On the Land Emissivity Assumption and Landsat-Derived Surface Urban Heat Islands: A Global Analysis

This is a Preprint and has not been peer reviewed. This is version 1 of this Preprint.


Download Preprint


TC Chakraborty , Xuhui Lee, Sofia Ermida, Wenfeng Zhan


The prescription of surface emissivity (ε) strongly controls satellite-derived es-timates of land surface temperature (LST). This is particularly important for studying surface urban heat islands (SUHI) since built-up and natural land-scapes are known to have distinct ε values. Given the small signal associated with the SUHI compared to LST, accurately prescribing urban and rural ε would improve our satellite-derived SUHI estimates. Here we test the sensitivity of SUHI to the ε assumption made while deriving Landsat LST for almost 10,000 global urban clusters for summer and winter days. We find that adjusting the ε values from the Advanced Spaceborne Thermal Emission and Reflection Ra-diometer (ASTER) dataset based on pixel-level normalized difference vegetation index (NDVI) improves our ability to capture the summer to winter contrast in SUHI. Overall, the difference between the two methods is moderate; around 10%during summer and around 20% during winter, though this difference varies by climate zone, showing higher deviations in polar and temperate climate. Using both methods of prescribing ε, we provide the first global estimates of SUHI derived from Landsat. During summer, the global mean SUHI varies between 2.31 °C (0.24 °C for winter) when using ASTER ε to 2.54 °C (0.29 °C for winter) after NDVI-adjustment. Compared to Moderate Resolution Imaging Spectrora-diometer (MODIS) Terra observations, Landsat data show higher SUHI daytime intensities during summer (by more than 1.5 °C), primarily due to its ability to better resolve urban pixels. We also find that the ε values prescribed for urban land cover in global and regional weather models are lower than the satellite-derived broadband ε values. Based on the sensitivities to ε we find for urban and rural LST, this would imply overestimation of SUHI by these models (by around 4 °C for both summer and winter), all else remaining constant. Our analysis provides a global perspective on the importance of better constraining urban ε for comparing satellite-derived and model-simulated urban heat islands. Since both the structural and geometric heterogeneity of urban areas controls the bulk ε, future studies should try to benchmark the suitability of existing LST-ε separation methods over urban areas.




Physical Sciences and Mathematics


Land Surface Temperature, Google Earth Engine, Urban Heat Island, Surface emissivity, Landsat, MODIS, Global


Published: 2021-05-07 16:48

Last Updated: 2021-05-07 23:48


CC BY Attribution 4.0 International

Add a Comment

You must log in to post a comment.


There are no comments or no comments have been made public for this article.