傳統上，海洋學的研究係以溫度與鹽度(Temperature-Salinity，T-S)關係來表示水團的特徵。然而，溫度與鹽度特性可能會隨著地理位置、年度、季節、或水層的不同而變化；即使在同一地點，其T-S特性也會受到深度很大的影響。因此，本論文研究了鹽度預測模型，並加入考慮深度，希望了解深度是否可以提高預測的準確性。我們分別使用多項式迴歸模型、以及多變數非線性迴歸模型來建立鹽度預測模型。實驗結果顯示，加入深度對於提高鹽度準確性非常有效，並且季節相關模型可以獲得比季節無關模型更好的預測效能。在對五年期資料進行分析時，也發現五年期預測效能明顯高於所有年份的結果；這顯示了台灣附近的水團特性很可能存在著長期的變化。另外，基於上述模型，本論文也提出了水團特徵的距離度量方法，並用來產生大海洋區域的相似圖。經過與黑潮樣式較早的研究結果比對，我們成功地驗證迴歸模型對水團特性有良好的鑑別力。雖然迴歸模型可根據溫度與深度資訊來精確預測鹽度的變化，惟迴歸函數之變數組合較多，需要繁複的變數關係假設與實驗驗證。為了簡化建模與比對，我們進一步提出了以高斯混合模型對水團特性建模的方法；實驗結果顯示所得到的黑潮樣式相似圖具有很好的鑑別力。高斯混合模型屬於非監督式學習，對於連續性的多維度資料有很好的描述能力。在使用此模型時，並不需要先知道變數之間的關係，即可學習到具有鑑別力的水團特性，還可彈性地加入新的特徵因子。因此，在未來的研究中也可使用此模型直接分析更多變量的水團特性，而不需要預先找出變量間的關係。
基於上述分析方法，本論文也提出了3D路徑模型和視覺化架構，用以呈現深度對T-S特性的影響，並據此建立視覺化互動網站系統。此系統提供使用者具易用性的視覺化介面，能呈現深度資訊和多重視角，可更有效率地查詢、探索、比較、和解釋水團特性，例如解釋黑潮入侵南中國海的研究。互動系統也提供了有效率的查詢工具，允許使用者查詢特定T-S資料之大海洋區域相似度圖，並可以根據指定的季節或年期範圍，顯示相對應的T-S曲線圖和三維T-S-D圖。這些技術的研究與整合使得大範圍海洋水團的知識探索和比較性研究更為便利。

Traditionally, temperature-salinity (T-S) relationship is analysed to indicate the characteristic of water mass and plays an important role in oceanography. Temperature-salinity characteristic however might change dynamically with respect to the geographic location, season, or layer of water, and is quite sensitive to the depth even at the same location. It is therefore of interest whether including depth into the prediction model of salinity could help to improve the prediction accuracy. In this research, multivariate statistical models, including polynomial regression and multivariate nonlinear regression, are investigated to predict the salinity according to both temperature and depth. Experimental results show that depth is very effective for improving the prediction accuracy, and season-dependent model may achieve better performance than season-independent model. Additionally, when the analysis was conducted for five-year range, it is found the prediction accuracy is significantly higher than the result for all years, which indicates there might exist long-term variation on the characteristics of the water masses near Taiwan. Though multivariate regression model may predict the salinity well according to its relationship with temperature and depth, it is however quite cumbersome to conduct a lot of experiments for the many alternative regression functions combined from multiple variables. For simplification, Gaussian mixture model that learns in an unsupervised way is further proposed to model the characteristic of a water mass. This model is flexible since, it does not require any knowledge about the relationships among the variables, and allows to add new variables without making the model more complicated. In addition, the distances of the models can be measured easily and discriminatively, which may facilitate the comparison for a lot of the water masses. This model is hence potential to be used in the future research.
Furthermore, 3D path model and visualization scheme were proposed to explore and interpret the effect of depth on the temperature-salinity characteristic from different perspectives, and an interactive visualization system was built accordingly. This system may present the T-S curve and 3D path model according to the assigned criteria of season or multi-year range, and allow the user to view the similarity map for the given T-S-D data of water masses so as to conduct comparative study of water masses for a wide area of ocean.

