Problems with forward models and retrieval algorithms (in view of feeding the simulator work)
It is useful to structure this short discussion in terms of time scales., with reference to the objectives :
I Short term
Detailed assessment must wait unil the actual content of the projected simulator is well known.
Certainly, many things are still lacking. Some of them (e.g. sea state effect) are being investigated. However in terms of forward modelling but especially of retrieval, a major need concerns the quality and statistical structure of ancillary data likely to be available. This is especially true over the ocean, where averaging the data should allow to reduce the random uncertainties down to the required levels. In addition, such informations are necessary to assess the possibility and performance of external calibration using natural targets.
These issues begin to be considered (e.g. ESA 's ITT for salinity retrieval).
2 Longer term
TO topic about ancillary data is raised again. Fro some of them maps are necessary. The possibility to make use of previous work (e.g. in the frame of the HYDROSTAR project) deserves some investigation.
In the retrieval analysis, one should not wait too much before
adressing the issue of varying pixel sizes.
Back to main page