受全球暖化、氣候變遷的影響，極端氣候頻傳，導致強降雨及短期乾旱的頻率上升，降雨量的預測日顯重要，以利於更有效的水資源管理和提供人民預警，為了評估降雨量的多寡造成的濕潤或乾旱程度，本研究採用能基於多時間尺度量化降雨量的標準化降雨指標(Standardized precipitation index, SPI)對台灣地區基隆、淡水、台中、高雄、恆春、成功、花蓮和阿里山8個降雨測站進行評估，將8個測站自西元1951至2014年之月雨量數據，結合觀測聖嬰現象之Niño區域的月平均海表面溫度，使用近期新發展的遞迴神經網路中的長短期記憶模型與門控遞迴單元，對8個降雨測站做出不同時間尺度(1、3、6與12個月)之SPI值，並與多元線性迴歸做出比較。各測站之SPI-1在測試集三種模型下，皆無法順利建立良好的模型，但是各測站SPI-3、SPI-6以及SPI-12在測試集三種模型下，效率係數與判定係數在多數的情況下分別都能保持在0.7與0.8以上的水準。本研究也試圖透過前人的研究結果嘗試探討聖嬰現象與台灣地區降雨之關聯，對不同的地理位置與時間尺度下之降雨量做出推測，用以解釋模型表現優劣的原因。最後本研究認為Niño區域的海表面溫度對於用以預測台灣地區時間尺度3個月以上之SPI值，不管位於何地理位置，都能使模型有良好的表現。

The prediction of dry and wet periods is becoming more and more important to facilitate more effective water resources management and provide people with early warnings for evaluation. This study uses standardized precipitation index (SPI) which can quantify rainfall based on multiple timescales to measure eight rainfall stations of Keelung, Tamsui, Taichung, Kaohsiung, Hengchun, Chenggong, Hualien and Alishan in Taiwan. In this work, two popular variants of recurrent neural network (RNN) named long short-term memory (LSTM) and gate recurrent unit (GRU) networks were employed to predict SPI at four different timescales (SPI-1, SPI-3, SPI-6, SPI-12), and compared them with multiple linear regression (MLR). The three models were applied to simulate multiple timescales of SPI, using monthly rainfall data from 8 stations from 1951 to 2014 and the monthly sea surface temperature (SST) of the Niño regions. Under the three models in the testing data, the SPI-1 of each station could not be successfully established an acceptable model. However, the efficiency coefficient and the determination coefficient of the three models in the testing data of SPI-3, SPI-6 and SPI-12 at each test station can be maintained at 0.7 and 0.8 in most cases respectively. This research also attempts to explore the relationship between the El Niño phenomenon and rainfall in Taiwan based on previous research results, and to discuss the performance of the model under different geographical locations with multiple time scales. Finally, this study believes that the SSTs of Niño regions could be used to predict the SPI value, which of time scale is more than 3 months in Taiwan, regardless of the geographical location, and make the model perform very well.

Abramovitz, M., & Stegun, I. Handbook of Mathematical Formulas Graphs and Mathematical Tables. US Department of Commerce, National Bureau of Standards, Washington, DC. (1966).
Apaydin, H., Feizi, H., Sattari, M. T., Colak, M. S., Shamshirband, S., & Chau, K.-W. Comparative analysis of recurrent neural network architectures for reservoir inflow forecasting. Water, 12(5), 1500. (2020).
Bi, X. Y., Li, B., Lu, W. L., & Zhou, X. Z. Daily runoff forecasting based on data-augmented neural network model. [Article]. Journal of Hydroinformatics, 22(4), 900-915. (2020).
Bordi, I., & Sutera, A. Drought monitoring and forecasting at large scale Methods and tools for drought analysis and management (pp. 3-27): Springer. (2007)
Chen, G. T.-J., Jiang, Z., & Wu, M.-C. Spring heavy rain events in Taiwan during warm episodes and the associated large-scale conditions. Monthly Weather Review, 131(7), 1173-1188. (2003).
Chen, J. F., Jin, Q. J., & Chao, J. Design of Deep Belief Networks for Short-Term Prediction of Drought Index Using Data in the Huaihe River Basin. Mathematical Problems in Engineering, 2012. (2012).
Chen, S.-T., Kuo, C.-C., & Yu, P.-S. Historical trends and variability of meteorological droughts in Taiwan/Tendances historiques et variabilité des sécheresses météorologiques à Taiwan. Hydrological sciences journal, 54(3), 430-441. (2009).
