Multi-Factor Intensity Estimation Method of Tropical Cyclone in the Western North Pacific with the Deviation Angle Variance Technique

• In this paper, the standardized infrared cloud images from Fengyun series geostationary satellites and Best Track Data from China Meteorological Administration (CMA-BST) within 2015–2017 are used to investigate the effects of two multi-factor models, generalized linear model (GLM) and long short-term memory (LSTM) model, for the tropical cyclone (TC) intensity estimation. The typical single-factor Sigmoid function model (SFM) with the map minimum value of deviation angle variance (DAV) is also reproduced for comparison. Through applying the sensitivity experiments to the DAV calculation radius and different training data groups, the estimation precision and optimum calculation radius for DAV in the western North Pacific (WNP) are analyzed. The results show that the root mean square error (RMSE) of the single-factor SFM is between 8.79 and 13.91 m s−1 by using the individual years as test sets and the remaining two years as training sets with the optimum calculation radius of 550 km. However, after selecting and using the high-correlation factors by GLM, the RMSE of GLM and LSTM models decreases to 5.93–8.68 and 4.99–7.00 m s−1 respectively with their own optimum calculation radius of 350 and 400 km. All sensitivity experiments indicate that the estimation results of SFM can be significantly influenced by the DAV calculation radius and characteristics of the training set data, while the results of multi-factor models appear more stable. Furthermore, the multi-factor models reduce the optimum radius within the process of DAV calculation and improve the precision of TC intensity estimation in the WNP, which can be an effective way for TC intensity estimation in marine area.
• Fig. 1.  The schematic of the deviation angle calculation process with the (a) cloud image (shaded; K) of Meranti (1614) and (b) corresponding deviation angle variance (DAV) map (shaded; deg2) at 0600 UTC 13 September 2016. Or is the reference point marked by the blue tringle. The blue circle denotes the calculation area within a given radius, taking 350 km as an example here. Two blue lines indicate the radial and Block-body Brightness Temperature (TBB) gradient direction of the given point A, and $\theta$ is their deviation angle. The blue square and red dot denote the location of the map minimum value (MMV) and recorded tropical cyclone (TC) center of Meranti. (c) The main region infrared cloud image (shaded; K) and (d) distribution of DAV (shaded; deg2) at the same time. The blue circles are centered on the position of each recording TC from the Best Track Data from China Meteorological Administration (CMA-BST) at that time with the radius of 350 km. The TC name and its intensity level are marked in red above each blue circle.

Fig. 2.  (a) The Sigmoid-based fitting curve of the FMMV (deg2) obtained from different calculation radii in the G-all test with the maximum wind speed (${V}_{\max}$; m s−1) in the CMA-BST dataset. The scattered points refer to the median of FMMV values corresponding to different ${V}_{\max}$; the solid line is the fitting curve based on the Sigmoid function; and the seven colors represent the results with seven different radii for DAV calculation varying from 250 to 550 km at an interval of 50 km, respectively. (b) The distribution of the RMSE (labeled with the straight line; m s−1) of the test results calculated at different radii in the four groups of experiments. Different colors refer to different experiments; G-2015, G-2016, and G-2017 represent the RMSE of the test groups, respectively; and G-all represents the fitting RMSE of the training set.

Fig. 3.  (a) Average variance contribution rates of different radii in the all-factor model with G-all as the validation data. The dark bar represents the variance contribution of the single factor, and the light bar represents the cumulative variance contribution of all pairwise interactions involving this factor. (b) The RMSE change rate of the all-factor model after the elimination of certain factor. (c) The RMSE change with the increasing number of factors in the model at different calculation radii. Bars and lines with different colors represent the results with different calculation radii.

Fig. 4.  The schematic diagram of the structure of (a) LSTM cell and (b) 2-D LSTM chains. ${f}_{t}$, ${i}_{t}$, and ${o}_{t}$ denotes the forget, input, and output gates, respectively. $\sigma$ and tanh are activate functions, in which $\sigma$ represents the Sigmoid function while tanh represents the hyperbolic tangent function. ${x}_{t}$ is the input value at the current time. ${C}_{t}$ and ${h}_{t}$ are the state of the cell and hidden layer at the current moment, respectively.

Fig. 5.  Boxplots for biases of the estimation results in the (a) single-factor SFM, (b) multi-factor GLM, and (c) LSTM model, from the CMA-BST dataset at different stages of the TC. “A”, “B”, “C”, and “D” in the four boxplots of different colors at each stage of the TC correspond to the G-2015, G-2016, G-2107, and G-all tests, respectively.

