Reconciling remote sensing and reanalysis land surface temperatures: How surface conditions shape bias between GOES-16 and MERRA-2 across the contiguous US

Dhruva Kathuria , Alexandra G Konings, Jana Kolassa, Yanlan Liu, Meng Zhao, Alexey N Shiklomanov

Land surface temperature is a key variable governing land–atmosphere energy and water exchanges. Despite its importance, satellite observations and reanalysis products often differ in how they define the effective depth of land surface temperature and in the assumptions underlying their estimates, making comparisons and interpretation challenging. In this study, we present a detailed comparison of land surface temperature from GOES-16 (satellite) and MERRA-2 (reanalysis) across the contiguous United States for 2022 and 2023. The results reveal systematic diurnal and seasonal biases: GOES-16 tends to be warmer than MERRA-2 in the afternoon and at night, but cooler in the morning. The magnitude of these biases varies by season. At night, GOES-16 is warmest relative to MERRA-2 for forests; in the morning, it is coolest for croplands and grasslands; and in the afternoon, it is warmest for barren and shrublands. Within individual land cover types, variability in surface conditions—such as soil moisture and elevation—modulates the bias at night and in the morning, with GOES-16 LST being warmer at night and cooler in the morning for wetter soils and at higher elevations. Our analysis also indicates that Leaf Area Index plays a role in bias patterns during spring and autumn, likely due to the association of temperature with leaf emergence and senescence. These findings provide new insights into the mechanisms underlying land surface temperature bias patterns and highlight the importance of accounting for surface condition variability in bias correction and data assimilation workflows.