Assessment and inter-comparison of five high-resolution sea surface temperature products in the shelf and coastal seas around China

Recently the global ocean data assimilation experiment (GODAE) High-Resolution SST Pilot Project (GHRSST-PP) has delivered several sea surface temperature (SST) products based on satellite remote-sensing observations. These cloud-free, high spatial (up to 2–5 km) and temporal resolution (daily or 6-hourly) SST products provide important data sources for operational ocean forecast of both global and shelf/coastal seas. Before applying the GHRSST products to initialization or data assimilation for ocean forecasting of shelf and coastal seas, it is necessary to make inter-comparison and assessment of these products. In this paper, five products were investigated by means of inter-comparison and comparison to in situ observations at the shelf and coastal sea around China and in the northwest Pacific over 1-year period (from October 2005 to September 2006). The inter-comparison shows that in the coastal and the shelf sea around China, there are some significant differences among these products. Some have warm bias with reference to the five-member mean, while some have cold biases. In shelf seas around China, the spread of the five products is between 0.4 and 0.6 °C. In the coastal sea (shallower than 40 m), the spread increases dramatically to 0.95 °C. Further, the comparison of daily GHRSST products with in situ night-time SST observations from surface drifters and ship reports indicated that most of these SST products are excellent in deep water region. The root mean square differences (RMSDs) of some products with reference to in situ observations are less than 0.4 °C. But near the coastal, relatively large bias and RMSD exist in all five products. In order to reduce the bias and RMSD in near coast, a simple, depth-dependent bias correction scheme is applied to two of the five products. After correction their biases can be reduced, and their RMSDs could be also decreased about 6–14%. This study suggests that the GHRSST products can be used for operational ocean forecasting with proper error estimation for shelf seas and the open ocean. For the coastal seas where the depths are shallower than 40 m, large uncertainties exist in these products. Improvements are needed for future application in the coastal seas.