%0 Journal Article
%T Simulation of Farmland Use Pattern in Zhangjiakou Based on Multinomial Logistic Regression Model<br>基于多分类Logistic回归模型的张家口市农用地格局模拟
%A TIAN Yuan
%A XU Yueqing
%A GUO Hongfeng
%A WU Yanfang
%A <br>田媛
%A 许月卿
%A 郭洪峰
%A 吴艳芳
%J 资源科学
%D 2012
%I 
%X Land use/cover change (LUCC) is the main reason as well as the direct manifestation of global change, so the research on it has become the front in global change and a hot issue at present. Spatial simulation of land use pattern is an important part of LUCC research. Multinomial logistic regression model used in this paper is expanded from binary logistic regression model and can predict the distribution probability of multi-types of land use at one time, and simulate land use pattern more accurately. Based on the achievements of the second national rural land survey, we have simulated the land use pattern of cultivated lands, garden plots, forests and grasslands in Zhangjiakou, Hebei province with data about DEM, rainfall, road, population density and urban and rural residential area. The results show that: 1) The main influencing factors of distribution probability of agricultural land in Zhangjiakou are slope, rainfall, elevation, population density and distance to the nearest village. Among which, slope, elevation, rainfall and distance to the nearest village are negatively correlated with the distribution probability of cultivated lands, while positively correlated with forests; 2) Where there is a higher population density, the distribution probability of cultivated lands and garden plots are higher. GDP and total investment have little influence on spatial distribution of agricultural land. Landforms and water conditions are major factors that influence the spatial distribution of agricultural lands as well as constrain agricultural development and ecological stability in Zhangjiakou; 3) ROC inspection results of cultivated lands, garden plots, forests and grasslands are 0.7732, 0.8800, 0.7432 and 0.6750 respectively, and the spatial distribution probability of each land use type fits well with the real corresponding land use pattern. The model has revealed the inner mechanism of the land use pattern, which can be used for land management decisions and future land use scenarios simulation. Besides, the multinomial logistic regression model takes full advantage of the whole information of land use system and is an effective tool to study land use system.
%K Land use pattern
%K Multinomial logistic regression
%K Driving forces
%K Zhangjiakou<br>土地利用格局
%K 多分类logistic回归
%K 驱动因素
%K 张家口市
%U http://www.alljournals.cn/get_abstract_url.aspx?pcid=B5EDD921F3D863E289B22F36E70174A7007B5F5E43D63598017D41BB67247657&cid=B47B31F6349F979B&jid=9DEEAF23637E6E9539AD99BE6ABAB2B3&aid=A77A33C0E8506BABF581472AF5AD92E6&yid=99E9153A83D4CB11&vid=339D79302DF62549&iid=5D311CA918CA9A03&sid=3DEC1A337B83DA73&eid=D56713C22DA9FDE8&journal_id=1007-7588&journal_name=资源科学&referenced_num=0&reference_num=17