Geostationary Satellite-measured Land Surface Temperature Measurement Statistical Comparisons with in situ Soil Temperature Observations for Land Data Assimilation

This investigation focuses on evaluating geostationary satellite-based land surface temperature (LST) observations from the International Satellite Cloud Climatology Project (ISCCP), comparing the measurements to in situ soil temperature measurements obtained from the International Soil Moisture Network (ISMN). The geostationary satellite-based LST measurements provide frequent assessments of LST during cloud-free periods, providing multiple measurements of the same location over a single day and capturing the diurnal cycle. Because of this, ISCCP is of potentially great value for land data assimilation systems. Accordingly, it’s important to understand and quantify the satellite LST bias and error. This analysis compares the geostationary LST measurements against in situ soil temperature measurements spread across the United States. Specifically, we compare the top layer soil temperature measurements from US sites within the ISMN archive to determine the error characteristics for assimilating the LST measurements into the Noah and Noah-MP land surface models. We partition the LST and ISMN statistical comparisons by land use and time of day, computing results to evaluate diurnal changes in correlation and bias. The results show that the geostationary LST measurements correlate to the 2-inch (5-cm) depth soil temperature observations across multiple land use classes. Still, the bias switches from negative at night to strongly positive during the day. Further, the LST measurements also strongly correlate to deeper soil temperature observations. The strongly correlated data, but with a shifting diurnal bias pattern, points to needing an assimilation approach that accounts for day and night statistical differences.