摘要
运用元胞自动机结合人工神经网络方法,以遥感和GIS技术为依托,对庆城县水土保持世行二期贷款项目区的土壤侵蚀进行了动态模拟和预测,探讨了元胞自动机模型在土壤侵蚀预测方面的重要作用。应用BP神经网络与元胞自动机相结合自动挖掘出元胞自动机模型的转换规则,不仅降低建立多种土壤侵蚀类型预测模拟的难度,而且减少了人为的主观因素,提高了准确性。预测的结果与实际土地覆被类型比较得出预测精度达到70%以上。通过对该地区未来5年土壤侵蚀变化的预测,反映出了该地区生态环境变坏的趋势,为土壤侵蚀的动态监测与水土保持规划提供了参考和决策支持。
It is very important to dynamicaly simulate and forecast the development and evolutionary process of soil erosion. This paper presented a model to simulate and forecast soil erosion with geographic cellular automata,in the foundation of remote sensing images and GIS softwares. The importance of cellular automata expounds in dynamically forecasting and simulating the trend of soil erosion was also expounded, which will provide the important reference for the management and development of water and soil loss. The model, which uses the back-propagation neural network to mine transition rules of cellular automata, has a lot of merits that not only reduces the difficulty to establish the transition rule because of many kinds of soil erosion types,but also reduces subjective idea and then improve the simulation accuracy. Finally,this model is applied to the Loess Plateau Qingcheng Project Area. The result,which owned accuracy of about 70 %, has showed that the model was suitable for simulating wetland soil erosion evlovement. The environment of study area will be improved in the coming years according to the simulation.
出处
《土壤通报》
CAS
CSCD
北大核心
2009年第4期902-906,共5页
Chinese Journal of Soil Science
基金
甘肃省自然科学基金项目(0710RJZA104)
黄土高原水保世行二期贷款新技术推广研究项目(050321)资助
关键词
元胞自动机
神经网络
土壤侵蚀
预测与模拟
Cellular automata
Neural network
Soil erosion
Forecast and simulation