Barker, P. A., Street-Perrott, F. A., Leng, M. J., Greenwood, P. B., Swain, D. L., Perrott, R. A., and Ficken, K. J. (2001), “A 14,000-Year Oxygen Isotope Record from Diatom Silica in Two Alpine Lakes on Mt. Kenya.” Science, 292 (5525), 2307–2310, doi: 10.1126/science.1059612.
Briggs SA, Blanar CA, Robblee MB, Boyko CB, Hirons AC (2017), “Host abundance, sea-grass cover, and temperature predict infection rates of parasitic isopods (Bopyridae) on caridean shrimp” Journal of Parasitology, 103:653–662., doi: 10.1645/16-126.
Butler H, Daly M, Doyle A, Gillies S, Hagen S, Schaub T (2016), “The geojson format” Technical report.
Boyer, T.P., O.K. Baranova, C. Coleman, H.E. Garcia, A. Grodsky, R.A. Locarnini, A.V. Mishonov, C.R. Paver, J.R. Reagan, D. Seidov, I.V. Smolyar, K. Weathers, M.M. Zweng (2018), World Ocean Database 2018. A.V. Mishonov, Technical Ed., NOAA Atlas NESDIS 87, 16.
Chen CTA (2005), “Tracing tropical and intermediate waters from the South China Sea to the Okinawa Trough and beyond” J Geophys Res, 110:C05012, doi: 10.1029/2004JC002494.
Chen CTA, S.-l. Wang (1998), “Influence of intermediate water in the western Okinawa Trough by the outflow from the south China sea.” Journal of Geophysical Research, 103: pp. 12683-12688, doi:10.1029/98JC00366.
David P. Woodruff (2014), “Sketching as a Tool for Numerical Linear Algebra.” Foundations and Trends in Theoretical Computer Science., Vol.10:No.1–2, pp1-157, doi: 10.1561/0400000060.
Dan R. Lipşa Robert S. Laramee Simon J. Cox Jonathan C. Roberts Rick Walker Michelle A. Borkin Hanspeter Pfister, (2012), “Visualization for the Physical Sciences.” Proc. Eurographics State of the Art Reports, pp. 49-73, doi: 10.1111/j.1467-8659.2012.03184.x.
D.L. Liu, Z.S. Xu, Y. Zhou, C.G. Fan, (2015), “Heat map visualisation of fire incidents based on transformed sigmoid risk model Fire Saf.” J., 109, p. 102863
Douglas C. Montgomery, Elizabeth A. Peck, G. Geoffrey Vining, (2015), “Introduction to Linear Regression Analysis.” John Wiley & Sons, 70 pp.
Emelianov, M., Font, J., Turiel, A., Millot, C., Solé, J., Poulain, P., Julià, A., Vitrià, M.R., (2006), “Transformation of Levantine Intermediate Water tracked by MedArgo floats in the Western Mediterranean.” Ocean Sci., 2, 281–290 pp.
Emery, W.J., Dewar, J.S., (1982), “Data Analysis Methods in Physical Oceanography.” Elsevier Science, 56pp.
Emery, W.J., Dewar, J.S., (1982), “Mean temperature‐salinity, salinity‐depth and temperature‐depth curves in the North Atlantic and North Pacific.” Progress in Oceanography, 16, Pergamon, 219–305.
Glantz, S. A. and Slinker, B. K, (1990), “Primer of applied regression and analysis of variance.” McGraw-Hill.
G. Nagino and M.Shozakai, (2006), “Distance Measure between Gaussian Distributions for Discriminating Speaking Styles.” Proc. ICSLP06, Pittsburg, pp. 657–660.
Hansen, D. V., & Thacker, W. C., (1999), “Estimation of salinity profiles in the upper ocean.” Journal of Geophysical Research , 104(C4), 7921–7933, doi: 10.1029/1999JC900015.
