Evaluations of land-ocean skin temperatures of the ISCCP satellite Retrievals and the NCEP and ERA reanalyses

This study evaluates the skin temperature (ST) datasets of the International Satellite Cloud Climatology Project (ISCCP) D satellite product, the ISCCP FD satellite product, the 40-yr ECMWF Re-Analysis (ERA-40), the NCEP-NCAR Reanalysis, and the NCEP-Department of Energy (DOE) Atmospheric Model Intercomparison Project (AMIP)-II Reanalysis. The monthly anomalies of all the datasets are correlated to each other and to most of the ground-truth stations with correlation coefficients > 0.50. To evaluate their qualities, the 5 ST datasets are used to calculate clear-sky (CS) outgoing longwave radiation (OLR) and upward surface longwave radiation (USLR); the results are compared with the Earth Radiation Budget Experiment (ERBE) satellite observation and 14 surface stations. The satellite-derived STs and ERA- 40 ST tend to bias high on hot deserts (e.g., Sahara Desert), and the reanalyzed STs tend to bias low in mountain areas ( e. g., Tibet). In Northern Hemisphere high-latitude regions ( tundra, wetlands, deciduous needle-leaf forests, and sea ice), the CS OLR anomalies calculated using the satellite-derived STs have higher correlations and lower root-mean-squared errors with the ERBE satellite observation than those derived from using the reanalyzed STs. ERA- 40 underestimates the amplitude of the seasonal ST over glaciers. All the reanalysis products (ERA-40, NCEP-NCAR, and NCEP-DOE AMIP-II) overestimate the ST during partial sea ice-covered periods in the middle-high-latitude oceans. Nonetheless, suspected spurious noises with an amplitude of 2 K in the satellite- derived STs produce a physically unviable anomaly over earth's surface where the amplitude of the anomaly is weak (such as open-water bodies, croplands, rain forest, grasslands, hot deserts, and cold deserts). Better land - ocean - ice schemes for a reanalysis should be developed for desert regions, high plateaus, fractional sea ice - covered oceans, and seasonally snow-covered lands, where the largest ST errors are identified.