| 摘要: |
| 以全国典型农业大县—吉林省梨树县为研究区，在1987年、2003年和2019年Landsat 5/TM、Landsat 8-OLI影像的基础上，进行支持向量机分类（SVM）。同时，以土地利用动态度、土地利用程度、信息熵等为主要方法实现对梨树县30余年土地利用动态的定量表达。结果表明：（1）耕地、林地、草地和建筑用地的变化最为显著，其中耕地面积增加态势明显，1987年、2003年、2019年面积占比均保持在75 %以上。（2）林地形态较为集中，面积持续减少，以分布在东南部为主；水域、草地主要接受来自耕地的转入，呈现出“先减后增”的态势，表明这一时期人类干预对土地利用起到积极作用。（3）建筑用地不同程度地接受来自耕地、林地、其他用地的转入。（4）从信息熵原理来看，梨树县土地利用类型分布不均匀，均衡度较低，均质性较弱。 |
| 关键词: 土地利用 梨树县 信息熵 土地利用变化模型 |
| DOI： |
| 分类号: |
| 基金项目:中央财政林业科技推广示范资金项目(吉推201416)；吉林省科技发展计划项目(20180101085JC) |
|
| Dynamic analysis of land use in Lishu County of Jilin Province from 1987 to 2019 |
|
YANG SHENGJIE
|
|
Central China Normal University, 152 Luoyu Road, Hongshan District, Wuhan City, Hubei Province
|
| Abstract: |
| The support vector machine method (SVM) is performed to classify the land cover types of Lishu County, Jilin Province, a typical county dominated by agricultural production, with the Landsat 5/TM and Landsat 8-OLI datasets in 1987, 2003 and 2019. Land use dynamic attitude, land use degree and entropy are used to quantitatively describe the land use dynamics of Lishu County over 30 years. Our experimental results show that cropland, forest land and grassland and construction land undergo the most obvious changes. The details are as follows: (1) cropland had an increased area, of which the proportion maintains at more than 75% in 1987, 2003 and 2019. (2) the distribution of forest land was mainly concentrated in the southeast of the study area, and the area of forest land continued to decline. (3) water area and grass land decreased first, then increased because of the sections transferred from cropland, which can indicate that human intervention played a positive role in land use during this period. (4) a part of construction land was transferred from cropland, forest land and other land use types to varying degrees. According to the principle of information entropy, the distribution of land use types in Lishu County is uneven, less balanced and less homogeneous. |
| Key words: land use Lishu County information entropy land use change model |