不确定性约束下的AOD空间覆盖STRE建模优化

An Aerosol Optical Depth Spatial Coverage Improvement Method Based on Spatial-Temporal Random Effects with Uncertainty Constraint

针对现有气溶胶光学厚度(aerosol optical depth,AOD)空间覆盖优化研究中存在融合过程时空联合关联特征考虑不全面(仅考虑AOD产品间的时间关联,而忽略了AOD产品间的空间关联关系)、估算结果不确定性未加以约束等问题,提出了一种不确定性约束下的AOD空间覆盖时空随机效应(spatio-temporal random effects,STRE)建模优化方法,并以2013-01-2017-12中国大陆东部地区日均中分辨率成像光谱仪(Moderateresolution Imaging Spectroradiometer,MODIS)AOD产品空间覆盖优化为例对该方法的有效性进行评价。结果表明,通过综合顾及AOD数据时空关联特征,STRE建模优化方法可有效提升卫星遥感AOD产品的空间覆盖度和时间完整度;且不确定性约束下的STRE建模可进一步提升AOD空间覆盖优化结果的精度、降低原始卫星遥感AOD产品长时间和大范围连续缺失所造成的不确定性。

According to the existing researches of aerosol optical depth(AOD) spatial coverage improvement, the spatial-temporal correlation of multi-source products is not fully considered in the fusion process,and the uncertainty of the estimated results is not considered. This paper proposes an AOD spatial coverage improvement method based on spatio-temporal random effects(STRE) modeling with the uncertainty constraint. And this method is applied to the spatial coverage improvement of daily Moderate-resolution Imaging Spectroradiomete(MODIS) AOD products in eastern China from January 2013 to December2017. The results show that the STRE model AOD spatial coverage improvement method can effectively improve the spatial coverage and temporal completeness of AOD by considering the temporal and spatial correlation of AOD data. And the introduction of uncertainty analysis in the STRE model AOD spatial coverage improvement can effectively reduce the accuracy of the overall coverage improvement results, especially for the region with serious missing original AOD data on time and space.