高空间分辨率全天候地表温度反演方法研究

发布时间：2019-01-03 13:14

【摘要】：地表温度(Land surface temperature, LST)是地表与大气相互作用过程中一个非常重要的物理量。及时掌握区域和全球尺度上的地表温度时空分布,尤其是全面、完整和连续的地表温度时空分布信息,对地气系统能量平衡和生态系统研究具有重要意义。卫星遥感是高效获取区域和全球尺度地表温度的最佳手段。本论文以极轨卫星数据和静止卫星数据作为数据源,开展了高空间分辨率全天候地表温度反演方法研究、基于地理加权回归的MODIS地表温度降尺度方法研究,以及MSG-SEVIRI与Terra/Aqua-MODIS地表温度产品的交叉验证方法研究。主要研究结论包括：(1)结合热红外反演地表温度与微波反演地表温度各自的优势,建立了像元内有云部分被动微波地表温度与无云部分热红外地表温度之间的有效融合模型,发展了高空间分辨率全天候地表温度反演方法。与原始的MODIS地表温度相比较,融合后地表温度在空间上变化连续,在时间上很好地反映了地表温度的年际变化。(2)利用地理空间变化的局部回归系数,建立了地表温度与归一化植被指数和地表高程之间动态回归关系,发展了基于地理加权回归的MODIS地表温度降尺度方法。通过与ASTER地表温度产品相比较,基于地理加权回归的地表温度降尺度方法的精度优于前人发展的UniTrad方法和TsHARP方法的精度。基于地理加权回归的地表温度降尺度方法的平均绝对误差和均方根误差分别约为2.3K和3.1 K。(3)通过选择SEVIRI和MODIS地表温度产品在时间、空间和观测角度上一致的像元作为匹配像元对,分析了这两种地表温度产品在不同季节、时间和地表覆盖类型下的差异。白天SEVIRI和MODIS地表温度产品的差异随季节呈现显著的变化,而晚上地表温度产品的差异随季节的变化不显著。白天地表温度产品的差异在不同地表覆盖类型下具有显著的变化,而晚上地表温度产品的差异在不同地表覆盖类型下的变化不显著。
[Abstract]:Surface temperature (Land surface temperature, LST) is a very important physical quantity in the process of interaction between surface and atmosphere. It is of great significance to grasp the temporal and spatial distribution of surface temperature on regional and global scale, especially the comprehensive, complete and continuous information of temporal and spatial distribution of surface temperature, which is of great significance to the energy balance of terrestrial and atmospheric systems and the study of ecosystem. Satellite remote sensing is the best way to get surface temperature at regional and global scale. In this paper, the polar orbit satellite data and geostationary satellite data are used as data sources to study the inversion method of high spatial resolution all-weather surface temperature, and the MODIS downscaling method based on geo-weighted regression. And the cross-validation method of MSG-SEVIRI and Terra/Aqua-MODIS surface temperature products. The main conclusions are as follows: (1) the advantages of microwave inversion and thermal infrared inversion of surface temperature, An effective fusion model between cloud partial passive microwave surface temperature and cloud free part thermal infrared surface temperature in pixel is established, and a high spatial resolution all weather surface temperature inversion method is developed. Compared with the original MODIS surface temperature, the ground surface temperature changes continuously in space after fusion, which reflects the interannual variation of surface temperature in time. (2) the local regression coefficient of geographical spatial variation is used. A dynamic regression relationship between land surface temperature (LST) and normalized vegetation index (NDVI) and surface elevation was established, and the method of MODIS surface temperature downscaling based on geo-weighted regression was developed. Compared with the ASTER surface temperature products, the precision of the ground temperature downscaling method based on geographical weighted regression is better than that of the UniTrad method and the TsHARP method developed by the predecessors. The average absolute error and root mean square error of the GST downscaling method based on geographical weighted regression are about 2.3 K and 3.1K respectively. (3) by selecting SEVIRI and MODIS surface temperature products, the mean absolute error and root mean square error are about 2.3K and 3.1krespectively. The spatial and observational pixels are used as matching pixel pairs to analyze the differences between the two surface temperature products in different seasons time and cover types. The difference of SEVIRI and MODIS surface temperature products showed significant change with the season during the day, but the difference of the surface temperature products at night was not significant with the season. The variation of surface temperature products in daytime was significant under different surface cover types, but not in different surface cover types at night.
【学位授予单位】：中国农业科学院
【学位级别】：博士后
【学位授予年份】：2016
【分类号】：P423.7;P407

【相似文献】