Fig. 6.  Comparison of the TC intensity among CMA-BST records (black heavy line), SFM estimations (green heavy line), GLM estimations (blue heavy line), LSTM estimations (purple heavy line), and trends of four high-correlation factors after normalization (light lines with different colors as shown in the figure legend). Three cases [Dujuan (1521), Meranti (1614), and Talim (1718)], are taken as examples, whose life cycles are divided by red dashed lines.

•  [1] Briegel, L. M., and W. M. Frank, 1997: Large-scale influences on tropical cyclogenesis in the western North Pacific. Mon. Wea. Rev., 125, 1397–1413. doi: 10.1175/1520-0493(1997)125<1397:LSIOTC>2.0.CO;2. [2] Combinido, J. S., J. R. Mendoza, and J. Aborot, 2018: A convolutional neural network approach for estimating tropical cyclone intensity using satellite-based infrared images. Proceedings of the 2018 24th International Conference on Pattern Recognition, IEEE, Beijing, China, 1474–1480, doi: 10.1109/ICPR.2018.8545593. [3] Demaria, M., M. Mainelli, L. K. Shay, et al., 2005: Further improvements to the Statistical Hurricane Intensity Prediction Scheme (SHIPS). Wea. Forecasting, 20, 531–543. doi: 10.1175/WAF862.1. [4] Dvorak, V. F., 1975: Tropical cyclone intensity analysis and forecasting from satellite imagery. Mon. Wea. Rev., 103, 420–430. doi: 10.1175/1520-0493(1975)103<0420:TCIAAF>2.0.CO;2. [5] Graves, A., 2012: Long short-term memory. Supervised Sequence Labelling with Recurrent Neural Networks, A. Graves, Ed., Springer, Berlin, Heidelberg, 37–45, doi: 10.1007/978-3-642-24797-2_4. [6] Guo, Z., M. H. Wang, Y. Qian, et al., 2014: A sensitivity analysis of cloud properties to CLUBB parameters in the single-column Community Atmosphere Model (SCAM5). J. Adv. Model. Earth Syst., 6, 829–858. doi: 10.1002/2014MS000315. [7] Hill, K. A., and G. M. Lackmann, 2009: Influence of environmen-tal humidity on tropical cyclone size. Mon. Wea. Rev., 137, 3294–3315. doi: 10.1175/2009MWR2679.1. [8] Hubert, L. F., and A. Timchalk, 1969: Estimating hurricane wind speeds from satellite pictures. Mon. Wea. Rev., 97, 382–383. doi: 10.1175/1520-0493(1969)097<0382:EHWSFS>2.3.CO;2. [9] Knaff, J. A., D. P. Brown, J. Courtney, et al., 2010: An evaluation of Dvorak technique-based tropical cyclone intensity estimates. Wea. Forecasting, 25, 1362–1379. doi: 10.1175/2010WAF2222375.1. [10] LeCun, Y., Y. Bengio, and G. Hinton, 2015: Deep learning. Nature, 521, 436–444. doi: 10.1038/nature14539. [11] Lee, J., J. G. Im, D. H. Cha, et al., 2020: Tropical cyclone intensity estimation using multi-dimensional convolutional neural networks from geostationary satellite data. Remote Sens., 12, 108. doi: 10.3390/rs12010108. [12] McCullagh, P., 1984: Generalized linear models. Eur. J. Oper. Res., 16, 285–292. doi: 10.1016/0377-2217(84)90282-0. [13] Olander, T. L., and C. S. Velden, 2007: The advanced Dvorak technique: Continued development of an objective scheme to estimate tropical cyclone intensity using geostationary infrared satellite imagery. Wea. Forecasting, 22, 287–298. doi: 10.1175/WAF975.1. [14] Piñeros, M. F., E. A. Ritchie, and J. S. Tyo, 2008: Objective measures of tropical cyclone structure and intensity change from remotely sensed infrared image data. IEEE Trans. Geosci. Remote Sens., 46, 3574–3580. doi: 10.1109/TGRS.2008.2000819. [15] Piñeros, M. F., E. A. Ritchie, and J. S. Tyo, 2010: Detecting tropical cyclone genesis from remotely sensed infrared image data. IEEE Geosci. Remote Sens. Lett., 7, 826–830. doi: 10.1109/LGRS.2010.2048694. [16] Piñeros, M. F., E. A. Ritchie, and J. S. Tyo, 2011: Estimating tropical cyclone intensity from infrared image data. Wea. Forecasting, 26, 690–698. doi: 10.1175/WAF-D-10-05062.1. [17] Ritchie, E. A., and G. J. Holland, 1999: Large-scale patterns associated with tropical cyclogenesis in the western Pacific. Mon. Wea. Rev., 127, 2027–2043. doi: 10.1175/1520-0493(1999)127<2027:LSPAWT>2.0.CO;2. [18] Ritchie, E. A., K. M. Wood, O. G. Rodríguez-Herrera, et al., 2014: Satellite-derived tropical cyclone intensity in the North Pacific Ocean using the deviation-angle variance technique. Wea. Forecasting, 29, 505–516. doi: 10.1175/WAF-D-13-00133.1. [19] Sun, Z. C., Y. B. Tao, S. Li, et al., 2013: Statistical strategies for constructing health risk models with multiple pollutants and their interactions: Possible choices and comparisons. Environ. Health, 12, 85. doi: 10.1186/1476-069X-12-85. [20] Wood, K. M., O. G. Rodríguez-Herrera, E. A. Ritchie, et al., 2015: Tropical cyclogenesis detection in the North Pacific using the deviation angle variance technique. Wea. Forecasting, 30, 1663–1672. doi: 10.1175/WAF-D-14-00113.1. [21] Yuan, M., and W. Zhong, 2019: Detecting intensity evolution of the western North Pacific super typhoons in 2016 using the deviation angle variance technique with FY data. J. Meteor. Res., 33, 104–114. doi: 10.1007/s13351-019-8064-7. [22] Zhang, C. J., J. F. Qian, L. M. Ma, et al., 2016: Tropical cyclone intensity estimation using RVM and DADI based on infrared brightness temperature. Wea. Forecasting, 31, 1643–1654. doi: 10.1175/WAF-D-15-0100.1. [23] Zhang, C. J., Q. Luo, L. J. Dai, et al., 2019: Intensity estimation of tropical cyclones using the relevance vector machine from infrared satellite image data. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens., 12, 763–773. doi: 10.1109/JSTARS.2019.2894654. [24] Zhang, Q. H., and C. Y. Guo, 2008: Overview of the studies on tropical cyclone genesis. Acta Oceanol. Sin., 30, 1–11. (in Chinese).
