Land Suitability Evaluation Research And Application Of Neural Network Model

In recent years, along with the fast development of economy, the inflation of population and the acceleration of urbanization, specially the increase of the area of the built region, the conflicts crop up between the relationships of population and land resources, environmental protection and economic development, the background and situation of the land construction and prospective planning have changed. With the new round of the land construction and prospective planning have begun, the land suitability evaluation is launching in order to alleviate the contradiction between population and land resource. The scientific evaluation of land suitability is an important basis and tool to achieve its goals of sufficient, reasonable and sustainable utilization of land resources, as well as to perfectly coordinate the relationship between land resources exploration, utilization and land protection.Based on the land suitability evaluation and BP neural network (BPNN) theory, this paper designs a commonly-used BPNN model to evaluate the land suitability, establishes the evaluational index system and the evaluation criterion for the land of construction, and explores a quantitative way to the suitability evaluation of the land of construction. In the end the systematization, automation, quantification and the veracity of evaluational results is realized and the efficiency of evaluation is improved. So this gives the further analysis of the land construction and prospective planning to provide the convincing basic and increase the persuadation of the program and decision, the objectivity of land planning.According to the land use planning of HengQin island, ZhuHai city, the paper mainly discusses the feasibility of application, network structure, learning algorithm and the improvements of BPNN model in land suitability evaluation, puts forward with the Levenberg-Marquardt algorithm of BPNN model to evaluate the suitability of the land of construction, establishes the evaluational index system and the evaluation criterion of the land of construction in HengQin island and uses VC++ 6.0 language to develop the land suitability evaluational neural network model. Through applying the data of the present land using diagram, the land type diagram, the present geology diagram, the present land resource diagram and the present traffic diagram of HengQin island, the writer takes the spatial analyst function of ArcGIS 9.0 to complete data operation and spatial analyst, including grid DEM, data converter, layer stitching, spatial overlay analyst, buffer analyst, scores computing of evaluational units and attribute value computing of some evaluational factors (for example, slope, height and traffic location), realizes the automatic evaluating process which includes choosing evaluational objects, determining evaluational units, selecting evaluational factors, building the standards of classifying and quantifying evaluational factors, performing the BPNN model, and outputting evaluation results, divides into the highly suitable region, the medium suitable region ,the narrowly suitable region, the unsuitable region of land developing suitability, and completes the land suitability evaluation of the land of construction in HengQin island so as to provide scientific basis to the land planning and management.At present the relative researches on the suitability evaluation of the land of construction at home or abroad are insufficient. There were three attempt in this paper: (1) This paper designs a BPNN model to evaluate the land suitability ; (2) The Levenberg-Marquardt algorithm is applied to the land suitability evaluation; (3) the evaluational index system and the evaluation criterion of the land of construction is established. The findings indicate that based on the combination of the land suitability evaluation and the BP neural network, the land suitability evaluational BP neural network model is feasible and deserves being explored further.