基于SOFM网络的云南省土地利用程度类型划分研究

土地利用程度研究是开展土地整理工作和土地可持续利用研究的重要内容之一。本文尝 试以人工神经网络技术作为土地利用程度类型划分工作的理论支撑, 构建了自组织特征映射 SOFM网络, 在分类过程中同时考虑土地利用程度现状情况和影响土地利用程度的社会经济因 素, 最终将云南省土地利用程度分成高土地利用程度- 高人口压力- 高经济压力区等六种类型, 以 为区域土地管理宏观调控提供科学依据。

Classification of Land Use Degr ee in Yunnan Province Based on SOFM Networks

Study on land use degree is one of the superiority fields of land arrangement and sustainable land use research.Classification of land use degree provides guidelines for utilization and conservation of regional land use as it can indicate regional differentiation regularity and existent problems.A considerable amount of research has been done on land use degree during the last decade.In this paper,Yunnan Province is taken as a case and unsupervised artificial neural network,namely Self-Organizing Feature Mapping(SOFM),is used in land use degree classification.The results indicate that classification of land use degree based on SOFM networks is a promising approach to land use studies.In this paper,Multiple Cropping Index is employed to the land use degree model so as to indicate the quality differences within a specific land use type.More improvements of the model should be brought through by further consideration.As for the data employed as input for training,not only the status quo of land use degree but also the influence factors are included.After the iterative learning phase in the SOFM analysis,six output units representing different classes of land use degree come forth,i.e.,High land use degree-high population pressure-high economy pressure region,High land use degree-medium population pressure-medium economy pressure region,Low land use degree-medium population pressure-medium economy pressure region,Low land use degree-low population pressure-low economy pressure region,Medium land use degree-medium population pressure-low economy pressure region,and medium land use degree-low population pressure-low economy pressure region.Accordingly,some advice on utilization and conservation of land use is proposed based on the studying result.From the results obtained so far,it seems that SOFM is superiors over others in many aspects and has been trained to perform complex functions in various fields of application,including land use degree classification.But more improvements should be conducted before further applications.