Advantages of fine resolution SSTs for small ocean basins: Evaluation in the Black Sea

Abstract

This paper examines monthly variability of climatological mean sea surface temperature (SST) in the Black Sea. A total of eight products, including observation‐ and model‐based SST climatologies, are formed and compared with each other. Some of the observation‐based SST data sets include only satellite measurements, while others combine in situ temperatures, such as those from moored and drifter buoys, with satellite data. Climatologies for numerical weather prediction (NWP) model‐based data sets are formed using high temporal resolution (6 hourly) surface temperatures. Spatial resolution of these SST products varies greatly (≈ 4 km to 280 km), with the observation‐based climatologies typically finer than the NWP‐based climatologies. In the interior, all data sets are in general agreement, with annual mean SST biases typically within ±0.2°C in comparison to the finest resolution (4 km) satellite‐based Pathfinder climatology. Major differences are near the land‐sea boundaries where model‐based SSTs pose serious biases (as much as >5°C). Such errors are due to improper contamination of surface temperatures over land since coarse resolution model‐based products cannot distinguish land and sea near the coastal boundaries. A creeping sea‐fill interpolation improves accuracy of coastal SSTs from NWP climatologies, such as European Centre for Medium‐Range Weather Forecast. All climatologies are also evaluated against historical in situ SSTs during 1942–2007. These comparisons confirm the relatively better accuracy of the observation‐based climatologies.