Knowledge Of Soil-Landscape Model Obtain From A Soil Map And Mapping

Soil is the loose mineral on the Earth’s surface that can support the growth of plants, which is the most basic resource of human survival. It provides soil nutrients of the physical and chemical places, which will directly affect the structure and nature of agriculture production. With the rapid development of precision agriculture and urbanization implemented, people need accurate soil information. So understanding the status of the soil, marking the latest soil maps is of great significance for land use and agriculture production. Digital soil mapping technique aging fast, cheaper manpower and resources, which was used to update soil maps have wider practical significance. Second National Soil Survey retained the precious soil map data, soil type are closely linked to various environmental factors. So the relationship between the soil and the landscape variations implied in the soil map is a key of mapping. In this paper, the original soil map is used for data mining. It take the relationship between the soil and the landscape variations to infer soil type map of the study area, when obtain the relationship. During the research process, part of the jintang and gaojian town in Chongyang County of Hubei province is the study area. The contents and main result of this study included three parts as follows:(1) Based on the mechanistic model for soil development, in the study area, terrain was the main factor. So six variations, such as elevation, slope, aspect, profile curvature, plan curvature, topographic wetness index were chosen to predict soil types in the study area. The data were obtained from the software of Arcgis9.3.(2) Extracting several samples from the original soil map by the grid sampling and manually add points method. After the specification of the data, using the decision tree algorithm to classify the data, and finally get the knowledge.In the study, cyclic delete misclassification was used to process the tree with See5, and reduced the misclassification rate from19.5%to5.2%.(3) The more detailed decision tree classification, the more rules. Reference to the relevant soil information, it reduced some meticulous branches. It took elevation as the main field, then sorted all the rules, at last the knowledge of soil-landscape model was finished. Quantified on the knowledge of soil-landscape model with the similarity model, expressed each rules in the form0to1. SolIM Solution5.0was used to predict the soil type, and obtained the membership distribution chart. After all the soil types charts were hardened, the soil type map of the study area was generated. The map would be verified, it took field sampling method to obtain accuracy. Finally, the Kappa coefficient was0.5. There was moderate consistent with the actual spatial distribution of the soil in the study area.