부산대학교 도서관

An analysis of three operational satellite-based thermal-infrared land surface temperature (LST) products is presented for conditions of heavy dust aerosol loading. The LST products are compared against ERA5’s skin temperature (SKT) across the Sahara Desert and Sahel region, where high concentrations of dust aerosols are prevalent. Large anomalous differences are found between satellite LST and ERA5’s SKT during the periods of highest dust activity, and satellite–ERA5 differences are shown to be strongly related to dust aerosol optical depth (DuAOD) at 550 nm, indicating an underestimation of LST in conditions of heavy dust aerosol loading. In situ measurements from two ground stations in the Sahel region provide additional evidence of this underestimation, showing increased biases of satellite LST with DuAOD, and no significant dependence of ERA5’s SKT biases on dust aerosol concentrations. The impact of atmospheric water vapor content on LST and SKT is also examined, but dust aerosols are shown to be the primary driver of the inaccurate LSTs observed. Based on comparisons with in situ data, we estimate an aerosol-induced underestimation of LST of approximately 0.9 K for every 0.1 increase in DuAOD. Analysis of brightness temperatures (BTs) in the thermal infrared atmospheric window reveals that dust aerosols have the opposite effect on BT differences compared to water vapor, leading to an underestimation of atmospheric correction by the LST retrieval algorithms. This article highlights a shortcoming of current operational LST retrieval algorithms that must be addressed.