土壤侵蚀已经成为全球最重要的环境问题之一，随着社会的不断发展，土壤侵蚀问题将变得日趋严峻。中国是土壤侵蚀严重的国家之一，进行土壤侵蚀预报治理已经迫在眉睫。土壤侵蚀按照作用力可分为水力侵蚀、风力侵蚀、冻融侵蚀和重力侵蚀，其中水力侵蚀在中国是最为活跃的侵蚀类型，降雨是导致水力侵蚀的主要动力。降雨侵蚀力（Rainfall Erosivity）反映了降雨引起土壤侵蚀的这种潜力。因此客观准确的预报降雨侵蚀力，分析其季节变化对于定量预报中国土壤流失、进行水土保持具有重要意义。本文对降雨侵蚀力的研究主要包括3部分：一是确立短历时降雨（小于30min）最大30min雨强的计算方法；二是确立次、日雨量计算降雨侵蚀力的简易计算模型，并且给出以EI30为降雨侵蚀力指标的计算误差；三是根据次EI30研究降雨侵蚀力的季节分布。研究得出的主要结论如下：1 一次降雨的平均雨强与其最大时段雨强是有差异的。当降雨历时小于30min时，由于无法用常规方法求I30而常用平均雨强代替，这样会导致降雨侵蚀力的计算误差，影响土壤流失量预报精度。通过对全国32个站共8055次降雨不同历时最大30min雨强与平均雨强关系的分析，提出了短历时情况下，利用平均雨强与最大时段雨强的幂函数关系，计算历时不足30min时的最大30min雨强的关系为：I30=1.0192Iave1.0113（R2＝0.995）；历时不足10min、15min、60min时I10、、I15和I60的公式为：I10=1.0575 Iave1.0457（R2=0.9770）；I15=1.0638Iave1.0351 （R2=0.9769）；I60=1.0923 Iave1.0126（R2=0.9847）。2 利用12个径流小区310次产流资料，建立了EI30与PI30的关系式为EI30 = 0.2143（PI30），R2 = 0.9758。与EI30相比，平均相对误差为33.69％，其中相对误差绝对值小于20％的占52.12％。EI30与PI10的关系式为 EI30 = 0.163（PI10），R2 = 0.9338。与EI30相比，平均相对误差是40.63％，相对误差绝对值小于20％的占33.31％。利用PI30估算效果更好。3 利用9个气象站1819日侵蚀性降雨量资料建立了日EI30的简易算法为：（EI30）日＝0.2407（P日I30日），R2＝0.9835。与日EI30比较，33.33％的降雨侵蚀力估计误差在20％以内。每日降雨侵蚀力估计的平均相对误差为53.42％。（EI30）日＝0.1716（P日I10日），R2＝0.9124。仅有18.58％的估计误差绝对值在20％以内，每日降雨侵蚀力平均误差为67.73％。4 利用13个站点均超过20年的次降雨过程资料分别模拟方程，简易算法参数α、β显出一定规律性，其中α值小于1的有5个站点均位于山区。次雨量简易算法计算次降雨侵蚀力时，相对误差较高，如果累加成年降雨侵蚀力时，平均相对误差均小于50％，计算多年平均值时，误差更小。通过研究分析，如果剔除一部分平均雨强较小的降雨，可以提高降雨侵蚀力的计算精度，特别是次降雨侵蚀力。另外以蒲洼、三家店、番字牌的806次降雨为例，对其进行历时分类模拟，其中历时小于等于6小时降雨模拟效果较好。次雨量模型要估计次降雨侵蚀力或者更短尺度的降雨侵蚀力，依靠雨量来估计较粗略，误差也较大。针对目前较易得到的气象常规观测资料，不适宜建立次雨量简易算法估计次降雨侵蚀力。但是次雨量模型计算年、多年平均降雨侵蚀力精度较高，鉴于资料以及已有研究基础，建议北京地区利用次雨量计算年或者多年平均降雨侵蚀力时使用两个模型：一是山区次雨量模型： ；二是平原次雨量模型： ；五寨代表山西西部山区站点，其方程为： ；阳城为山西平原站点，其方程为： ；宾县为黑龙江地区站点，方程为： 。以上方程在计算年、多年平均降雨侵蚀力时，平均误差均在50％以内。5 日雨量简易算法计算日降雨侵蚀力时仍然误差较大，不适宜使用。日雨量模型计算年、多年平均降雨侵蚀力以及计算降雨侵蚀力的季节分布时精确度较高。以现有资料为基础，建立北京日雨量模型： ，R2＝0.72，此方程估算年降雨侵蚀力的平均误差为11.3％，估算多年平均年R值平均相对误差6.6％。另外北京的月雨量模型为 ，年雨量模型为 。日、月或年雨量估算年降雨侵蚀力时，模型的误差均较小，约有一半的样本相对误差绝对值小于20％，三个模型相比，日雨量模型估算的年降雨侵蚀力平均相对误差最小。日、月或年雨量估算多年平均年降雨侵蚀力时，模型的误差最小，所有样本的相对误差绝对值均小于20％，平均相对误差绝对值最小值只有0.8％，最大值小于7％。三个模型相比，日雨量模型的估算精度最高。6 利用全国39个站点的次降雨过程资料求得多年平均半月EI30累积百分比。在限制SLogistic曲线上限等于100的情况下通过非线性模拟，各站点参数k值没有显著差异，均在0.3～0.5之间。 a值变化较大，较小的一般为南方地区，此地区出现侵蚀性降雨的时间比较早，并且最初降雨侵蚀力的增加速度也较大。北方地区在侵蚀性降雨较少地区，并且出现侵蚀性降雨时间较晚的地区，a值往往较大。各站点SLogistic曲线参数值，基本上有一定的地区差异，从侵蚀性降雨量多的地方到侵蚀性降雨量小的地方，参数a值逐渐增大。

Soil erosion has become one of the most important environmental problems in the whole world recently. With the development of the society, the soil erosion problem is getting into a serious one that will directly threaten the human being’s survive and development. It is very necessary for China to accelerate the prediction approach of soil erosion which is quite serious in this country. Depending on the effecting agents, soil erosion could be defined into four classes: water erosion, wind erosion, freeze-melt erosion and gravity erosion. In China, the water erosion aroused primarily by the rain is the most active factor in the erosive field. The potential erosion ability could be calculated by using the index of rainfall erosivity. It is the fundamental factor for quantitative soil erosion prediction. It is important for soil conservation plan to estimating the rainfall erosivity in China and analyzing the temporary and spatial variation of soil erosion. This scientific dissertation has three aims: One is research on methodology for calculating the maximum 30 minute rainfall intensity of short duration storms. The second is to establish the simple method of daily and event rainfall. The third part of this dissertation is the distributions of event rainfall erosivity. 