2006-08-012024-05-18https://scholars.lib.ntu.edu.tw/handle/123456789/697176摘要：本研究的目的是提出一內嵌自動參數檢定法的非線性降雨—逕流模式。其中的自動參數檢定法是以遺傳演算為基礎而發展。並且，本研究利用現代資訊科技提出以該非線性降雨—逕流模式為核心的水資訊系統。相較於傳統及以類神經網路為基礎的的降雨—逕流模式，本研究所提出的非線性降雨—逕流模式具有幾個特點。由於該模式具參數自動檢定法，避免傳統降雨—逕流模式參數難以獲得的困難。模式架構可配合流域特性而調整以及使用者的需要，亦為傳統降雨—逕流模式所沒有的一大特點。對以類神經網路為基礎的的降雨—逕流模式來說，為增進模擬的精確度，需進行系統識別且需要大量的輸入資料。而本研究提出的模式架構已隱含降雨與逕流間的時間相依性，只需降雨量作為其輸入項，因此可大幅減少輸入項的維度並免去系統識別的工作。且本研究所提出的模式不需要逕流量作為模式的輸入，使得本模式的使用範圍更為廣泛。 不僅如此，本研究更將所提出的降雨—逕流模式以物件導向設計的方法加以實作，提出以該降雨—逕流模式為核心的水資訊系統。其包含可調整模式架構、因應不同模式架構的自動化參數檢定法、圖形化使用者介面以及報表及圖形輸出等功能。該水資訊系統可使各不同單位的水利工程師毋需花費大量時間級金錢重複發展本研究提出的降雨—逕流模式的電腦程式，只需受簡單的教育訓練後即可輕易使用。 綜合來說，本研究整合了降雨—逕流過程模擬的理論、最佳化理論以及現代資訊工程的技術開發水資訊系統，對於臺灣發展基礎水文科學及掌握水資訊領域中關鍵性技術發展具有極大的意義與助益。 <br> Abstract: The purpose of this project is to propose a nonlinear rainfall-runoff model embedded with an automated calibration method. The automated calibration method of the proposed model is based on genetic algorithms. Furthermore, a hydroinformatics system containing the proposed model is developed using the modern information technologies. The proposed model has several advantages over the conventional rainfall-runoff models and artificial neural network (ANN) based rainfall-runoff models. Unlike the conventional rainfall-runoff models, the proposed model possesses an automated calibration method so that the appropriate model parameters can be easily obtained. The configuration of the proposed model can be adjusted to the characteristics of the watershed and the requirements of the modelers. This is another advantage of the proposed model over the conventional ones. The only requirement of the input of the proposed model is the sequential rainfall depth records. The proposed model prevents the requirement of the several preceding rainfall and runoff records as input, while the ANN based rainfall-runoff models require them as input. Therefore, the proposed model doesn’t need the work of system identification that is necessary for ANN based rainfall-runoff models to improve the modeling accuracy. This property also reduces the dimensions of the input data and makes the applicability of proposed model be more comprehensive than the ANN based rainfall-runoff models. Considering the easy usability of the proposed model, a hydroinformatics system packaging the proposed model is then developed. The popular paradigm of modern information system design, object-oriented design (OOD), is adopted here as the guideline of the development of the hydroinformatics system. The functionalities of the hydroinformatics system involve the graphical user interface (GUI), text file processing module, system kernel, a database and the corresponding accessing module. The GUI helps the modelers and engineers utilize the hydroinformatics system easily. The system kernel that processes the simulation and the calibration of the proposed model is the key part of the hydroinformatics system. With the hydroinformatics system, modelers and engineers can utilize the proposed model easily. In summary, this project integrates the modeling theory of the rainfall-runoff process, the theory of optimization and the modern information technologies into the hydroinformatics system. The achievements of this project improve the development of the fundamental hydrological science in Taiwan and help Taiwan master the unique technology of hydroinformatics. The hydroinformatics system developed in this project also facilitates the modelers and engineers to model the rainfall-runoff pro降雨—降雨—逕流模式自動化檢定法遺傳演算法物件導向水資訊學rainfall-runoff modelautomated calibration methodgenetic algorithmsobject-orientedhydroinformatics