Present: Steve English, Catherine Prigent, Stu Newman, Nick Nalli, Lise Kilic, Susanne Crewell, Thomas Meissner, Magdalena Anguelova, Carlos Jiminez, Christophe Accadia, James Hocking, Jacob Hoyer, Ad Stoffelen 

1. Infrared assessment of PARMIO (Stu Newman, Nick Nalli)

   Presentations were given by Stu Newman (Met Office) and Nick Nalli (NOAA). Stu showed realistic simulations from PARMIO for comparison against the existing RTTOV IR emissivity model (ISEM), when making similar assumptions, and against other sources. Some difference were found, and it was thought possibly the treatment of double reflections in IREMIS and PARMIO is different. This needs to be checked (Action: Catherine and Steve to check with Emmanuel and Jacqueline). However Thomas, Steve and Catherine noted you would expect this to be treated in a GO model, and Stu confirmed from checking the code that wave shadowing effects definitely are included. 

   There was a discussion on improving the physical consistency of the PARMIO calculations in the IR and MW so that, ideally, they are physically consistent, or if not the reasons for bespoke treatments for different spectral bands are documented and justified. It was noted that the foam model probably could be applied to IR frequencies, though the treatment of scattering may need to change. The permittivity model has the same optical formulation, but different data (e.g. to different resonant frequencies) as input. It was noted it was worth investigating use of the permittivity “rescued” by Nick Nalli in Stu Newman’s studies. There was willingness to try and explore further harmonisation from MW to IR, but on a best endeavours basis so possibly not before the meeting in October. 

   It was noted that polarisation is generally ignored in the IR, and whilst in principle this can be modelled as it is in the microwave, at present this does not appear to be justified. 

2. SURFEM-Ocean (Lise Kilic)

   The fast ANN formulation to replicate PARMIO results for microwave frequencies was presented. The ANN is doing a very good job for direct calculations. Lise showed a range of validation that showed SURFEM-Ocean is a highly competitive ocean emissivity model. 

   The Matlab code now has analytical Jacobians so coding the full TL/Adjoint should be straightforward. The Met Office (James Hocking and Emma Turner) will convert the Matlab code to Fortran and ensure TL and adjoint code (as well as Jacobian code) is working. It is expected this code will be available in August and will be included in the RTTOV release later in 2022 (late October, or early November). The ISSI team will support the Met Office / NWPSAF effort, helping with interpretation of results to ensure the RTTOV code set is working.  

   The surface-atmosphere reflectance correction term is still taken from Fastem in SURFEM-Ocean as PARMIO can’t handle the full atmosphere, so it is difficult to model this from PARMIO. However the objective remains in a later version to enable this in PARMIO so this element of the ANN can also be taken straight from PARMIO data, losing this one handover from Fastem. 

3. October meeting plan

   At present only Stu, Maggie, Catherine and Steve from those present are confirmed for the 18-19 October meeting in Bern, but it is hoped others will make it. The suggested agenda for the October meeting is:

   • Reminder of goals of team (Steve and Catherine) 
   • Summary of what the team achieved – 30 minute presentations from Emmanuel, Maggie, Thomas and Lise. 
   • Plan for documentation of code (Emmanuel to lead) 
   • Plan for publication of model and model application/validation (All, round table) 
   • Plan for future maintenance of PARMIO code (Emmanuel and Ben to lead for PARMIO, Lise and James for SURFEM-Ocean) 
   • Post ISSI plan: discussion of prospects and aspirations for 
   • Day-2 for MW and IR 
   • Extension to Visible 
   • Application to profiling radar (precip, cloud, oxygen) 
   • Application to surface radar (scatterometer, SAR) 
   • AOB 

4. AOB   

(Post meeting as Ad Stoffelen needed to leave before the end: KNMI are working with the CMA WindRad data which is both C and Ku and both VV and HH. It turns out that the polarization ratio HH/VV of Ku behaves as expected, but for C it does not. The Ku polarization ratio now appears close to the C one, which is physically plausible. In the next months work will be done to quantify this using the WindRad data.)

.