1 The maximum 30 minute rainfall intensity for a storm, which is calculated from breakpoint rainfall data taken from continuous recording rain gauge charts. However, to estimate maximum 30min rainfall intensity, different methods were used. The objective of this study was to provide a method for calculating the I30 When the duration of a storm is less than 30min. 8055 storm events from 32 stations in China were used. 4792 storms of 30 stations which the duration was more than 30min were used to develop equations of I30 and the average intensity (Iave). The results showed that the power function was better fitted than the linear function. The duration of the storms had influenced on the I30 estimations. The less the duration of the storms, the better the I30 estimations. We used storms with 30-40min duration and gave the estimated function：I30=1.0192Iave1.0113，R2＝0.995. The comparison analysis showed that rainfall erosivity could increase by 34.2％ by using this equation to estimate I30 compare to those by using usual method of Iave. The method was further developed for estimating maximum 15min intensity for storms with less 15min duration, and maximum 60min intensity for storms with less 60min duration, and maximum 10min intensity for storms with less 10min duration: I15=1.0638 Iave1.0351, (R2=0.9769), I60=1.0923 Iave1.0126(R2=0.9847), I10=1.0575 Iave1.0457（R2=0.9770）.2 310 event rainfall from 12 plots were to estimate the relation for transferring PI10、PI30 into EI30. The first transferring formula is EI30 = 0.2143（PI30）, R2 = 0.9758；The second is EI30 = 0.163（PI10），R2 = 0.9338. The above simple model can give rise to error. The average absolute relative error of the first model is 33.69%, and the relative error below 20% is 52.12%. To the EI30 = 0.163（PI10）, the above error value is 40.63%、33.31%.3 Using the data of the erosive rainfall based on the data in 9 stations and 1919days records, establish the simple formula of daily EI30 is: （EI30）d＝0.2407（PdI30d），R2＝0.9835. Compare with daily EI30, 33.33% of rainfall erosivity error less than 20%, and the average relative error for the rainfall erosivity for every rainfall is 53.42%. (EI30）d＝0.1716（PdI10d），R2＝0.9124, only 18.58% of the data has an average absolute relative error within 20%, and the average error of rainfall erosivity for every rain event is 67.73%.4 By using the 13 stations’ records in over 20 years time, we can simulate the formula, and we can find there is some rules between α、β: The 5 stations which have “α<1” are all located in the mountain area. When we use the “event rainfall simple method” to calculate the rainfall erosivity, there will always be a big relative error, but if we calculate it into yearly rainfall erosivity, and the average