Chen, S.-T., Yang, T.-C., Kuo, C.-M., Kuo, C.-H., & Yu, P.-S. Probabilistic drought forecasting in southern Taiwan using El Niño-Southern oscillation index. TAO: Terrestrial, Atmospheric and Oceanic Sciences, 24(5), 911. (2013).
Chiew, F. H., Piechota, T. C., Dracup, J. A., & McMahon, T. A. El Nino/Southern Oscillation and Australian rainfall, streamflow and drought: Links and potential for forecasting. Journal of hydrology, 204(1-4), 138-149. (1998).
Cho, K., Van Merriënboer, B., Gulcehre, C., Bahdanau, D., Bougares, F., Schwenk, H., et al. Learning phrase representations using RNN encoder-decoder for statistical machine translation. arXiv preprint arXiv:1406.1078. (2014).
Chung, J., Gulcehre, C., Cho, K., & Bengio, Y. Empirical evaluation of gated recurrent neural networks on sequence modeling. arXiv preprint arXiv:1412.3555. (2014).
Dai, A., Trenberth, K. E., & Qian, T. A global dataset of Palmer Drought Severity Index for 1870–2002: Relationship with soil moisture and effects of surface warming. Journal of Hydrometeorology, 5(6), 1117-1130. (2004).
de Oliveira-Júnior, J. F., de Gois, G., de Bodas Terassi, P. M., da Silva Junior, C. A., Blanco, C. J. C., Sobral, B. S., et al. Drought severity based on the SPI index and its relation to the ENSO and PDO climatic variability modes in the regions North and Northwest of the State of Rio de Janeiro - Brazil. Atmospheric Research, 212, 91-105. (2018).
Dikshit, A., Pradhan, B., & Alamri, A. M. Temporal Hydrological Drought Index Forecasting for New South Wales, Australia Using Machine Learning Approaches. Atmosphere, 11(6). (2020).
Djerbouai, S., & Souag-Gamane, D. Drought Forecasting Using Neural Networks, Wavelet Neural Networks, and Stochastic Models: Case of the Algerois Basin in North Algeria. [Article]. Water Resources Management, 30(7), 2445-2464. (2016).
Feng, P., Wang, B., Luo, J. J., Liu, L., Waters, C., Ji, F., et al. Using large-scale climate drivers to forecast meteorological drought condition in growing season across the Australian wheatbelt. Sci Total Environ, 724, 138162. (2020).
Fung, K. F., Huang, Y. F., Koo, C. H., & Soh, Y. W. Drought forecasting: A review of modelling approaches 2007-2017. [Review]. Journal of Water and Climate Change, 11(3), 771-799. (2020).
Ganguli, P., & Reddy, M. J. Ensemble prediction of regional droughts using climate inputs and the SVM-copula approach. Hydrological Processes, 28(19), 4989-5009. (2014).
Gao, S., Huang, Y., Zhang, S., Han, J., Wang, G., Zhang, M., et al. Short-term runoff prediction with GRU and LSTM networks without requiring time step optimization during sample generation. Journal of Hydrology, 589, 125188. (2020).
Gers, F. A., Schmidhuber, J., & Cummins, F. Learning to forget: Continual prediction with LSTM. (1999).
Gers, F. A., Schmidhuber, J., & Cummins, F. Learning to forget: Continual prediction with LSTM. Neural computation, 12(10), 2451-2471. (2000).
Hayes, M. J., Svoboda, M. D., Wilhite, D. A., & Vanyarkho, O. V. Monitoring the 1996 drought using the standardized precipitation index. [Article]. Bulletin of the American Meteorological Society, 80(3), 429-438. (1999).
Heddinghaus, T. R., & Sabol, P. A review of the Palmer Drought Severity Index and where do we go from here. Paper presented at the Proc. 7th Conf. on Applied Climatology. (1991).
Hochreiter, S., & Schmidhuber, J. Long short-term memory. Neural computation, 9(8), 1735-1780. (1997).
Kalra, A., Miller, W. P., Lamb, K. W., Ahmad, S., & Piechota, T. Using large‐scale climatic patterns for improving long lead time streamflow forecasts for Gunnison and San Juan River Basins. Hydrological Processes, 27(11), 1543-1559. (2013).
