區域降雨模擬時空架構發展應用

Development of Spatio-temporal Framework for Regional Precipitation Simulation

傅冠禎(Kuan-Cheng Fu)

指導教授 ： 游景雲

國立臺灣大學/工學院/土木工程學研究所/碩士(2017年)

https://doi.org/10.6342/NTU201703205

摘要

隨著都市化的發展以及氣候變遷的影響，乾旱及豪大雨等極端氣候發生的強度與頻率逐漸增加，並預期會對於環境造成更大之衝擊，因此也更加彰顯都市排水及防洪議題之重要性。而防洪措施之施作以及洪水管理方法之建立，皆仰賴於降雨序列作為輸入之參數，但由於測站歷史資料取得不易，且常有資料長度不足之問題，因此推估降雨量於時間與空間上之分佈逐漸成為一重要課題。
水文學家過往基本上專注於發展單一測站之降雨序列模擬，較少著重於降雨之空間分布特性，因此本研究將基於非參數方法之精神，提供多測站之降雨序列模擬之方法。本研究方法主要分為三部分，首先藉由馬可夫序列判釋空間中各測站是否發生降雨；第二部分則假定某站於時間 下之降雨量與時間 下之各站降雨量存在線性關係，並利用該線性關係建立多變數自回歸模式，若測站發生降雨則進一步利用此自回歸模式估計降雨量；最後則根據改進之最近鄰居法處理自回歸模式中之殘差項，藉以提升雨量序列之多樣性。最後此研究方法將基於歷史資料之特性，針對淡水河流域中三十五個測站進行降雨序列之模擬。

關鍵字

日雨量模擬 ； 時空架構 ； 馬可夫鏈 ； 多變數自回歸模式 ； 最近鄰居法

並列摘要

With the change of future climatic conditions, the environment will face more severe hydrological events. Under these situations, urban flooding is one of serious emerging problems. For this concern, simulating spatial-temporal distribution of precipitation has become an important issue.
During the past decades, hydrologists have improved the development of generating synthetic sequences of precipitation in a given rain gauge, but seldom focused on the spatial characteristics of precipitation in a given space region. Therefore, the aim of this study is to extend the concept of nonparametric approaches to develop a framework which is capable of simultaneously generating multisite precipitation amounts in a given space region over time.
The purposed framework is composed of three parts: multisite precipitation occurrence process, multisite precipitation amounts process, and modified K-NN method. First, a first-order Markov chain is used to simultaneously simulate precipitation occurrences across the network of multiple gauges. Subsequently, the correlations between precipitation amount of the given rain gauge and precipitation amounts of all rain gauges across the network are obtained by a multivariate autoregressive model. Moreover, the concepts of spatial uncertainty, based on the residual sampling in the modified K-NN method, are added to the model but moderately adjusted to deal with the change of the correlation functions.
After that, a case study is demonstrated in Danshui River Basin with the historical daily precipitation data from thirty-five rain gauges for seven-year period. Besides, the simulation results of the purposed framework are expected to preserve the statistical properties of the observed data and also possess a rich variety in time sequences.

並列關鍵字

daily precipitation simulation ； spatial-temporal framework ； Markov chain ； multivariate autoregressive model ； K-nearest neighbor method

參考文獻

Aksoy, H., & Bayazit, M. (2000). A model for daily flows of intermittent streams. Hydrological Processes, 14(10), 1725-1744.

Charles, S. P., Bates, B. C., & Hughes, J. P. (1999). A spatiotemporal model for downscaling precipitation occurrence and amounts. Journal of Geophysical Research: Atmospheres, 104(D24), 31657-31669.

Cliff, A. D., & Ord, J. K. (1975). Space-time modelling with an application to regional forecasting. Transactions of the Institute of British Geographers, 119-128.

Cowpertwait, P. S. P., Kilsby, C. G., & O'Connell, P. E. (2002). A space‐time Neyman‐Scott model of rainfall: Empirical analysis of extremes. Water Resources Research, 38(8).

Dalezios, N. R., & ADAMOWSKI, K. (1995). Spatio-temporal precipitation modelling in rural watersheds. Hydrological sciences journal, 40(5), 553-568.