, 70(1): 146-162.
[8]	CHEN X M, WANG Y Q, FANG J, et al. A numerical study on rapid intensification of typhoon Vicente (2012) in the South China Sea. Part II: Roles of inner-core processes[J]. Journal of the Atmospheric Sciences, 2018, 75(1): 235-255.
[9]	HAZELTON A T, HART R E, ROGERS R F. Analyzing simulated convective bursts in two Atlantic hurricanes. Part II: Intensity change due to bursts[J]. Monthly Weather Review, 2017, 145(8): 3095-3117.
[10]	JIANG H Y. The relationship between tropical cyclone intensity change and the strength of inner-core convection[J]. Monthly Weather Review, 2012, 140(4): 1164-1176.
[11]	MOLINARI J, VOLLARO D. Rapid intensification of a sheared tropical storm[J]. Monthly Weather Review, 2010, 138(10): 3869-3885.
[12]	ROGERS R F, REASOR P D, ZHANG J A. Multiscale structure and evolution of hurricane Earl (2010) during rapid intensification[J]. Monthly Weather Review, 2015, 143(2): 536-562.
[13]	SHI D L, CHEN G H, XIE X R. Revisiting the relationship between tropical cyclone rapid intensification and the distribution of inner-core precipitation[J]. Geophysical Research Letters, 2023, 50(17): e2023GL104810.
[14]	GAO S, ZHANG W, LIU J, et al. Improvements in typhoon intensity change classification by incorporating an ocean coupling potential intensity index into decision trees[J]. Weather and Forecasting, 2016, 31(1): 95-106.
[15]	GENG H T, SHI D W, ZHANG W, et al. A prediction scheme for the frequency of summer tropical cyclone landfalling over China based on data mining methods[J]. Meteorological Applications, 2016, 23(4): 587-593.
[16]	MERCER A, GRIMES A. Diagnosing tropical cyclone rapid intensification using Kernel methods and reanalysis datasets[J]. Procedia Computer Science, 2015, 61: 422-427.
[17]	SU H, WU L T, JIANG J H, et al. Applying satellite observations of tropical cyclone internal structures to rapid intensification forecast with machine learning[J]. Geophysical Research Letters, 2020, 47(17): e2020GL089102.
[18]	RUAN Z X, WU Q Y. Precipitation, convective clouds, and their connections with tropical cyclone intensity and intensity change[J]. Geophysical Research Letters, 2018, 45(2): 1098-1105.
[19]	WU Q Y, HONG J C, RUAN Z X. Diurnal variations in tropical cyclone intensification[J]. Geophysical Research Letters, 2020, 47(23): e2020GL090397.
[20]	JANOWIAK J E, JOYCE R J, YAROSH Y. A real-time global half-hourly pixel-resolution infrared dataset and its applications[J]. Bulletin of the American Meteorological Society, 2001, 82(2): 205-217.
[21]	WU Q Y, RUAN Z X. Rapid contraction of the radius of maximum tangential wind and rapid intensification of a tropical cyclone[J]. Journal of Geophysical Research: Atmospheres, 2021, 126(3): e2020JD033681.
[22]	KAPLAN J, ROZOFF C M, DEMARIA M, et al. Evaluating environmental impacts on tropical cyclone rapid intensifi-cation predictability utilizing statistical models[J]. Weather and Forecasting, 2015, 30(5): 1374-1396.
[23]	KAPLAN J, DEMARIA M, KNAFF J A. A revised tropical cyclone rapid intensification index for the Atlantic and eastern North Pacific basins[J]. Weather and Forecasting, 2010, 25(1): 220-241.