Levine, D. M., P. P. Ramsey, and R. K. Smidt, (2001), “Ap¬plied Statistics for Engineers and Scientists: Using Mi¬crosoft Excel & Minitab”, Pearson.
Liang, W.-D., Y. J. Yang*, T. Y. Tang, and W.-S. Chuang, (2008), “Kuroshio in the Luzon Strait.” Journal of Geophysical Reaserch, 113, doi: 10.1029/2007JC004609.
L.P. Heck, K.C. Chou, (1994), “Gaussian mixture model classifier for machine monitoring.” Proceedings of the IEEE world Congress on Computational Neural Network and International Conference on Intelligence, Vol. 7, pp. 4493–4496.
Mamayev, O. I., (1975), “Temperature–Salinity Analysis of World Ocean Waters.” Elsevier
K. Kim , K. Kim , T. Rhee , H. Rho , R. Limeburner and R. Beardsley K, (1991), “Identification of water masses in the Yellow Sea and the East China Sea by cluster analysis.” Elsevier Oceanography Series, 54 (C), 253-267.
Jan, S., Y. J. Yang, J. Wang, V. Mensah, T.-H. Kuo, M.-D. Chiou, C.-S. Chern, M.-H. Chang, and H. Chien, (2015), “Large variability of the Kuroshio at 23.75°N east of Taiwan.” J. Geophys. Res. Oceans, 120, 1825– 1840, doi:10.1002/2014JC010614.
Jian-Heng Wu, Bor-Shen Lin, and Jia-Yu Kuo, (2014), “Querying Similar Water Masses Visualization Tool Design and Implementation Based on Polynomial Regression.” Terr. Atmos. Ocean. Sci., Vol. 25, No. 5, 727-741, doi:10.3319/TAO.2014.03.17.01(Oc)
Johns WE, Lee TN, Zhang D, Zantopp R, Liu C-T, Yang Y, (2001), “The Kuroshio east of Taiwan: moored transport observations from the WOCE PCM-1 array.” J Phys Oceanogr. 31:1031–1053, doi:10.1175/1520-0485(2001)031<1031:TKEOTM>2.0.CO;2
Lien, R.-C., B. Ma, C.M. Lee, T.B. Sanford, V. Mensah, L.R. Centurioni, B.D. Cornuelle, G. Gopalakrishnan, A.L. Gordon, M.-H. Chang, and others, (2015), “The Kuroshio and Luzon Undercurrent east of Luzon Island.” Oceanography, 28(4), 54-63, doi:10.5670/oceanog.2015.81.
Lien-Kwei Chien, Tsung-Shen Feng, Chieh-Cheng Yen,Beng-Chun Lee and Heng-Wen Chang, (2010), “The Application of e-Technology for Marine Information Service.” Journal of Marine Science and Technology, 18, 797-808.
Mamayev O.I., (1975), “Temperature-salinity analysis of world ocean waters.”, Elsevier, Amsterdam, 247 pp.
Moore, D.S., Notz, W.I. and Fligner, M.A., (2015), “The Basic Practice of Statistics”, 7th ed., Macmillan,New York,
Mensah, V., Jan, S., Chiou, M., Kuo, T., Lien, R., (2014), “Evolution of the Kuroshio Tropical Water from the Luzon Strait to the east of Taiwan.” Deep Sea Research Part I: Oceanographic Research Papers, 86, 68-81, doi:10.1016/j.dsr.2014.01.005.
Mensah V, Jan S, Chang M-H, Yang Y-J, (2015), “Intraseasonal to seasonal variability of the intermediate waters along the Kuroshio path east of Taiwan.” J Geophys Res, 120:5473–5489, doi:10.1002/2015JC010768.
Nan, F., Xue, H. & Yu, F., (2015), “Kuroshio intrusion into the South China Sea: A review.” Progress in Oceanography, 137, 314–333, doi: 10.1016/j.pocean.2014.05.012.
Jenkins WJ, Smethie WM Jr, Boyle EA, Cutter GA., (2015), “Water mass analysis for the U.S. GEOTRACES (GA03) North Atlantic sections.” Deep Sea Res. II Top. Stud. Oceanogr., 116, 6–20, doi: 10.1016/j.dsr2.2014.11.018.
