SURFEM-ocean microwave surface emissivity evaluated

SURFEM-ocean microwave surface emissivity evaluated
Technical memorandum
Date Published
Secondary Title
ECMWF Technical Memoranda
Niels Bormann
Stephen English
Abstract The assimilation of satellite microwave radiance observations for weather forecasting requires a model for the ocean surface emissivity. At the European Centre for Medium-range Weather Forecasts (ECMWF) the Fast Emissivity model (FASTEM) has been used for many years. A new surface emissivity model, SURFEM-ocean, is available with Radiative Transfer for TOVS (RTTOV) version 13.2 and is tested here in the context of weather forecasting. Simulations from the Integrated Forecasting System (IFS) are compared to observations. Compared to FASTEM-6, SURFEM-ocean generates better fits to observations at low frequencies (6 – 10 GHz) and in strong wind speed conditions.
However, there are worse fits to observations and larger biases in mid and high frequencies (e.g. 19–166 GHz) especially around nadir and in horizontally polarised channels at higher zenith angles (e.g. 53◦). Although the permittivity model for sea water has changed between FASTEM-6 and SURFEMocean, this is not a main cause of the changed surface emissivity. Instead the main explanation must be the new representation of the wind-speed dependence in SURFEM-ocean. Background departure standard deviations become fractionally worse for sensors like Advanced Microwave Sounding Unit A (AMSU-A), even in tropospheric temperature-sounding channels. However, the overall impact of moving to SURFEM-ocean is to improve weather forecasts, particularly at short range in the southern ocean. This is partly because the channels and situations that are most affected by the increase in bias, for example 90 GHz horizontally polarised channels, are not actively assimilated. A filtering effect of the data assimilation system also likely means that the effect of better performance of SURFEM-ocean in high wind speed situations at low frequencies outweighs the worse performance in other situations. However, a side effect of the changing surface emissivity biases at 37 GHz is to increase the assigned observation error for microwave imagers, and this also contributes to the apparently improved forecast scores. From a future perspective, SURFEM-ocean provides a built-in handling of all four Stokes parameters, making it able to support future full-polarimetric microwave missions such as Weather System Follow-on – Microwave (WSF-M) and Copernicus Imaging Microwave Radiometer (CIMR). Also it supports sub-millimetre applications, whereas FASTEM-6 was not valid beyond 200 GHz, and the clear improvements at 6 GHz and 10 GHz support future use of these channels for inferring sea surface temperature. Hence the decision is clearly in favour of activating SURFEM-ocean in the next operational cycle instead of FASTEM-6.
DOI 10.21957/0af49d82e2