石門水庫集水區土壤沖蝕與水文地文影響因子研究

A Study on Soil Erosion and Influential Hydrologic and Geographic Factors for Shihmen Reservoir Watershed

林韋成(Wei-Cheng Lin)

指導教授 ： 張德文

淡江大學/工學院/土木工程學系碩士班/碩士(2014年)

https://doi.org/10.6846/TKU.2014.00842

摘要

本研究以石門水庫集水區水文與地文資料以及土壤沖蝕監測資料為主，考量不同雨場代表特性，分析土壤沖蝕量之相關因子，並依據觀察結果建立土壤沖蝕經驗式。該經驗式可模擬不同降雨境況，並用於預測該水庫集水區可能之土壤沖蝕量。研究首先蒐集2008~2011年間石門水庫各區雨量資料，並將各子集水區內之降雨測站資料以徐昇氏法合成集水區內降雨情形，再透過合宜的雨場判定方法，萃取出各集水區內降雨特性。其次再以集水區內現地沖蝕針資料及地理資訊系統資料，透過Minitab統計軟體進行敏感度分析及多項式迴歸分析，得到不同坡面狀況之土壤沖蝕經驗公式。最後再利用本研究所迴歸之土壤沖蝕經驗式、通用土壤流失公式(USLE)、水保局經驗式及水蝕預估模式程式(WEPP)，配合代表性之子集水區沖蝕數據進行比較驗證。

研究結果顯示：1.植生程度將影響邊坡的年平均沖蝕量。依其植生程度而不同，裸露邊坡沖蝕量約自然植生邊坡之1.3倍，約人工整治邊坡之1.6倍，坡面的附蓋率及人工整治能有效抑制土壤沖蝕行為。2.本研究將可能影響土壤沖蝕量之相關因子進行檢定，以篩選相關因子，研究發現水文因子以雨場次數、降雨動能、降雨延時及累積雨量較能反映沖蝕量，地文因子則以坡度為主要影響因子。對土壤沖蝕深度敏感度以累積降雨量為最高，其次降雨延時，再者為雨場發生次數，最後則為邊坡坡度。3.降雨季節的迴歸分析可區分為颱風季節、春雨季節及梅雨季節。春雨季節模擬最好，R2值達0.837；颱風季節次之，R2值達0.605，梅雨季節迴歸模擬因降雨強度不高，降雨延時長，分析關聯性不高。該項觀察和雨場判定方式有關，本研究係以詹錢登(2002)所建議者為主。4.若以WEPP模擬沖蝕量，在相關參數的決定上，尤為困難；其考慮的參數多，結果將相對準確。由於本研究除氣候資料是由中央氣象局取得外，土壤參數皆為集水區平均值，若能進一步掌握當地鑽探資料，應能更準確預測土壤沖蝕深度。5.在案例比較中發現：(1) 植生邊坡迴歸式估算坡度超越20度之邊坡時，有低估沖蝕之情況，建議在經驗式上依集水區加入不同標準偏差進行土壤沖蝕深度之修正，以使結果能滿足所需。(2) 依據水保局建議所採用的USLE經驗式模擬結果在案例分析中顯示其過度保守，預測結果約為實測值之2至5倍間。(3) 氣候變遷導致年平均雨量有上升的趨勢，採WEPP模擬近年平均土壤沖蝕深度發現，近兩年的土壤沖蝕深度約為近十年的1.4~2倍之間，約為近五年的1.2~1.5倍之間。

