Publications

Jiao, ZH; Mu, XH (2022). Global validation of clear-sky models for retrieving land-surface downward longwave radiation from MODIS data. REMOTE SENSING OF ENVIRONMENT, 271, 112903.

Abstract
Surface longwave radiation (SLR) plays an important role in the energy budget of the Earth's climate system. Remote sensing provides various data sources to retrieve SLR on a large scale and with high spatial resolution (e. g., 1 km). Multiple retrieval methods of surface downward longwave radiation (SDLR) based on satellite thermal infrared data produce different retrieval results for the same scenario. Therefore, validating these models is necessary to understand their characteristics and limitations. To this end, ground-based measurements were used to provide independent validation of six widely used SDLR models for clear sky conditions over 41 Baseline Surface Radiation Network (BSRN) stations worldwide. The Wang2020 model had the best overall performance (bias of similar to 5.480 W/m(2), root-mean-square errors [RMSE] of 23.226 W/m(2), R-2 of 0.879), and Tang2008 model had similar retrieval capability. The errors of LST had limited influence on the retrieval accuracy of SDLR models. When using the near-surface air temperature, the retrieval accuracy of the Zhou2007 model was significantly improved with a range of similar to 9.5 W/m(2) for the RMSE. The uncertainty of TCWV had significant effect on all model performances, wherein the Zhou2007 model had stronger error resilience of TCWV. Moreover, MODIS TIR TCWV data provided better performance than NIR TCWV in most situations, and thereby are preferred to use in the SDLR retrieval. Surface altitude had a lesser impact on SDLR retrieval than terrain effects. All models overestimated SDLR for peak stations in mountainous areas, with biases reaching 56.614 W/m(2) and RMSE reaching 63.909 W/m(2). Land cover type also had a significant effect on retrieval accuracy; model performances were poorer in the desert and barren where atmospheric conditions are extremely dry and hot. Remote sensing SDLR data with high accuracy are needed for hydrological, agricultural, and climate change applications. The results of this study provide a reference for the SDLR retrieval accuracy based on clear-sky models.

DOI:
10.1016/j.rse.2022.112903

ISSN:
1879-0704