| 近25年来西安地区土地利用变化及驱动力研究 |
| |
| 引用本文： | 张海龙,蒋建军,吴宏安,张丽,解修平.近25年来西安地区土地利用变化及驱动力研究[J].资源科学,2006,28(4):71-77. |
| |
| 作者姓名： | 张海龙 蒋建军 吴宏安 张丽 解修平 |
| |
| 作者单位： | 1. 南京师范大学,地理科学学院,南京,210097
2. 中国科学院地球环境研究所,西安,710075 |
| |
| 摘 要： | 选取整个西安地区作为研究区，以1978年、1988年和2003年3期Landsat MSS/TM影像为基本数据源，以DEM及派生的坡度数据作为分类的辅助因子，利用改进的BP神经网络算法，提取研究区25年来的土地利用/覆被信息，并结合相应的社会经济统计数据，进一步分析引起变化的根源。研究表明，25年来：建设用地面积通过侵占周边的大量农田以及砍伐林地而得到迅速扩大，扩张了1.95倍；耕地流失严重，共有69 411.61hm2被挪为它用，减少率为每年1%；园地面积有较大增加，是1978年初面积的3.4倍，有力的推动了地区经济发展；同时水资源短缺问题日渐突出。国家政策法规对土地利用变化起宏观调控作用，铁路和公路等主要交通干线的发展是城镇建设用地迅猛发展的直接动力，人口的不断增长和快速的城市化进程导致耕地面积大量减少，而第一产业的发展和比较效益的作用使得园地面积有大幅增长。
|
| 关 键 词： | 土地利用/覆被变化 BP神经网络分类 驱动力 西安地区 |
| 文章编号： | 1007-7588(2006)04-0071-07 |
| 收稿时间： | 2005-08-25 |
| 修稿时间： | 2006-01-19 |
Analyzing Land Use Changes and Its Driving Forces in Xi'An Region During the Past 25 Years |
| |
| ZHANG Hai-long,JIANG Jian-jun,WU Hong-an,ZHANG Li and XIE Xiu-ping.Analyzing Land Use Changes and Its Driving Forces in Xi''An Region During the Past 25 Years[J].Resources Science,2006,28(4):71-77. |
| |
| Authors: | ZHANG Hai-long JIANG Jian-jun WU Hong-an ZHANG Li and XIE Xiu-ping |
| |
| Abstract: | The land use change in Xi'an during the last 25 years was investigated through land use classification of three sets of Landsat MSS/TM images acquired in September of 1978,August of 1988 and May of 2003 respectively.The BP(back propagation) neural network model has the advantage of non-linear characteristic,it can solve the problems existing in the traditional classifiers which classify images just based on the comparability of the spectral characteristic and has been gradually used in the classification of remote sensing image.In this study,the BP neural network classifier was used to classify the three sets of images.Combing with the statistic data and spatial analysis technique of GIS,the land use change and its driving forces in Xi'an in the past 25 years were analyzed.Due to the deficiency of the initial BP neural network model,great improvement was made to optimize it.In order to make the BP model converge faster,the program was made with the software of Matlab 6.5 to make the learning rate an autoregulative one,and besides,the offset and the momentum were added to the input layer and the hidden layer to store the result of the former process and to make the model easily converge to the overall minimum in stead of the partial minimum.In order to eliminate the misclassification caused by the shadow of the terrain and the angle of incidence,the DEM and the slope that was derived from the DEM were used as two accessorial nerve cells in the input layer of the BP neural network.By doing so,the BP neural network can have a shorter runtime and a higher accuracy than the maximum likelihood classification.The results showed that the area of the build-up increased sharply through occupying arable land and forestland.It was nearly twice the area of what it was in1978.The area of arable land decreased sharply with a rate of 1 percent annually.The area of the orchard increased greatly,and it was 3.4 times the area of what it was in 1978.Besides,the problem of lack water resources was getting more and more serious.Land use change is a complex process,which is affected by many comprehensive factors including both socioeconomic and natural and environment.Generally the governmental policy and the rules play a decisive role to the land use change.The expansion of the railway and the highway was the direct driving force for the rapid growth of the build-up land.The increase of the population and the rapid expansion of the build-up land resulted in large amount of plantation vanished.The increase of the orchard resulted from the development of the primary industry. |
| |
| Keywords: | LUCC BP neural network classification Driving force Xi'an region |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
| 点击此处可从《资源科学》浏览原始摘要信息 |
| 点击此处可从《资源科学》下载免费的PDF全文 |
|
