Submitted to: Journal of Geophysical Research Atmospheres
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 12/10/2012
Publication Date: 3/1/2012
Citation: Holmes, T.R., Crow, W.T., Yilmaz, M.T., Jackson, T.J. 2012. Enhancing model-based land surface temperature estimates using multi-platform microwave remote sensing products. Journal of Geophysical Research Atmospheres. 11:577-591.
Interpretive Summary: Land surface temperature is defined as the temperature of individual surface (or near-surface) elements (e.g., soil, vegetation or vegetation residue) that are in energetic contact with the overlying atmosphere. It plays an important role in land surface processes and is a key input into physically-based, satellite retrieval algorithms for a number of important land surface variables (e.g., soil moisture and evapotranspiration). However, large systematic differences exist between surface temperatures as observed or modelled by different methods. This study presents a framework to reconcile various independent estimates of land surface temperature. After implementing this method on five independent temperature records (from both models and satellites), it is shown that the information contained in each record can be combined into a single improved data product, with a 10% reduction in random error. The results of this study give detailed new insights into the relative properties of different temperature data sets from numerical models and retrievals from satellite sensors. It presents a methodology that will be implemented to improve a decade-long record of surface temperature, which, in turn, will benefit scientists that rely on such records to study anomalies in the global hydrologic cycle.
Technical Abstract: Land surface temperature plays an important role in land surface processes, and it is a key input to physically-based retrieval algorithms of important hydrological states and fluxes, such as soil moisture and evaporation. This study presents a framework to use independent estimates of land surface temperature from four satellite sensors to improve the sub-daily accuracy of land surface temperature output from a numerical weather prediction system. First, structural differences in timing and amplitude of the temperature signal are addressed. Then, the satellite observations are used to update an auto-regressive model of the offset in the numerical weather prediction output, treating the offsets in daily mean and amplitude separate. This study gives detailed new insights in the relative properties of different temperature data sets from numerical based estimates and retrievals from passive microwave sensors. It is shown that the satellite observations may be used to reduce random errors in surface temperature estimations by 10%. Together, the pre-processing and assimilation presents a framework for merging satellite temperature observations with model surface temperature output.