關鍵字

土壤沖蝕 ； 雨場切割 ； WEPP ； Minitab ； 多項式回歸

並列摘要

This study mainly takes the hydrology and physiographic factors, as well as the soil erosion pins data of Shihmen Reservoir watershed to analyze the soil erosion related factors by considering the characteristics of different raining sites, and to establish the soil erosion empirical formula based on the observation results. This empirical formula can not only simulate various rainfall conditions, but also predict the possible amount of soil erosion in Shihmen Reservoir watershed. The study firstly collects the rainfall data in Shihmen Reservoir during the period of 2008-2011, and composes the rainfall condition of the watershed based on Thiessen method and by using the data collected at the rainfall prediction station of each sub watershed. Moreover, it extracts the characteristics of rainfall in each sub watershed by appropriate cutting method. After that, it conducts sensitivity analysis and polynomial regression analysis with Minitab statistical software and based on current erosion indicators and GIS data within the watershed, so as to obtain the empirical soil erosion regression formula for different slope conditions. Finally, with the soil erosion data in representative sub watershed, it conducts comparison and validation of the empirical soil erosion regression formulas with other prediction methods such as USLE and WEPP.

The study shows that：1. Planting will affect the amount of soil erosion on the slope. Dependent on planting conditions, the amount of soil erosion on bare slope is about 1.3 times of that of vegetation slope, and 1.6 times of that of artificial remediation slope. It shows that coverage and remediation can effectively inhibit soil erosion. 2. The study tested the related factors that might possibly affect the soil erosion, so as to screen out the related factors. The soil erosion amount could be reflected by such hydrologic factors as the rainfall frequency, rainfall energy, duration and cumulative rainfall; in terms of the geographic factors, the slope gradient shows the major influence. As for the sensitivity of the soil erosion depth, the cumulative rainfall takes the first place, followed by duration and rainfall frequency, and slope gradient takes the last place. 3. Regression analysis of rainfall seasons can be distinguished as the typhoon season, spring season and plum rains season. The regress simulation in spring season obtains the best result, with the R2 value reaching 0.837. It is followed by typhoon season, with the R2 value reaching 0.605. Due to low rainfall intensity and long duration, the correlation of analysis is relatively low for the plum rains season. The observation is correlated to the cutting method. The study mainly employs the method recommended by Zhai Qian-Deng (2002). 4. If WEPP is used to simulate the soil erosion amount, it will be quite difficult to determine the related parameters. By taking a lot of parameters into account, it will obtain accurate results. Except for the weather data obtained from the Central Weather Bureau, the study takes the mean values in the watershed as the soil parameters. Thus, it will be able to predict the soil erosion depth more accurately if local drilling data were available. 5. The case comparison shows that: (1) When the regression formula of planting slope estimates the slope gradient is higher than 20 degrees, the simulation might underestimate the erosion. It is suggested that the soil erosion depth could be corrected by adding different standard deviations into the empirical formula, so as to make the results satisfy the demands. (2) The simulation results of the USLE empirical formula as suggested by Water Conservation Bureau show that it is overly conservative, the prediction results are 2~5 times of the measured ones. (3) The climate change leads to the increase of annual average rainfall. By simulating the average soil erosion depth in recent years with WEPP, it is found that the soil erosion depth in the last two years is about 1.4~2 times of that in the last decade, and about 1.2~1.5 times of that in the last five years.

並列關鍵字

soil erosion ； cutting method ； WEPP ； Minitab ； polynomial regression

參考文獻

37. 陳怡睿、張雲岳(2013)，「現地沖蝕釘試驗應用於集水區降雨影響土壤沖蝕之探討」。長榮大學土地管理與開發學系碩士論文。

35. 林俐玲(2010)，「通用土壤流失公式(USLE)之應用與誤用」。水保技術 5:2 2010.04[民99.04] 頁119-121。

44. 梁惠儀、許振崑、林伯勳、鄭錦桐、冀樹勇(2010)，極端暴雨事件於石門水庫集水區之土壤沖蝕量估算及探討，中興工程季刊，第106期，第5-15頁

47. 陳樹群、簡如宏、馮智偉、巫仲明(1998)，「本土化土壤沖蝕指標模式之建立」，中華水土保持學報，第29 卷，第3 期。

48. 黃俊德(1979)，「台灣降雨沖蝕指數之研究」，中華水土保持學報，20(2):17-4