Kao, I. F., Zhou, Y. L., Chang, L. C., & Chang, F. J. Exploring a Long Short-Term Memory based Encoder-Decoder framework for multi-step-ahead flood forecasting. Journal of Hydrology, 583. (2020).
Khan, M. M. H., Muhammad, N. S., & El-Shafie, A. Wavelet based hybrid ANN-ARIMA models for meteorological drought forecasting. [Article]. Journal of Hydrology, 590, 9. (2020).
Kousari, M. R., Hosseini, M. E., Ahani, H., & Hakimelahi, H. Introducing an operational method to forecast long-term regional drought based on the application of artificial intelligence capabilities. Theoretical and applied climatology, 127(1-2), 361-380. (2017).
Kratzert, F., Klotz, D., Brenner, C., Schulz, K., & Herrnegger, M. Rainfall–runoff modelling using long short-term memory (LSTM) networks. Hydrology and Earth System Sciences, 22(11), 6005-6022. (2018).
Krause, P., Boyle, D., & Bäse, F. Comparison of different efficiency criteria for hydrological model assessment. Advances in geosciences, 5, 89-97. (2005).
Legates, D. R., & McCabe Jr, G. J. Evaluating the use of “goodness‐of‐fit” measures in hydrologic and hydroclimatic model validation. Water resources research, 35(1), 233-241. (1999).
Lin, C.-C., Liou, Y.-J., & Huang, S.-J. Impacts of two-type ENSO on rainfall over Taiwan. Advances in Meteorology, 2015. (2015).
Lloyd‐Hughes, B., & Saunders, M. A. A drought climatology for Europe. International Journal of Climatology: A Journal of the Royal Meteorological Society, 22(13), 1571-1592. (2002).
McKee, T. B., Doesken, N. J., & Kleist, J. The relationship of drought frequency and duration to time scales. Paper presented at the Proceedings of the 8th Conference on Applied Climatology. (1993).
Moriasi, D. N., Gitau, M. W., Pai, N., & Daggupati, P. Hydrologic and water quality models: Performance measures and evaluation criteria. Transactions of the ASABE, 58(6), 1763-1785. (2015).
Nash, J. E., & Sutcliffe, J. V. River flow forecasting through conceptual models part I—A discussion of principles. Journal of hydrology, 10(3), 282-290. (1970).
NOAA. El niño (ENSO) Related Rainfall Patterns Over The Tropical Pacific. Retrieved 4/21, 2021, from https://www.cpc.ncep.noaa.gov/products/analysis_monitoring/ensocycle/ensorain.shtml. (2021a)
NOAA. La Niña Related Rainfall Patterns Over The Tropical Pacific. Retrieved 4/21, 2021, from https://www.cpc.ncep.noaa.gov/products/analysis_monitoring/ensocycle/laninarain.shtml. (2021b)
NOAA. Niño Regions. Retrieved 4/22, 2021, from https://www.ncdc.noaa.gov/teleconnections/enso/indicators/sst/. (2021c)
Palmer, W. C. Meteorological drought. (1965).
Palmer, W. C. Keeping track of crop moisture conditions, nationwide: the new crop moisture index. (1968).
Pascanu, R., Mikolov, T., & Bengio, Y. On the difficulty of training recurrent neural networks. Paper presented at the International conference on machine learning. (2013).
Paulo, A. A., & Pereira, L. S. Drought concepts and characterization: Comparing drought indices applied at local and regional scales. [Article]. Water International, 31(1), 37-49. (2006).
Poornima, S., & Pushpalatha, M. Drought prediction based on SPI and SPEI with varying timescales using LSTM recurrent neural network. Soft Computing, 23(18), 8399-8412. (2019).
Rumelhart, D. E., Hinton, G. E., & Williams, R. J. Learning representations by back-propagating errors. nature, 323(6088), 533-536. (1986).
Saadat, H., Adamowski, J., Bonnell, R., Sharifi, F., Namdar, M., & Ale-Ebrahim, S. Land use and land cover classification over a large area in Iran based on single date analysis of satellite imagery. Isprs Journal of Photogrammetry and Remote Sensing, 66(5), 608-619. (2011).
Shafer, B., Dezman, L., Simpson, H., & Danielson, J. Development of surface water supply index-A drought severity indicator for Colorado. Paper presented at the Proceedings of Western Snow Conference. (1982).
