节点文献

武陵山区低山丘陵小流域土壤侵蚀特征及产流产沙模拟预测

Characters of Soil Erosion and Runoff and Sediment Yield Simulation of Watershed in Low-mountain Hilly Region of Wuling Mountian

【作者】 周璟

【导师】 彭镇华;

【作者基本信息】 中国林业科学研究院 , 森林培育, 2009, 博士

【摘要】 长江中下游低山丘陵生态退化区为防治好水土流失已开展了大规模的植被恢复工作,要了解植被恢复所产生的水土保持效应乃至生态环境效应,研究流域土壤侵蚀特征及规律,是必不可缺的。本文以了解植被恢复对长江中下游低山丘陵区所产生的水土保持效应为主要目的,选取湖南西北部武陵山区女儿寨小流域为研究流域,对降雨侵蚀力、土壤可蚀性K值等因素进行详细分析,在总结流域不同尺度产流产沙特征与植被恢复对水土流失改善效果的基础上,选用GIS结合USLE及分布式水文模型SWAT两种方式对流域产流产沙进行模拟预测,对模型在流域的适宜性进行评价,通过比较找出更适合于流域的预测方式,为流域今后水土流失治理提供基础数据,为今后指导武陵山区及长江中下游低山丘陵区相似流域的水土保持工作提供借鉴与参考。主要得出如下结论:(1)流域降雨侵蚀力时间分布不均匀,主要集中在4-8月。2000-2008年9年间降雨侵蚀力波动变幅较大。次降雨量与次降雨侵蚀力、月降雨量与月降雨侵蚀力、年降雨量与年降雨侵蚀力均为乘幂关系,其拟合的函数可用于相应降雨侵蚀力的计算,其中以月侵蚀性降雨量作为变量计算精度最大,决定系数达到0.8605。在流域内长期计算与分析降雨侵蚀力,EI结构中以EI30最为适宜。半月步长日雨量模型R=α(Dj)β和非线性多参数模型Rj=a{1+bsin[Π(j-1)/12]} Pkc适合流域降雨侵蚀力的评价。(2)小流域土壤可蚀性K值介于0.2451—0.4623之间,其变幅较大。土壤表层质地普遍偏粗,粘土所占比例较小。K值离差系数Cv为0.1295,存在中等程度的变异性。流域几种主要土壤类型K值为:黄壤0.332,黄棕壤0.326,红壤0.318。通过Kriging插值得到K值在流域范围内的空间分布图。今后,增加采样密度和优化采样方法将是提高K值空间分布图精度的有效途径。(3)流域内几种具有代表性的植被恢复模式径流小区,其产流产沙量在2000-2008年9年间存在巨大的差异,自2004年后产流产沙均显著减少并基本稳定,植被恢复对水土流失的效益从2004年开始得到明显体现。灌木林和润楠林小区具有良好的保水减沙效益,退耕还林措施同样带来了良好的效益。从产流产沙的年内分布来看,4-8月是水土流失的主要时期。随着植被恢复年限增大,林分郁闭度增加,降雨因子对产流产沙的显著作用有所增加,体现在2004年后降雨因子与产流产沙的相关系数明显增大,产流产沙回归方程的回归相关系数从2004年后除润楠林小区外普遍有了增加。各个小区产沙量均随产流量的增大而增大,以润楠林小区产沙随产流的变化最大。(4)对于植被恢复径流小区,乔层盖度是对产流产沙影响最大的环境因子,其次是毛管持水量,地形因子的坡度对产沙有较大影响;对于退耕还林小区,坡度是对产流产沙影响最大的环境因子,其次是土壤渗透速度,灌层盖度和草层盖度对产流产沙有较大影响且影响作用相同。(5)从流域尺度上来看,汛期是流域径流产生的主要时期,洪水是流域泥沙产生的主要来源。径流量2004年前和2004年后没有太大变化,而产沙量自2004年后显著减少,2004-2008年来泥沙特征值的变化幅度大于降雨与径流特征值变化幅度。流域径流量随着降雨量的增加而增加,汛期月径流深与月输沙量呈正相关关系。水文指标对流域土壤侵蚀量的相关性要高于降水指标。选择3种降水指标、4种水文指标及径流系数Cr,建立流域土壤侵蚀量S的定量计算方程式,经检验方程复相关系数平方R2达到0.96,可用于流域土壤侵蚀量的定量计算。(6)利用GIS并以USLE为评价模型预测了流域土壤侵蚀量。<5t/(hm2·a)的微度侵蚀占侵蚀总面积的66.02%。林地除竹林外均为微度侵蚀,也说明了林地防治土壤侵蚀的效益。海拔300-400m与200-300m地带的侵蚀量分别占到侵蚀总量的34.99%和29.89%,是需要预防土壤侵蚀的重点地带。坡度20-40o地带侵蚀量占总侵蚀量的55.5%,也是防治土壤侵蚀的主要区域。整个流域年均侵蚀模数为652.37 t/(km2·a),与实测值相比未达到较为满意的结果,只能用于粗略的估算。(7)应用分布式水文模型SWAT对流域产流产沙进行预测。经过参数校准和模型验证,模型对径流深的模拟值相关系数(R2)与Nash-Suttclife系数(Ens)两个评价指标均在0.8以上,相对误差系数(RE)也仅为0.004;输沙模数整体值小于实测值,但R2为0.88,说明模拟值可以很好的反映泥沙变化的趋势,Ens为0.78也是模型可以接受的精度范围。结果表明,模型可以较好的模拟流域产流产沙,在流域内的适用性好于USLE模型。SWAT模型对产流的模拟效果要好于对产沙的模拟效果。

