應用 RiverFlow2D 與SRH-2D 探討野溪泥沙運移-以那托爾薩溪為例

Abstract

SRH-2D二維水理輸砂模式目前已廣泛應用於水利工程，其模擬器主要採CPU中央處理器為主，其善於處理單一邏輯繁複的任務，然其並不善於處理大量瑣碎的任務。RiverFlow2D模式結合CPU中央處理器、GPU圖形化加速器的優點，過往研究顯示其有效縮減模擬時間。本文採兩輸砂模式進行分析，項目包含時間步距與網格精度的敏感度分析、實際降雨事件、構造物模擬等。 本研究以那托爾薩溪為案例，其過往曾發生土石流，目前仍屬於土石流高潛勢溪流，上游河道內仍有大量土石淤積尚未穩定，底床載運移頻繁值得進步研究。 透由現地量測與Python爬蟲收集降雨量資料、水文分析、空拍地形建置、河床質粒徑採樣等調查項目，獲取兩模式之參數，成果說明如下： 1.與過往研究比較，本研究透過現場監測與空拍機詳細調查模式所需參數，包含河床質參數、河道地形、集水區水情(降雨量、水位)、河道構造物等。 2.經由平均絕對誤差(MAE)與均方根誤差(RSML)評估兩模式的敏感度，SRH-2D和RiverFlow2D兩模式對於網格精度均屬敏感，此外針對時間步距SRH-2D的敏感度分析結果亦顯示良好，隨著時間步距縮減，模擬時間增加。 3.本研究使用底床載運移公式進行模擬，RiverFlow2D支援多達十個公式，而SRH-2D目前僅支援三個公式，然而SRH-2D多考慮二次流、泥沙尺寸大小等級、底床護甲層作用，較能準確預估本研究區彎道土砂運移。 4.本研究使用RiverFlow2D和SRH-2D兩個模式進行二維輸砂模擬，其模擬結果顯示兩模式相互比較沖淤結果相近、河道整體多為沖刷，模擬結果與驗證資料之淨沖淤體積趨勢相同。

The two-dimensional hydraulic sediment transport model SRH-2D has been widely used in water conservancy projects. Its simulator is mainly based on CPU (Central Processing Unit). It is good at dealing with single logic complicated tasks, but it is not good at handling a lot of trivial tasks. RiverFlow2D mode combines the advantages of CPU central processor and GPU (Graphics Processing Unit) to effectively reduce the simulation time from tens of hours to several minutes.In this paper, two sediment transport moduls are analyzed, and the project includes sensitivity analysis of time space and grid accuracy, rainfall - runoff events, structure simulation, etc. The research project includes model parameter collection and selected. This study takes the case of Atul Creek. In the past, earth-rock flow occurred. It is still a high-potential stream of earth-rock flow. There is still a large amount of earth-rock deposition in the upper channel, which is not stable yet. Through the survey data of rainfall data collection, hydrological analysis, aerial terrain construction, riverbed quality sampling, etc. The results collected are as follows: 1.Comparison with previous studies, this study used on-site monitoring and UAV detailed survey model of the parameters required, including the nature of the iverbed, river terrain, Rainfall and water levels at the scene, river structures and so on. 2.The Sensitivity of the Two models is evaluated by average absolute error (MAE) and mean square root error (RSML), the SRH-2D and RiverFlow2D modes are sensitive to mesh accuracy, and the sensitivity analysis results for the time step SRH-2D are also good, with reduced step length over time. Simulation time has increased significantly. 3.This study used a formula to simulate sediment transport, RiverFlow2D although support up to ten formulas, and SRH-2D is currently only supports three formulas, but more consideration to the SRH-2D secondary flow, sediment size grades, the bed Armor layer effect, the more accurate estimate of this study bend earth and sand transport. 4.The simulation of two-dimensional sediment transport using RiverFlow2D and SRH-2D in this study, and the simulation results show that the two model silting results are similar, like the tendency of the data of aerial verification.