The Study On Typical Granite Basin On Simulating Of Runoff And Soil Erosion Monitoring

The data for studying in this paper was based on the documents recorded by the relative meteorological stations located at the Tangbei and Lanshui river basins in Xingguo county of Jiangxi province, which were mainly about the daily rainfall for many years, runoff, and individual rainfall for flood. After correlative analyzing these data with the runoff, some values were got and eventually chosen as parameters for BP Artificial Neural Network Model which was used to predicting the rainfall and runoff of the above two rivers' basins. Then for the quantitative research on soil erosion in Lianshui river basin was done by such ways as RUSLE modeling, basin bayonet observing and visual interpretation of remote sensing image. Some conclusions were got as follows:1) Model for the runoff yield of daily rainfall along Tangbei River: Pt (mm) meant rainfall for observing day, Pt-1(mm) meant rainfall for the day before the observing day, ft-1(mm) meant day evaporation and output variable Q(m~3/s) meant average daily discharge. In 1987, the relative error was 7.6% as wellas in 1983 the relative error was 30.9%.2) Model for the runoff yield of individual rainfall along Tangbei River: the input variable such as P(mm) meant individual rainfall,duration of individual rainfall meant T(min) and duration of flood (min), while the output variable Q(m~3/s) was mean discharge for individual flood and the Nash Efficiency Coefficient of model was 0.94. Then the prediction for total discharge after 30 times rainfall along Tangbei River in 1987 was 37.51M~3/S, the observing was 35.24 M~3/S, and relative error was 6.4%.3) Model for the runoff yield of daily rainfall along Lianshui River: In 2000, the relative error was -3.6%. As for 2008, with the error of 4.6%.4) If a basin with medium size was monitored by RUSLE Model, no scale between 1:50,000 and 1:100,000 would affect the measuring result a lot. The difference only manifested on the degree of soil erosion which would be determined as mild on the one hand and micro-degree on the other.5) the interactive visual interpretations between men and mechanic were able to monitor the largest amount of soil erosion, which was about 6.2 times as that for observing at basin bayonet station as well as 1.9 times as that for model monitoring. The difference was caused by the difference between soil erosion and soil loss under the effect of space and scale. The value got from visual interpretation should be considered as the amount for soil erosion, the observing value from basin bayonet station might be seen as amount for soil loss, and the result from model with the large scale could be nearest to the real condition.