臺灣博碩士論文加值系統

為了解決傳統雷達降水估計法不易和地面降雨量連結在一起的困難，並利用四維時空的雙偏極化雷達產品來改進降雨估計的準確度，本研究發展了一套基於氣象雷達(Weather Radar)特徵及支持向量機(Support Vector Machine, SVM)的地面降雨估計方法，而為了便於嘗試各種雷達特徵組合，還建立了一套完整的批次處理流程，以支援雷達特徵的萃取與標記、特徵進一步計算與組合以及降雨估計模型訓練與測試等作業。
本研究使用了中央氣象局五分山(Wu-Fen-Shan)雷達站2012年~2014年間降雨較顯著的案例，除了各個不同仰角掃描得到的雷達回波(Reflectivity, Z)外，還用了差異反射率(Differential Reflectivity, ZDR)以及比相位偏移差(Specific Differential Phase, KDP)等產品做為特徵的來源，而要估計的目標則為臺北氣象站的地面降雨量。當我們把同一個小時內的數筆特徵取聚合平均後，再將10分鐘地面累積雨量數值劃分為六個區間作為類別標記，並以支持向量迴歸(SVR)進行實驗，針對仰角0.5度的KDP測試資料所能得到的最佳估計結果均方根誤差(RMSE)為0.54、相關係數(CC)為0.95，不過若將測試資料中極大雨量值(9.4mm)去掉，結果的均方根誤差變為0.5，相關係數則降為0.72；此外，本估計方法的表現大致上和傳統的R(KDP)差不多，並比R(ZDR,KDP)略佳。
總括來說，使用低仰角KDP的統計特徵，如：基於5x5平面特徵算出來的中間值、平均值及前五大數值等進行組合，並搭配支持向量迴歸(SVR)進行模型訓練，為本研究所能得到的最佳地面降雨估計解決方案；雖然本研究也有針對Z、ZDR及KDP，或各仰角的產品進行特徵組合，但未能得到較好的實驗結果。

Significant problems exist for traditional precipitation estimates from radar when it comes to comparing with rain gauges on the ground. To avoid the problem and better utilize the information in the four-dimensional structure of the atmosphere, this research proposes a precipitation estimation method based on support vector machine and regression in the hopes to improve the accuracy of precipitation estimates. To generate various feature sets as inputs for the experiments, a set of tools have been developed, including a radar feature extraction and labeling tool based on AWIPS II (Advanced Weather Interactive Processing System) and a library for further batched processing of the extracted features such as calculation of statistics and combinations of features from various elevations and/or products.
For experiments of this research, Taiwan Central Weather Bureau’s Wu-Fen-Shan weather radar products, including reflectivity (Z), differential reflectivity (ZDR) and specific differential phase (KDP) from rainy days in the years of 2012, 2013 and 2014 are used as the source of features, and the target location for precipitation estimation is Taipei weather station.
In one of the experiments, we first aggregated over the feature vectors within the same observation hour to derive a mean feature vector, then partitioned 10-minute precipitation accumulations into 6 classes as labels of the features, by using the resulting feature set along with support vector regression, the best result gives a root mean squared error of 0.54 and correlation coefficient of 0.95, corresponding to features derived from 0.5-degree elevation specific differential phase (KDP), when the extreme 10-minute precipitation value of 9.4(mm) in the test set is removed, the corresponding root mean squared error becomes 0.5 and correlation coefficient dropped to 0.72. Results from this reaearch also indicate the proposed support vector regression estimation method has about the same performance as traditional R(KDP) given by Sachidananda and Zrnic(1987) and is better than R(ZDR,KDP) by Ryzhkov and Zrnic(1995).
In summary, using combinations of statistics such as middle value, mean, and maximum 5 values calculated from 5x5 feature vectors of low-elevation KDP products, along with support vector regression, is the best solution found for radar precipitation estimation in this research. While combinations of Z、ZDR and KDP as well as combinations of various elevations are also tried, no significant improvements can be derived in this research.

謝誌 I
摘要 II
Abstract III
目錄 V
圖目錄 VII
表目錄 IX
第一章 緒論 1
第一節 研究動機 1
第二節 研究目的 3
第三節 論文組織與架構 4
第二章 文獻回顧 5
第一節 雷達降水估計在防洪監測的應用 5
第二節 傳統雷達降雨估計方式 6
第三節 雙偏極化雷達降雨估計 11
第四節 美國國家氣象局降雨估計系統(NMQ) 13
第五節 工程統計/機器學習方式 14
第六節 支持向量機與類神經網路 15
第七節 雷達降水估計的準確度評估 25
第三章 研究方法 26
第一節 研究方法與限制 26
第二節 雷達資料的萃取與標記 32
第三節 模型訓練與測試實驗方法 37
第四章 研究結果與討論 39
第一節 實驗軟硬體環境 39
第二節 系統架構設計 40
第三節 雷達基礎特徵萃取成果 64
第四節 特徵進一步處理及組合 70
第五節 模型訓練與測試成果分析 73
第五章 結論與建議 100
第一節 結論 100
第二節 建議與未來展望 101
參考文獻 105

1. Laws, J.O. and D.A. Parsons, The relation of raindrop-size to intensity. Transactions, American Geophysical Union, 1943. 24(2): p. 452-460.
