基于中高分辨率遥感的植被覆盖度时相变换方法

Abstract

PDF HTML阅读 XML下载 导出引用 引用提醒 基于中高分辨率遥感的植被覆盖度时相变换方法 DOI: 10.5846/stxb201305020904 作者: 作者单位: 黄河中下游数字地理技术教育部重点实验室,中国科学院遥感与数字地球研究所 作者简介: 通讯作者: 中图分类号: 基金项目: 中国科学院知识创新工程重大项目(KZCX1-YW-08-03);水利部-官厅密云水库上游水土保持遥感监测二期工程(HW-STBC2004-03);中国博士后科学基金资助项目(20100470994) A temporal transformation method of fractional vegetation cover derived from high and moderate resolution remote sensing data Author: Affiliation: Institute of Remote Sensing Applications,Chinese Academy of Sciences,Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences Fund Project: 摘要 | 图/表 | 访问统计 | 参考文献 | 相似文献 | 引证文献 | 资源附件 | 文章评论 摘要:植被覆盖度是衡量地表植被状况、指示生态环境变化的一个重要指标,也是许多学科的重要参数。传统的测量方法难以获取时间连续的面状数据,且耗时、耗力,很难大范围推广。遥感估算方法虽然可以弥补传统方法的不足,但由于云覆盖等天气条件的影响,获得同一时相覆盖整个研究区的遥感影像非常困难,时相的差异必然导致研究结果产生误差。针对植被覆盖度这一重要生态参数,结合低分辨率遥感数据的时间优势和中高分辨率遥感数据的空间优势,提出一种时相变换方法,将源于中高分辨率影像的植被覆盖度变换到研究需要的时相上。首先,利用像元二分模型计算MODIS尺度的时间序列植被覆盖度,并利用已经获得的SPOT影像计算其获取时相上的植被覆盖度;其次,利用土地利用图划分植被覆盖类型,并利用MODIS数据和土地利用数据之间的空间对应关系制作MODIS像元内各类植被覆盖的面积百分比数据;再次,利用面积百分比数据提取各类植被覆盖的纯像元,结合MODIS植被覆盖度时间序列,从而提取各类植被覆盖纯像元的植被覆盖度时间序列曲线;最后利用像元分解的方法提取MODIS像元内各类植被覆盖组分的植被覆盖度的变化规律,将其应用到该组分对应位置上SPOT像元的植被覆盖度上,从而将其变换到所需要的时相上。在密云水库上游进行试验,将覆盖研究区的10 景SPOT5多光谱影像计算的植被覆盖度统一变换到7 月上旬,结果显示:视觉效果上明显好转,且空间上连续一致;变换前后植被覆盖度的统计量对比结果也符合植被生长规律;利用外业样点数据与对应位置的植被覆盖度变换结果进行回归分析,结果发现各植被覆盖类型的R2均在0.8左右,表明变换结果与实测值非常接近,时相变换的效果较好,从而可以很好地促进相关研究精度的提高。 Abstract:Fractional vegetation cover (FVC) is an important index of land surface vegetation status. It is also an indicator of ecological environment changes and an important spatial parameter for various ecological modeling. The traditional methods of FVC measurement are time-consuming and labor-intensive, and thus difficult to obtain large-scale time series FVC data. Remote sensing technique is an effective approach to estimate FVC, but it is very difficult to acquire high and moderate resolution remote sensing images covering the entire study area during the same period because of the cloud cover and other weather conditions. Consequently, the FVC data derived from multi-temporal images inevitably lead to uncertain research results. To address the problem, this paper proposes a novel method to eliminate the impact of acquisition time differences on FVC from high and moderate resolution remote sensing images. For FVC data derived from images with different resolutions and acquisition dates, this proposed temporal transformation method is used to estimate high resolution FVC combined with a low-resolution time series FVC. Firstly, low and high resolution FVC data can be calculated from time series MODIS images and acquired SPOT images respectively using the dimidiate pixel model. Secondly, the vegetation cover is divided into different vegetation types based on the land use map derived from these SPOT images. And for each MODIS pixel, the area percentages of various types of vegetation cover are calculated based on the spatial overlay of MODIS image and land use data. As a result, the area percentage data represent that the area ratio of the different vegetation types within each MODIS pixel. Thirdly, the pure pixels of various types of vegetation cover can be extracted based on the area percentage data where the ratio is equal to 1, and the FVC time series curve of each type of vegetation cover can be generated based on these pure pixels and time series MODIS FVC data. Finally, the sub-pixel FVC variation of each type of vegetation cover can be extracted from MODIS pixels based on the pixel unmixing technique, and then apply them to the same location of SPOT FVC. Thus, the SPOT FVC can be transformed from its acquisition date to the specific date, which satisfies the need of our research. The feasibility of this temporal transformation method is examined in the upstream of Miyun Reservoir. The FVC data derived from 10 SPOT images are transformed to the same date of early July. The case study results show that: (1)The visual effects of the transformed FVC are significantly improved and consistent with the spatial patterns of vegetation cover; (2)The changes of FVC statistics information before and after the transformation are also in line with the laws of vegetation growth; (3) The linear regression of the FCV data on field measurement samples shows strong positive correlations between them, and the R2 is about 0.8 for each vegetation cover indicating the transformation results is close to the field measured values. The transformation results with higher precision can promote the accuracy of related researches. This method has also a certain reference value for the transformation of other parameters. 参考文献 相似文献 引证文献