【Abstract】 It has been implemented large-scale vegetation restoration in order to prevent soil and water loss in low-mountain hilly ecological degraded region of middle-lower Yangtze River. In order to realize the effect of soil and water conservation even ecological environment which produced by vegetation restoration,study on soil erosion characters of watershed is necessary. The main intention of this paper were to realize the effect of soil and water conservation which produced by vegetation restoration in low-mountain hilly region of middle-lower Yangtze River. This paper selected Nverzhai small watershed which lies in Wuling Moutain in northwest of Hunan province as study area,analyzed some factors such as rainfall erosivity、soil erodibility(K factor) in detail. Then generalized soil erosion characters on different scale and improvment of vegetation restoration to prevent soil and water loss. Based on above,this paper simulated runoff and sediment yield combining with GIS and USLE and distributed hydrological model SWAT. Then,This paper evaluated which model can be used better in watershed. This can provide basal data for soil and water conservation of watershed in future,and also can provide experience learned and reference for soil and water conservation in future in Wuling Moutain as well as low-mountain hilly region of middle-lower Yangtze River. Main results and conclusions showed as follows:Time distribution of rainfall erosivity in watershed were not symmetrical. It was mainly concentrated from April to August. Rainfall erosivity had comparatively high fluctuation in 9 years from 2000 to 2008. Single rainfall and single rainfall erosivity,monthly rainfall and monthly rainfall erosivity,yearly rainfall and yearly rainfall erosivity were in exponential function relation. Their regressive functions can calculate corresponding rainfall erosivity. Determinative coefficient of regressive functions were 0.8605,this can gain by ways of taking monthly rainfall as variable. To calculate and analyse rainfall erosivity for long-timg,EI30 were best applicable in EI configuration. Daily rainfall precipitation model of half month periods of time R=α∑(Dj)βand non-linear and more parameters model Rj=a{1+bsin[Π(j-1)/12]} Pkc also have good applicability for evaluating rainfall erosivity.The value of soil erodibility(K factor) were between 0.2451 to 0.4623,it’s variation were relative large. Soil texture of surface layer were generally coarse and the proportion of clay were little. K value which Cv was 0.1295 were in moderate variability. K value of main soil types of watershed can be calculated. Yellow earth,yellow brown earth and red earth were 0.332,0.326 and 0.318 respectively. Spatial distribution map of K value in watershed can be made by Kriging interpolation. Precision of Spatial distribution map of K value can be improved by increasing sampling density and optimizing sampling method in future.Runoff and sediment yield had largely difference on typical vegetation patterns plots in 9 years from 2000 to 2008. It greatly decreased since 2004 year and stability hereafter. Benefit from vegetation restoration to soil and water loss were evidently received since 2004. Shrub and Pingii community had preferable benefit to prevent soil and water loss. The measure of returning land for farming to forest also had better benefit. Main period of soil and water loss was April to August for distribution of runoff and sediment within year. Following with the time increment of vegetation restoration and increment of forest canopy density,the effect which rainfall factor on runoff and sediment yield were obviously increased. The representative was that correlative coefficient between rainfall factors and runoff and sediment increased obviously,and also regressive correlative coefficient of regressive functions which between rainfall factors and runoff and sediment were generally increasing except Pingii community since 2004. Sediment yield were increasing follow with the increment of runoff yield on all plots,change of Pingii community were maximal.On runoff plots of vegetation restoration,the most impacting factor to runoff and sediment yield was arbor cover. The next was capillary water holding. Slope had relatively large impacting to sediment yield. On returning land for farming to forest plots,the most impacting factor to runoff and sediment yield was slope. The next was soil infiltration speed. Shrub cover and grass cover also had great impact and their effect were the same.On watershed scale,flood season was the main time of runoff yield,flood was the main source of sediment yield. Runoff had no high fluctuation,but sediment obviously decreased since 2004. Changing extent of sediment eigenvalue were higher than rainfall and runoff eigenvalue from 2004 to 2008. Runoff increased following with the increment of rainfall,monthly runoff depth and monthly sediment yield took on positive correlation in flood season. The relativity of hydrology indexes to soil erosion were higher than rainfall indexes.The calculative equation about sediment yield of watershed can be established after 3 rainfall indexes,4 hydrology indexes and runoff coefficient Cr were selected. R2 of the equation was 0.96 by testing,this showed the equation can be used to calculate sediment yield of watershed.Soil erosion of watershed were simulated based on combining with GIS and USLE. Feeble erosion which less than 5t/(hm2·a)occupied 66.02 percent in total erosion. Forest land were feeble erosion except bamboo,this showed the benefit which forest land to prevent soil erosion. Soil erosion in elevation of 300-400m and 200-300m occupied 34.99 and 29.89 percent in total erosion respectively. This was the main area which needed to prevent soil erosion. Soil erosion between slope of 20-40o occupied 55.5 percent in total,this was also the main area which needed to prevent soil erosion. The average of yearly sediment yield was 652.37 t/(km2·a)in watershed,this was not satisfying result comparing with actual observed result and only can be used to gross estimation.Runoff and sediment yield were simulated combining with distributed hydrological model SWAT. After parameter adjustment and model test , correlative coefficient ( R2 ) and Nash-Suttclife coefficient(Ens)about simulative value of runoff depth were higher than 0.8. Relative error coefficient was only 0.004. Whole sediment yield were less than actual observed result,but it’s correlative coefficient(R2) was 0.88. This showed simulative value can reflect the trend of sediment change preferable. The value of Ens was 0.78 also in precision range that model can accept. Result showed that SWAT model can simulate runoff and sediment yield preferable and had better applicability to USLE in watershed. Simulative result to runoff were better than sediment for SWATmodel.

节点文献中: 

本文链接的文献网络图示:

本文的引文网络