2. Marshall, J.S., R.C. Langille, and W.M.K. Palmer, MEASUREMENT OF RAINFALL BY RADAR. Journal of Meteorology, 1947. 4(6): p. 186-192.
3. Bringi, V.N. and V. Chandrasekar, Polarimetric Doppler Weather Radar, Principles and Applications. 2001.
4. Marshall, J.S. and W.M.K. Palmer, THE DISTRIBUTION OF RAINDROPS WITH SIZE. Journal of Meteorology, 1948. 5(4): p. 165-166.
5. Battan, L.J., Radar observation of the atmosphere. L. J. Battan (The University of Chicago Press) 1973. PP X, 324; 125 figures, 21 tables. £7•15. Quarterly Journal of the Royal Meteorological Society, 1973. 99(422): p. 793-793.
6. Zawadzki, I. Factors affecting the precision of radar measurements of rain. in Conference on Radar Meteorology, 22 nd, Zurich, Switzerland. 1984.
7. Seliga, T.A. and V.N. Bringi, Potential Use of Radar Differential Reflectivity Measurements at Orthogonal Polarizations for Measuring Precipitation. Journal of Applied Meteorology, 1976. 15(1): p. 69-76.
8. Hall, M.P.M., et al., Rain drop sizes and rainfall rate measured by dual-polarization radar. Nature, 1980. 285(5762): p. 195-198.
9. Doneaud, A., et al., The Area-Time Integral as an Indicator for Convective Rain Volumes. Journal of Climate and Applied Meteorology, 1984. 23(4): p. 555-561.
10. Calheiros, R.V. and I. Zawadzki, Reflectivity-Rain Rate Relationships for Radar Hydrology in Brazil. Journal of Climate and Applied Meteorology, 1987. 26(1): p. 118-132.
11. Rongrui, X. and V. Chandrasekar, Development of a neural network based algorithm for rainfall estimation from radar observations. Geoscience and Remote Sensing, IEEE Transactions on, 1997. 35(1): p. 160-171.
12. Chen, C.-Y., et al., Improving debris flow monitoring in Taiwan by using high-resolution rainfall products from QPESUMS. Natural Hazards, 2007. 40(2): p. 447-461.
13. Ulbrich, C.W., Natural Variations in the Analytical Form of the Raindrop Size Distribution. Journal of Climate and Applied Meteorology, 1983. 22(10): p. 1764-1775.
14. 陳盈臻, 臺灣北部鋒面強降水個案之雨滴粒徑觀測比較研究. 2013.
15. Zhang, J., K. Howard, and J.J. Gourley, Constructing Three-Dimensional Multiple-Radar Reflectivity Mosaics: Examples of Convective Storms and Stratiform Rain Echoes. Journal of Atmospheric and Oceanic Technology, 2005. 22(1): p. 30-42.
16. Xin, L., G. Reuter, and B. Larochelle, Reflectivity‐rain rate relationships for convective rainshowers in Edmonton: Research note. Atmosphere-Ocean, 1997. 35(4): p. 513-521.
17. Kumar, L.S., et al., Tropical rain classification and estimation of rain from ZR (reflectivity-rain rate) relationships. Progress In Electromagnetics Research B, 2011. 32: p. 107-127.
18. 吳舜華, 利用雨滴譜儀分析不同降水系統之微物理特性研究. 2006.
19. Sachidananda, M. and D.S. Zrnić, Rain Rate Estimates from Differential Polarization Measurements. Journal of Atmospheric and Oceanic Technology, 1987. 4(4): p. 588-598.
20. Zrnić, D.S., et al., Sensitivity Analysis of Polarimetric Variables at a 5-cm Wavelength in Rain. Journal of Applied Meteorology, 2000. 39(9): p. 1514-1526.
21. Guifu, Z., J. Vivekanandan, and E. Brandes, A method for estimating rain rate and drop size distribution from polarimetric radar measurements. Geoscience and Remote Sensing, IEEE Transactions on, 2001. 39(4): p. 830-841.
