Combining model and geostationary satellite data to reconstruct hourly SST field over the Mediterranean Sea

Abstract

This work focuses on the reconstruction of Sea Surface Temperature (SST) diurnal cycle through combination of numerical model analyses and geostationary satellite measurements. The approach takes advantage of geosta- tionary satellite observations as the diurnal signal source to produce gap-free optimally interpolated (OI) hourly SST fields using model analyses as first-guess. The resulting SST anomaly field (satellite-model) is free, or nearly free, of any diurnal cycle, thus allowing one to interpolate SST anomalies using satellite data acquired at different times of the day. The method is applied to reconstruct the hourly Mediterranean SST field during summer 2011 using SEVIRI data and Mediterranean Forecasting System analyses. A synthetic cloud reconstruction experiment demonstrated that the OI SST method is able to reconstruct an unbiased SST field with a RMS = 0.16 °C with respect to SEVIRI observations. The OI interpolation estimate, the model first guess and the SEVIRI data are evaluated using drifter and mooring measurements. Special attention is devoted to the analysis of diurnal warming (DW) events that are particularly frequent in the Mediterranean Sea. The model reproduces quite well the Mediterranean SST diurnal cycle, except for the DW events. Due to the thickness of the model surface layer, the amplitude of the model diurnal cycle is often less intense than the corresponding SEVIRI and drifter observations. The Diurnal OI SST (DOISST) field, resulting from the blending of model and SEVIRI data via optimal interpolation, reproduces well the diurnal cycle including extreme DW events. The evaluation of DOISST products against drifter measure- ments results in a mean bias of −0.07 °C and a RMS of 0.56 °C over interpolated pixels. These values are very close to the corresponding statistical parameters estimated from SEVIRI data (bias = −0.16 °C, RMS = 0.47 °C). Results also confirm that part of the mean bias between temperature measured by moorings at 1 m depth and the satellite observations can be ascribed to the different nature of the measurements (bulk versus skin).