•  [1] Cheng Zhaohui, Kang Di, Chen Lianshou, Xu Xiangde. INTERACTION BETWEEN TROPICAL CYCLONE AND MEIYU FRONT*. Journal of Meteorological Research, [2] Lijuan CHEN, Zhensong GONG, Jie WU, Weijing LI. Extremely Active Tropical Cyclone Activities over the Western North Pacific and South China Sea in Summer 2018: Joint Effects of Decaying La Niña and Intraseasonal Oscillation. Journal of Meteorological Research,  doi: 10.1007/s13351-019-9009-x. [3] LEI Xiaotu, CHEN Lianshou. DYNAMICAL STUDY OF DIABATIC HEATING INFLUENCE ON RADIAL INHOMOGENEOUS STRUCTURE OF TROPICAL CYCLONE*. Journal of Meteorological Research, [4] Minmin WU, Lei WANG, Baiyang CHEN. Recent Weakening in Interannual Variability of Mean Tropical Cyclogenesis Latitude over the Western North Pacific during Boreal Summer. Journal of Meteorological Research,  doi: 10.1007/s13351-020-0039-1. [5] Meng YUAN, Wei ZHONG. Detecting Intensity Evolution of the Western North Pacific Super Typhoons in 2016 Using the Deviation Angle Variance Technique with FY Data. Journal of Meteorological Research,  doi: 10.1007/s13351-019-8064-7. [6] YU Hui, HU Chunmei, JIANG Leyi. Comparison of Three Tropical Cyclone Intensity Datasets. Journal of Meteorological Research, [7] CHEN Peiyan, DUAN Yihong, YU Hui, HU Chunmei. Application of Equivalent Black Body Temperature in the Forecast of Tropical Cyclone Intensity. Journal of Meteorological Research, [8] CHEN Peiyan, YU Hui, Johnny C. L. CHAN. A Western North Pacific Tropical Cyclone Intensity Prediction Scheme. Journal of Meteorological Research, [9] YU Hui, DUAN Yihong, LIANG Xudong. Effect of Horizontal Nonuniformity of Diabatic Heating on Tropical Cyclone Intensity and Structure. Journal of Meteorological Research, [10] Duan Yihong, Qin Zenghao, Gu dianfeng, Li Yongping. NUMERICAL STUDY ON THE EFFECTS OF SEA SURFACE TEMPERATURE ON TROPICAL CYCLONE INTENSITY-PART I:NUMERICAL EXPERIMENT OF THE TROPICAL CYCLONE INTENSITY RELATED TO SST *. Journal of Meteorological Research, [11] Liwen WANG, Suhong MA. Effect of Initial Vortex Intensity Correction on Tropical Cyclone Intensity Prediction: A Study Based on GRAPES_TYM. Journal of Meteorological Research,  doi: 10.1007/s13351-020-9093-y. [12] SUN Yuan, ZHONG Zhong, HA Yao, WANG Yuan, WANG Xiaodan. The Dynamic and Thermodynamic Effects of Relative and Absolute Sea Surface Temperature on Tropical Cyclone Intensity. Journal of Meteorological Research,  doi: 10.1007/s13351-013-0105-z. [13] DUAN Yihong, WU Rongsheng, YU Runling, LIANG Xudong. Numerical Simulation of Changes in Tropical Cyclone Intensity Using a Coupled Air-Sea Model. Journal of Meteorological Research, [14] DUAN Yihong, QIN Zenghao, GU Jianfeng, LI Yongping. NUMERICAL STUDY ON THE EFFECTS OF SEA SURFACE TEMPERATURE ON TROPICAL CYCLONE INTENSITY-PART Ⅱ:COUPLING MODEL AND EXPERIMENT*. Journal of Meteorological Research, [15] Yuxing WANG, Yuan SUN, Qianfeng LIAO, Zhong ZHONG, Yijia HU, Kefeng LIU. Impact of Initial Storm Intensity and Size on the Simulation of Tropical Cyclone Track and Western Pacific Subtropical High Extent. Journal of Meteorological Research,  doi: 10.1007/s13351-017-7024-3. [16] Xiaojiang SONG, Yuejian ZHU, Jiayi PENG, Hong GUAN. Improving Multi-Model Ensemble Forecasts of Tropical Cyclone Intensity Using Bayesian Model Averaging. Journal of Meteorological Research,  doi: 10.1007/s13351-018-7117-7. [17] ZHENG Yongguang, CHEN Jiong, TAO Zuyu. Distribution Characteristics of the Intensity and Extreme Intensity of Tropical Cyclones Influencing China. Journal of Meteorological Research,  doi: 10.1007/s13351-014-3050-6. [18] CHU Huiyun, WU Rongsheng. Environmental Influences on the Intensity Change of Tropical Cyclones in the Western North Pacific. Journal of Meteorological Research,  doi: 10.1007/s13351-013-0307-4. [19] YU Yubin, YAO Xiuping. Thermodynamic Characteristics of Tropical Cyclones with Rapid Intensity Change over the Coastal Waters of China. Journal of Meteorological Research, [20] YUAN Jinnan, LIN Ailan, LIU Chunxia. Spatial and Temporal Variations of Tropical Cyclones at Different Intensity Scales over the Western North Pacific from 1945 to 2005. Journal of Meteorological Research,
通讯作者: 陈斌, bchen63@163.com
• 1.