22. 紀博庭;陳台琦;呂崇華, 雙偏極化雷達觀測參數反求雨滴粒徑分布及降雨估計. 大氣科學, 2007. 第三十六期第一號.
23. Zhang, J., et al., National Mosaic and Multi-Sensor QPE (NMQ) System: Description, Results, and Future Plans. Bulletin of the American Meteorological Society, 2011. 92(10): p. 1321-1338.
24. Collier, C.G., Applications of Weather Radar Systems: A Guide to Uses of Radar Data in Meteorology and Hydrology (Wiley-Praxis Series in Atmospheric Physics). 1996.
25. 雷鳳;周仲島, C頻偏振雷達總傳播相位偏移差面積雨量估計法之研究. 大氣科學, 2006. 第三十四期第二號.
26. Hsu, C.-W., C.-C. Chang, and C.-J. Lin, A practical guide to support vector classification. 2003.
27. Vapnik, V. and A. Lerner, Pattern Recognition using Generalized Portrait Method. Automation and Remote Control, 1963. 24.
28. Vapnik, V.N., Estimation of Dependences Based on Empirical Data (Series: Information Science and Statistics). 1979.
29. Vapnik, V.N., An overview of statistical learning theory. IEEE Transactions on Neural Networks, 1999. 10(5): p. 988-999.
30. Burges, C.C., A Tutorial on Support Vector Machines for Pattern Recognition. Data Mining and Knowledge Discovery, 1998. 2(2): p. 121-167.
31. Cortes, C. and V. Vapnik, Support-Vector Networks. Machine Learning, 1995. 20(3): p. 273-297.
32. Chih-Wei, H. and L. Chih-Jen, A comparison of methods for multiclass support vector machines. Neural Networks, IEEE Transactions on, 2002. 13(2): p. 415-425.
33. Vapnik, V.N., The nature of statistical learning theory. 1995: Springer-Verlag New York, Inc. 188.
34. Chang, C.-C. and C.-J. Lin, LIBSVM: A library for support vector machines. ACM Trans. Intell. Syst. Technol., 2011. 2(3): p. 1-27.
35. Rosenblatt, F., Principles of neurodynamics: perceptrons and the theory of brain mechanisms. 1962: Spartan Books.
36. Aydin, K., H. Direskeneli, and T.A. Seliga, Dual-Polarzation Radar Estimation of Rainfall Parameters Compared with Ground-Based Disdrometer Measurements: October 29, 1982 Central Illinois Expenment. Geoscience and Remote Sensing, IEEE Transactions on, 1987. GE-25(6): p. 834-844.
37. Ryzhkov, A.V., S.E. Giangrande, and T.J. Schuur, Rainfall Estimation with a Polarimetric Prototype of WSR-88D. Journal of Applied Meteorology, 2005. 44(4): p. 502-515.
38. Gourley, J.J. and B.E. Vieux, A Method for Evaluating the Accuracy of Quantitative Precipitation Estimates from a Hydrologic Modeling Perspective. Journal of Hydrometeorology, 2005. 6(2): p. 115-133.
39. Young, C.B., et al., An evaluation of NEXRAD precipitation estimates in complex terrain. Journal of Geophysical Research, 1999. 104(D16): p. 19691.
40. Austin, P.M., Relation between Measured Radar Reflectivity and Surface Rainfall. Monthly Weather Review, 1987. 115(5): p. 1053-1070.
41. Ryzhkov, A.V. and D.S. Zrnić, Comparison of Dual-Polarization Radar Estimators of Rain. Journal of Atmospheric and Oceanic Technology, 1995. 12(2): p. 249-256.
42. Multi-Radar/Multi-Sensor System (MRMS/q3)網站
http://mrms.ou.edu
43. 維基百科過度擬合(Overfitting)簡介網站
http://en.wikipedia.org/wiki/Overfitting
44. 五分山氣象雷達站介紹網頁http://www.cwb.gov.tw/V7/eservice/docs/overview/organ/stations/46685/
45. 臺北氣象站介紹網頁http://www.cwb.gov.tw/V7/eservice/docs/overview/organ/stations/46692/
46. 類神經網路與深度學習(Neural network and deep learning)電子書首頁 http://neuralnetworksanddeeplearning.com/
47. 筆者的Github首頁
https://github.com/joehuang74
48. 第二代氣象資料整合處理及顯示系統(Advanced Weather Interactive Processing System II, AWIPS II)
http://www.unidata.ucar.edu/software/awips2/
49. Reflectivity-Rainfall Rate Relationships In Operational Meteorology
　　http://www.srh.noaa.gov/tlh/?n=research-zrpaper