Surface emissivity working group, Bredbeck workshop Current status in ARTS - General solution for full Stokes vecttor with hemispherical integration of reflected term, but not actual physics to predict reflectivities. What is available - Fastem-2 in RTTOV7. This calculate V, H given windspeed and skin temperature for ocean. There is a land model but this is entirely empirical. It can be fed with coefficients based on the Prigent atlases, or if surface type is known from the list from Hewison. Note Fastem-2 already allows for the effect of full integration and should be fed with the specular angle only (i.e. specular angle, windspeed and skin temperature). - Fastem-3 in RTTOV8 makes an ad hoc correction to the ocean model emission term based on the model of Poe as modified by Weng to allow for azimuthal variations in all 4 Stokes vector elements. Fastem-2 should be feed with wind vector rather than windspeed. - Weng's ice and snow emissivity model. - IOMASA snow and sea ice emissivity model. - Matzler's snow emissivity model. Recommendations to ARTS model developers - Incorporate Fastem3 for ocean. This should be easy. The Fastem land code is probably of less value. - Adapt the ARTS input interface to accept wind vector and skin temperature and use these with Fastem-3 as an alternative to using an input emissivity. However input emissivity should still be available for other surfaces. - Consider replacing Fastem-3 with Geometric optics model for the ocean, which would exploit the general solution already in place. However the gains in accuracy doing this will be small, and it may not be necessary for ARTS. - Explore using AAPP or Weng method or other sources of surface estimate to drive empirical sea ice or snow emissivity models to provide more realistic emissivity, but at present do all this outside of ARTS as no method is mature enough to justify inclusion within ARTS - best to keep interface just as emissivity for these surfaces at present. - As emissivity schemes with geophysical inputs for these surfaces mature consider adding more geophysical inputs to the interface, but only where it is clear they are readily available (e.g. from an NWP model). Note Keep in mind there is more than one definition of the Stokes vector! Written by Steve English.