Ocean Data Bank, (2020),” Ocean Data Bank of the Ministry of Science and Technology, Republic of China.” Available at http://www.odb.ntu.edu.tw/.
P. C. Woodland, J. J. Odell, V. Valtchev, and S. J. Young, (1994), Large vocabulary continuous speech recognition using HTK. Proc. IEEE ICASSP., vol. 2, 1994, pp. II/125–II/128.
Qi Jifeng, Yin Baoshu, Zhang Qilong, Yang Dezhou, XU Zhenhua (2014), “Analysis of seasonal variation of water masses in East China Sea.” Chinese Journal of Oceanology and Limnology. Vol. 32 No. 4, P. 958-971
QU, T., MITSUDERA, H., and YAMAGATA, T. (2000), “Intrusion of the North Pacific waters into the South China Sea.” Journal of Geophysical Research. ,105, 6415–6424
Razi MA, Athappilly K, (2005) , “A comparative predictive analysis of neural networks (NNs), nonlinear regression and classification and regression tree (CART) models.” Expert Syst Appl 29:65–74
Rufu Qin, Bin Feng, Zhounan Xu, Yusheng Zhou, Lixin Liu, Yineng Li (2020), “Web-based 3D visualization framework for time-varying and large-volume oceanic forecasting data using open-source technologies. Environmental Modelling & Software.”, Elsevier Science vol 135, 104908.
Richard B. Darlington, Andrew F. Hayes, (2017), “Regression analysis and linear models: Concepts, applications, and implementation.”, The Guilford Press, 341pp.
Schmidt,E. and Stewart,G., (1992), “On the early history of the singular value decomposition.” Univeristy of Maryland. SIAM Rev 35(4):551–566
Shaw, P., Greenstein, D., Lerch, J., Clasen, L., Lenroot, R., Gogtay, N., Evans A., Rapoport J., and Giedd, J.s, (2006), “Intellectual ability and cortical development in children and adolescents.” Nature, 440 (7084), 676–679, doi: 10.1038/nature04513.
Stigler, Stephen M., (1971) “Optimal experimental design for polynomial regression.” Optimal experimental design for polynomial regression., 66, 334, 311–318, doi: 10.1080/01621459.1971.10482260.
Teague, William J., Michael J. Carron, and Patrick J. Hogan, (1990), “A comparison between the generalized environmental model and Levitus climatologies.” Journal of Geophysical Research: Oceans (1978–2012), 95(C5), 7167–7183, doi: 10.1029/JC095iC05p07167.
Troccoli A, Haines K, (1999), “Use of temperature-salinity relation in a data assimilation context.” J Atmos Oceanic Technol, 16:2011–2025, doi:10.1175/1520-0426(1999)016<2011:UOTTSR>2.0.CO;2.
V. Epitropou, T. Bassoukos, K. Karatzas, A. Karppinen, L. Wanner, S. Vrochidis, I. Kompatsiaris, J. Kukkonen, (2016), “Environmental data extraction from heatmaps using the air merge system Multimed.” Tool. Appl., 75, 1589-1613
Xie, C., Li, M.K., Wang, H.Y., Dong, J.Y., (2019), “A survey on visual analysis of ocean data.” Visual Informatics. 3 (3), 113–128.
Yang H, Griffiths PR, Tate JD, (2003), “Comparison of partial least squares regression and multi-layer neural networks for quantification of nonlinear systems and application to gas phase Fourier transform infrared spectra.” Analytica Chimica Acta 489:125–136
Yang, Y.J., S. Jan, M.-H. Chang, J. Wang, V. Mensah, T.-H. Kuo, C.-J. Tsai, C.-Y. Lee, M. Andres, L.R. Centurioni, and others, (2015), “Mean structure and fluctuations of the Kuroshio East of Taiwan from in situ and remote observations.” Oceanography. 28(4):74–83, doi:10.5670/oceanog.2015.83.