Sheela, K. G., & Deepa, S. N. Review on methods to fix number of hidden neurons in neural networks. Mathematical Problems in Engineering, 2013. (2013).
Stathakis, D. How many hidden layers and nodes? International Journal of Remote Sensing, 30(8), 2133-2147. (2009).
Szalai, S., & Szinell, C. Comparison of two drought indices for drought monitoring in Hungary—a case study Drought and drought mitigation in Europe (pp. 161-166): Springer. (2000)
Trenberth, K. E. Signal versus noise in the Southern Oscillation. Monthly Weather Review, 112(2), 326-332. (1984).
Trenberth, K. E., & Hoar, T. J. The 1990–1995 El Niño‐Southern Oscillation event: Longest on record. Geophysical research letters, 23(1), 57-60. (1996).
Troup, A. The ‘southern oscillation’. Quarterly Journal of the Royal Meteorological Society, 91(390), 490-506. (1965).
Vujicic, T., Matijevic, T., Ljucovic, J., Balota, A., & Sevarac, Z. Comparative analysis of methods for determining number of hidden neurons in artificial neural network. Paper presented at the Central european conference on information and intelligent systems. (2016).
Wilhite, D. A., & Glantz, M. H. Understanding: the drought phenomenon: the role of definitions. Water international, 10(3), 111-120. (1985).
Wilhite, D. A., Hayes, M. J., Knutson, C., & Smith, K. H. Planning for drought: Moving from crisis to risk management 1. JAWRA Journal of the American Water Resources Association, 36(4), 697-710. (2000).
Woli, P., Jones, J. W., Ingram, K. T., & Fraisse, C. W. Agricultural reference index for drought (ARID). Agronomy Journal, 104(2), 287-300. (2012).
Wu, P.-Y., You, G. J.-Y., & Chan, M.-H. Drought analysis framework based on copula and Poisson process with nonstationarity. Journal of Hydrology, 588, 125022. (2020).
Zhang, D., Peng, Q. D., Lin, J. Q., Wang, D. S., Liu, X. F., & Zhuang, J. B. Simulating Reservoir Operation Using a Recurrent Neural Network Algorithm. [Article]. Water, 11(4), 18. (2019).
Zhang, Y. G., Tang, J., He, Z. Y., Tan, J. K., & Li, C. A novel displacement prediction method using gated recurrent unit model with time series analysis in the Erdaohe landslide. [Article]. Natural Hazards, 105(1), 783-813. (2021).
Zhao, X. H., Lv, H. F., Wei, Y. Z., Lv, S. J., & Zhu, X. P. Streamflow Forecasting via Two Types of Predictive Structure-Based Gated Recurrent Unit Models. [Article]. Water, 13(1), 17. (2021).
吳定澄，許友貞和李忠潘，聖嬰－南方振盪現象對台灣氣溫及雨量之影響，海洋工程學刊，15(1)，57-65，(2015)。
李昱祺，王嘉琪，翁叔平，陳正達和鄭兆尊，臺灣氣象乾旱特性未來趨勢推估，大氣科學，47(1)，66-93，(2019)。
陳弘，遙相關月雨量預報模式應用於石門水庫乾旱預警，國立成功大學水利及海洋工程學系碩士論文，台南市，取自https://hdl.handle.net/11296/ddutj8，(2019)。
陳昭銘和汪鳳如，台灣地區長期暖化現象與太平洋海溫變化之關係，大氣科學，28(3)，221-241，(2000)。
陳儒賢，洪毓婷和陳清田，台灣地區乾旱特性之研究，中華水土保持學報，43(1)，85-95，(2012)。
游保杉，發展資料挖掘為基礎的乾旱預警研究：科技部補助專題研究計畫報告，(2019)。
楊道昌，郭俊超，呂季蓉和游保杉，標準化降雨指標應用於農業乾旱監測之研究，農業工程學報, 51(2)，11-25，(2005)。
蔡雨璇，結合乾旱指標之多水庫系統環境流量管理策略，國立成功大學水利及海洋工程學系碩士論文，台南市，取自https://hdl.handle.net/11296/xtrqru，(2019)。
蕭政宗，聖嬰現象與臺灣地區降雨量之相關分析，農業工程學報，46卷(1期)，頁93-110，(2000)。
蕭政宗和楊志傑，台灣地區之區域乾旱頻率分析，農業工程學報，52(2)，83-101，(2006)。