江苏地区MODISLST产品重建研究  

Research on the Reconstructing of MODIS LST Product of Jiangsu Province

作　　者：严婧 沈润平[1,2] 鲍艳松[3] 李鑫川[3] 

机构地区：[1]南京信息工程大学气象灾害省部共建教育部重点实验室 [2]南京信息工程大学遥感学院 [3]南京信息工程大学大气物理学院,江苏南京210044

出　　处：《环境科学与技术》2014年第1期160-167,共8页Environmental Science & Technology

基　　金：国家重点基础研究发展计划“973”计划(2010CB950700;2005CB121108-6)

摘　　要：地表温度(LST)是评价地表热环境的重要指标,但受云等大气状况影响,MODIS LST时间序列产品存在大量噪音像元,严重影响LST数据使用。以江苏省为研究区,用2003-2011年MYD11A1 LST日时间序列产品为基础,结合质量控制信息(QC,quality control),基于改进时空滤波(mTSF,modified Temporal Spatial Filter)方法和最小二乘滑动滤波(S-G,Savitzky-Golay)方法逐步判断,建立LST背景数据库,综合考虑时空尺度效应,提出基于背景数据和mTSF原理的温度重建方法,重建了2009-2011年LST日产品,并校正云覆盖下重建结果。结果表明,江苏地区LST日产品全年受云污染像元(QC=2)比例较高,重建后的LST和0 cm实测地表温度平均相关系数为0.87,同时重建前后平均绝对误差变小。说明重建技术重构LST日产品保证了时空尺度的连续性,提高了原始LST数据的质量和使用效率。Land surface temperature （LST） is an important parameter for evaluating urban heat environment. MODIS LST time-series products have a lot of void value pixels that badly damage the use of data because of cloud and other atmospheric conditions. MYDllA1 daily LST time-series products from 2003 to 2011 in Jiangsu Province were used and modified Temporal Spatial Filter （mTSF） and Savitzky-Golay （S-G） methods were introduced to gradually establish a LST background database together with quality control information （QC）. Then considering temporal and spatial scale, daily LST were reconstructed from 2009 to 2011 based on background database and mTSF principle. Finally, LST under cloud cover were calibrated. Results showed that the ratio of the pixels which were affected by cloud （QC=2） was the highest that ranks the top of the year. The average correlation between reconstructed LST and 0cm land surface measured temperature was 0.87 and mean absolute error decreased after reconstruction, which revealed that reconstructed LST based on the methods ensure the continuity in temporal and spatial scale and improve the quality and efficiency of the original LST.

关 键 词：MODIS LST 时间序列 mTSF Savitzky-Golay 

分 类 号：X827[环境科学与工程—环境工程]