沈阳化工大学材料科学与工程学院 沈阳 110142

Multi-Factor Intensity Estimation Method of Tropical Cyclone in the Western North Pacific with the Deviation Angle Variance Technique

Corresponding author: Meng YUAN, yuanm2012@foxmail.com;
• 1. College of Meteorology and Oceanology, National University of Defense Technology, Nanjing 211101
• 2. Submarine College of Navy, Qingdao 266199
Funds: Supported by the National Key Research and Development Project of China (2018YFC1507402) and National Natural Science Foundation of China (42075011)

Abstract: In this paper, the standardized infrared cloud images from Fengyun series geostationary satellites and Best Track Data from China Meteorological Administration (CMA-BST) within 2015–2017 are used to investigate the effects of two multi-factor models, generalized linear model (GLM) and long short-term memory (LSTM) model, for the tropical cyclone (TC) intensity estimation. The typical single-factor Sigmoid function model (SFM) with the map minimum value of deviation angle variance (DAV) is also reproduced for comparison. Through applying the sensitivity experiments to the DAV calculation radius and different training data groups, the estimation precision and optimum calculation radius for DAV in the western North Pacific (WNP) are analyzed. The results show that the root mean square error (RMSE) of the single-factor SFM is between 8.79 and 13.91 m s−1 by using the individual years as test sets and the remaining two years as training sets with the optimum calculation radius of 550 km. However, after selecting and using the high-correlation factors by GLM, the RMSE of GLM and LSTM models decreases to 5.93–8.68 and 4.99–7.00 m s−1 respectively with their own optimum calculation radius of 350 and 400 km. All sensitivity experiments indicate that the estimation results of SFM can be significantly influenced by the DAV calculation radius and characteristics of the training set data, while the results of multi-factor models appear more stable. Furthermore, the multi-factor models reduce the optimum radius within the process of DAV calculation and improve the precision of TC intensity estimation in the WNP, which can be an effective way for TC intensity estimation in marine area.

Reference (24)

Catalog

/

DownLoad:  Full-Size Img  PowerPoint