中国耕地熟制变化遥感监测研究

发布时间：2018-03-17 20:28

本文选题：GLASS　切入点：LAI　出处：《西安科技大学》2015年硕士论文　论文类型：学位论文

【摘要】：多熟种植是我国重要的耕作方式,它对保持和增加粮食的产量以及促进农村经济的发展都有十分重要的意义。我们国家耕地的熟制格局复杂多样,在过去的三十年里,我国的耕地资源在数量、质量和利用结构方面因受自然条件以及农村社会状况的影响发生了复杂的变化,这就需要我们能够很好地掌握全国耕地资源的情况和粮食生产的形势,以对估计粮食产量的变化及其原因做出科学的决策。本文以全国范围为研究区域,利用北京师范大学提供的GLASS LAI数据,用TIMESAT软件对时间序列的GLASS LAI数据进行重构,得到了能体现不同熟制的时序GLASS LAI平滑曲线,运用R语言编程求平滑曲线极大值个数(即曲线波峰数目)的方法实现了全国范围内耕地熟制信息的自动提取,将2000年中国1:10万的地表覆盖数据的耕地层数据对结果进行裁剪,最终得到1982年、1992年、2002年、2012年的全国耕地熟制格局分布图,对我国近三十年来耕地的熟制变化进行了分析,并计算得出复种指数,将所得结果分别与统计资料及前人的研究成果进行了对比验证。主要的研究方法及结果如下:(1)对我国32个省(直辖市、自治区)共12年的时间序列GLASS LAI数据进行重投影、拼接、裁剪等预处理,建立时间序列GLASS LAI曲线,由于影像中存在噪声的影响,时序GLASS LAI曲线偶尔会出现锯齿形,利用以Matlab为运行载体的TIMESAT3.0软件中提供的Gaussian滤波、Logistic滤波以及Savitzky-Golay滤波(简称S-G滤波)对曲线进行重构。结果显示,Gaussian和Logistic两种滤波不能很好的拟合二熟、三熟作物的曲线,这会造成监测结果与实际情况的严重不符,而S-G滤波能够更好的拟合不同熟制的时序曲线,这能够较好地展现农作物季节性的变化情况,有利于提取波峰数量。(2)依据植物的物候特征所形成的时序曲线的变化特点,耕地熟制就相当于时序曲线上满足条件的极大值点的个数。将上述得到的能体现不同熟制的时序GLASS LAI平滑曲线以时序数字的形式存储成文本格式,运用R语言编程的方法来提取曲线上满足条件的极大值点的个数。结果显示,此方法可以提取每个像元的时序曲线波峰个数(0代表非耕地,1代表一年一熟,2代表一年两熟,3代表一年三熟)。(3)从最终得到的1982年、1992年、2002年、2012年的全国耕地熟制格局分布图中可以明显地看出我国耕地熟制从北到南由单一到复杂,由一熟区到二熟区再到三熟区的渐变分布,一熟区、二熟区多集中分布,三熟区在零星分布的基础上存在部分小区域的集中。近三十年来,我国耕地熟制的动态变化情况为:一熟区总体变动不明显,但区域范围内有变动,部分区域增加的同时部分区域缩减;二熟区明显增加,以河南省为中心向四周扩展;三熟区呈零星区域的出现并扩展缓慢。(4)从ArcGIS成图的属性信息中获取我国32个省(直辖市、自治区)一熟、二熟、三熟的像元个数来计算复种指数。结果表明,近三十年来我国的复种指数整体呈持续上升的趋势,这说明越来越多的耕地实现了一年两熟和一年三熟的耕作方式。将2002年的复种指数与统计资料以及前人的研究结果进行对比。结果表明,本方法监测的结果与统计资料和前人监测的结果均呈正相关关系,相关系数分别为0.788和0.903,验证了本文结果的可靠性。
[Abstract]:Multiple cropping is an important way of farming, it is very important to maintain and increase food production and promote the development of rural economy. Our country farmland cropping system pattern is complicated, in the past thirty years, China's cultivated land resources in quantity, quality and structure for use affected by natural conditions and rural social status changed complicatedly, which requires us to have a good grasp of the situation of cultivated land resources and grain production situation, to make a scientific decision for the estimation of grain yield changes and its reasons. Based on the country as the study area, GLASS using LAI data provided by Beijing Normal University. GLASS LAI, reconstruct the data of time series using TIMESAT software obtained can reflect the timing of GLASS LAI smooth curves of different cropping system, for the smooth curve is the use of R language programming A large number of values (i.e. curve peak number) method to realize the automatic extraction of nationwide land cropping information, cutting results are 2000 China 1:10 million land cover data of arable layer data obtained in 1982, 1992, 2002, 2012, the national cultivated land cropping pattern distribution, cropping system the change of cultivated land in China in the past thirty years were analyzed, and calculated the cropping index, the results with statistical data and research results were verified. The previous research methods and results are as follows: (1) in 32 provinces in China (municipalities, autonomous regions) splicing time sequence GLASS LAI data for 12 years of heavy cutting, projection, pretreatment, a time series GLASS LAI curve, because of the influence of noise existing in images, temporal GLASS LAI curve occasionally appear jagged, using Matlab as its carrier TIME Gaussian filter SAT3.0 software, Logistic filter and Savitzky-Golay filter (S-G filter) to reconstruct curve fitting. The results showed that Gaussian and Logistic two kinds of filter is not well cooked two, three crops of the curve, this will cause the monitoring results and the actual situation is not serious, but the S-G filter can better fitting curve of different cropping system, which can show the changes of seasonal crops, is conducive to extract the peak number. (2) according to the changing characteristics of the timing curve formed by phenological characteristics of plants, the maximum value of the number of arable cropping system is equivalent to the timing curve to meet the conditions. The timing of GLASS LAI can reflect the smooth curve of different cropping system stored in sequential digital text format, using the method of R programming language to extract the maximum point on the curve to meet the conditions of the The number of the number sequence. The results show that this method can extract the peak curve of each pixel (representing 0 of non arable land, 1 represents a year one ripe, 2 representatives of two crops a year, 3 representatives of three crops a year). (3) from the final in 1982, 1992, 2002, the national arable cropping system the pattern of distribution map in 2012, you can clearly see that China's cultivated land cropping system from north to South and from single to complex, from single cropping area to two cropping zone to the gradient distribution of three cropping area, a ripe area, two cropping areas are more concentrated distribution areas of three crops are mainly concentrated in the base part of the small area scattered on the distribution. In the past thirty years, the dynamic changes of cultivated land in China cropping system: the total change in cropping area is not obvious, but there are also some changes in the region, part of the increase in regional area reduced; two cropping area increased significantly in Henan Province as the center to the surrounding areas of three crops showed expansion; zero star region and slow expansion Slow. (4) China's 32 provinces to obtain attribute information in the graph from ArcGIS (municipalities, autonomous regions) a two cooked, cooked, cooked three pixel number of the cropping index. The results show that in the past thirty years cropping index showed an overall upward trend, indicating that more and the more arable land achieved two crops a year and three crops a year of farming methods. By comparing the 2002 cropping index with statistical data and previous research results. The results show that the method of monitoring results and statistical data and the previous monitoring results showed a positive correlation, the correlation coefficient was 0.788 and 0.903 respectively. And verify the reliability of the results.

【学位授予单位】：西安科技大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：S127;S344

【相似文献】