The Experimental Study Of Soil Erosion Law In Western Slope Of Heilongjiang

The western semi-arid area of Heilongjiang is an important grain production base in china, however, it caused successive years of drought by complicated variable natural conditions, uneven distribution of precipitation in time and space as well as environment changes caused by long-term human intervention, drought and soil erosion have become main factors which constraint economic development, among that sloping land is the main region of soil erosion in this area, its rational utilization and management is an effective way to reduce soil loss and achieve virtuous circle of ecosystem. As a result, according to local conditions based on the natural characteristics and disaster status of the region, the study selected some conservational tillage technologies such as ridge direction farming, no-tillage with straw cover, minimum tillage subsoiling, straw cover and ridge direction farming and so on, conducted runoff plot experiment in field, with natural rainfall combined with artificial rainfall, studying slope characteristics of soil erosion in different tillage measures, different rainfall intensity, and different soil water content, and using universal soil loss equation, artificial neural network models and support vector machine models to simulate soil erosion in different tillage measures, the main conclusions are as follows:(1) The amount of soil erosion shows the characteristics of uneven distribution in season, corresponding well to rainfall, and mainly concentrated in July and August.(2) Various tillage measures have a certain influence to soil erosion, the degree is different with measures. According to merits and demerits, the control effect of soil erosion arranges: film mulching and district filed> no-tillage> ridge direction farming>minimum tillage and subsoiling>general tillage>bare land. Except general plot and bare plot, the plot's annual amount of soil erosion is at the scope of soil erosion permit.(3) It establishes daily rainfall model of rainfall erosion force, of which the precision is relatively high by testing.(4) It determines the soil loss equation in western slope of Heilongjiang by using the experimental plot data of rainfall, run-off and sediment. And the scope of error is in the permit, the equation itself and its parameters are reasonable.(5) It introduces neural network to the prediction of soil erosion, according to the model training results, selects eight hidden units to train and simulate, the result is very satisfactory and the error is also relatively small.(6) The introduction of SVM to the prediction of soil erosion, provides a new way of thinking for the prediction model of soil erosion, having certain practical value.