The Method Compare Research Of The Cultivated Land Evaluation Based On Kaijiang County

With such an immense population base and scarce mineral land resources, the contradiction between populations and land are extremely serious, the per-capita possession of land resources is far below the world average. Along with rapid economic and social development, the ever-increasing demand of land used for city construction causes sharp reduction of farmland; on the other hand, the rapid growth of GDP caused a serious impact on ecological environment, following by the quality of farmland getting worse. All of these not only constitute a serious challenge on food security but also make the committee growing concern, more and more researches are on evaluation of farmland.In the present study, selecting Kaijiang Country, Sichuan Province as studying area, in the support of geographic information system and computer science technology, using the second national soil survey data and measured soil testing and fertilizer data, utilizing the Delphi method to select the11influencing factors as evaluation indicators for Farmland of the study area, and performing Analytic Hierarchy Process and the neural network model to complete the evaluation of Farmland.In the process of evaluation using the analytic hierarchy, with the help of Photoshop for processing of scan maps of the study area, and then using professional drawing software to the formation of vector data; Using ArcGIS9.3as spatial data processing tool, property data were handled in Excel, Access and other statistical analysis software; Based on the County Land resource Management information System, establishing judgment matrix and determining the weight of each factor; Using fuzzy mathematics approach, according to the characteristics of each factors and the impact on farmland fertility, membership function was established, and then was used for calculating the composite index of farmland fertility and the division of soil fertility level, and evaluation of farmland was ultimately finished.Using neural network model for evaluation of farmland, mainly with the BP in the MATLAB software, a random sampling, taking into account the principle of the balance of geographical distribution and level evenly distributed, we selected8664evaluation unit for exercitations sample set, after58exercitations with the target error achieved, the exercised neural network model was employed to evaluate soil fertility for the sample. Finally, the evaluation results obtained by both methods were verified using the level of food production as standard. Through the comparative analysis of the distribution of regional variation and distribution of area changes, it was indicated that in the fitness with food production level, the results obtained by the analytic hierarchy process was significantly higher than the results obtained by the neutral network model.Using analytic hierarchy process for farmland evaluation of Kaijiang country is in line with universal trend of current National Evaluation of Farmland, with a high scientific, rationality and practicality. It can guide agricultural production of Kaijiang County, promote scientific protection and rational use of farmland, and create conditions for sustainable development of agriculture.