随着人口的增长和技术的进步，人类对土地资源的开发强度越来越大，由此产生的土壤侵蚀也成为世界性的环境问题之一。定量计算土壤流失量是合理利用土地资源、布设水土保持措施以及设计水利设施的重要科学依据之一，土壤可蚀性是定量计算土壤侵蚀量的重要指标。通用土壤流失方程（USLE，universal soil loss equation）中选用的土壤可蚀性指标K为单位降雨侵蚀力在标准小区上引起的土壤流失量，其中标准小区定义为坡度9%、坡长20 m的连续清耕休闲地。由于依靠天然降雨测定土壤可蚀性需要时间长且费用昂贵，所以需利用微小区进行人工降雨试验。研究短坡长与长坡长小区上的土壤可蚀性模拟实验结果之间的关系，有利于实验结果之间的比较和正确应用人工模拟降雨的结果。本文利用人工模拟降雨实验研究了坡长对侵蚀影响。采用变雨强，每场实验降雨2h，总雨量121mm，选择1m、2.5m、4m、5.5m、7m、8.5m、和10m等7个坡长，每个坡长进行干运行和湿运行。分析了侵蚀产沙过程、侵蚀产沙与坡长的关系、侵蚀产沙的空间分布及坡面流速，得到了以下结论：（1）坡长大于4 m的小区的含沙量及单位时间侵蚀模数随降雨过程发生跳跃性波动。这是由于细沟的发生和发展导致径流含沙量间歇性突增而产生的。（2）在降雨的最后30-40 min，各坡长单位时间侵蚀模数差别减小，这是主要是结皮和雨强减小造成的。（3）从1 m到2.5 m，次降雨平均含沙量和侵蚀模数变化不大，而4 m以上，随坡长的增加而增大。湿运行的侵蚀模数小于干运行的侵蚀模数，这是土壤表面结皮的结果。空间分布上，在2.5-4 m坡段，含沙量和侵蚀模数增大显著，5.5 m以后的坡段的含沙量和侵蚀模数显著大于1-4m的坡段。（4）细沟间流速从0.5 m到1.5 m迅速增加，2.5 m以后，随着坡长增加，流速保持不变甚至减小，且小于同坡段的细沟流速。在测量流速的时段，细沟沟头达到距坡顶2-4 m的位置，从此至5.5 m，细沟中径流的速度逐渐增加，增加的斜率逐渐减小，超过5.5m以后，流速趋于稳定。（5）利用本实验的数据得出在1-10 m范围内，坡长因子的指数m值为0.55405，用微小区测定土壤可蚀性K值时，可用该数值将侵蚀结果订正到标准小区求K值。本实验的局限。首先，在4m以下的坡长选择得较少，不能充分揭示短坡范围内坡长对侵蚀产沙的影响。其次，实验重复次数较少，重复间数据有一定差异。第三，本研究缺乏人工模拟降雨同天然降雨结果的比较，使得本研究结果的推广受到限制。所以，在今后的研究中，可以在4m以下取更多的坡长进行实验；同时进行更多组重复；有条件的情况下应将降人工模拟降雨结果同适当的天然降雨资料进行比较，为人工模拟降雨实验结果的应用提供基础。

When using the empirical model, Universal Soil Loss Equation (USLE), to predict soil loss, Soil erobility (K) is an indispensable factor, which is defined in USLE as rate of soil loss per rainfall erosion index (R) as measured on a unit plot. Soil erodibility factors are best obtained from direct measurements on natural runoff plots of a long observing period. Time and economic factors have limited the establishment of long-term runoff plots and therefore the development of plot research with simulated rainfall was promoted. However, the size of unit plot (22.1 m in length and 5 m in width) is too large for simulating rainfall experiment; therefore, microplots of shorter length are employed. Unfortunately, erosion processes on microplots are different from those on unit plots because of the difference in slope lengths; consequently, the K values obtained on microplots are not necessarily representative. The purpose of this research is to examine the relationship of soil loss on different slope lengths and set up a slope length standard, below which the microplots are suitable for determine K values.For the above purpose, microplots of 7 slope lengths, 1m, 2.5 m, 4 m, 5.5 m 7 m, 8.5 m, and 10 m, were selected. Simulated rainfalll of Multi-intensity were employed to imitate a representative storm. A single rainfall, consisting of 12 periods, lasted 2 hours, with a total rainfall amount of 121 mm. Experiments for each slope length consisted of dry run and wet run, which were conducted 24 hours after dry run. During the rainfall, runoff samples, which were used to calculate the totle runoff and soil loss, were taken every 5 min. The flow velocity of runoff in both rill and interrill area were measured with dyeing method during the same period in every rainfall.By analyzing the data, the following conclusions were drawn:(1) On plots longer than 4m, sediment concentration and soil loss modulus per unit time pulse during the process of rainfall. This pulse is probably attributed to the formation and development of rill.(2) During the last 30-40 min of rainfall, the difference in soil loss modulus per unit time between plots of different slope length decreases. This decreasing is due to the crusting on surface, which prevents the soil from being detached by raindrop, and the decreasing in rainfall intensity, which is unable to detach and carry soil particles.(3) On the basis of single rainfall event, sediment concentration and soil loss modulus don't show a significantly increase from plots of 1 m to 2.5 m, while increase considerably with slope length above 4m. Moreover, the soil loss moduluses for wet run are small than dry run, and this can also be attributed to crusting.(4) From slope 0.5 m down slope from upper end of plot to 1.5 m, flow velocities on interrill area increase and level beyond 1.5 m, while flow velocities in rill become approximately stable beyond 5.5 m.(5) Within 10 m, 0.55405 is recommended for the exponent m for slope length factor, L, as defined in USLE, and this value can be used to calculate soil erodibilty with microplot experiment.There exist some drawbacks of this study and suggestions about future experiment. Firstly, experimenting slope lengths less than 4 m are not enough, so the effects of slope lengths on soil loss are not revealed sufficiently. Secondly, the longest testing slope length, 10 m, is still short compared to unit plot, and longer plots should be involved to establish the relationship between plots shorter than 10 m and units. Thirdly, the replications the treatment are few; more replications are needed to get a stable data. Finally, the results of this study are not compared with natural rainfall data, that restricts the extrapolation of this study to actual conditions.