error will less than 50%. Further more, if we calculate the rainfall erosivity into average yearly erosivity, the error will be smaller. According to the research, if those rainfalls with weaker rainfall intensity can be ignored, the precision of the rainfall erosivity calculation can be increased, especially for the calculation for every rain event. By setting a time-classified simulation based on the data of 806 rain events from Puwa、Sanjiadian、Fanzipai，and the results show that the event rain with 6 hours for one single event rain base on the simple weather records that we can find. However, it has a great precision on the calculation of yearly or average yearly R value. According to those resources and research, we suggest in Beijing area there are two models can be used to calculate the yearly or average yearly rainfall erosivity, The first one is mountain area model: ；The second one is the plain model: . Wuzhai is the station in west mountain area of Shanxi province, the formula is ；Yangcheng is in the plain area of Shanxi province，and the formula is ；Binxian is the station in Heilongjiang province，formula is： . All of the above formulas have an average error of less than 50% when we calculate the yearly or average yearly R value.5 There is a big error when we use the daily rainfall volume model to calculate the daily rainfall erosivity, and it’s not suitable for use. However, the daily rainfall volume model has a very high precision on the calculation of yearly or average yearly R values and the seasonal rainfall erosivity distribution. By using the recently resources, establish the Beijing daily rainfall volume model: R2＝0.72, this model has a error rate of 11.3% on the calculation of yearly rainfall erosivity, and 6.6% on average yearly R value. And also, the monthly rainfall volume model for Beijing is , yearly rainfall volume model is .All of those three models have a small error rates on the yearly rainfall erosivity calculation, compare the three models, the daily rainfall model has the smallest error. Those three models have a small error rate on the yearly rainfall erosivity calculation, all of the samples have an average absolute error less than 20%, and the minimum average absolute error is 0.8% and the maximum error less than 7%. Compare the three models, the daily rainfall model has the smallest error.6 Based on the event rainfall data from 39 stations all around the country, we can get the semi month EI30 for years. When given a top limited of 100 for the SLogistic curve, the nonlinear simulation shows that every station’s “K” value has no significant different, all between 0.3 and 0.5. The “a” value has a great change, normally, due to the south area has the erosive rainfall earlier, and the rainfall erosivity increases faster, so the “a” value only slightly changed；On the other hand, the north area has less and later erosive rainfall, so the “a” value often higher. There shows a regionally different on the values of the SLogistic curve for different stations, and the value of “a” increases from the more erosive rainfall area to the less erosive rainfall area.