I have a very general question. There are quite many ways to force the WRF model, as some free data is available at http://www2.mmm.ucar.edu/wrf/users/down
I have the application of a forecast sometimes in the past, yet I want to improve the forecast in a case study using some ensembles. What forcing data (boundary and input values) shall I use?
I guess I have the oppertunity to use a Forecast, an Analysis, a Reanalysis or a Reforecast. In my opinion when I use an Analysis I kind of "cheate", because in the lateral boundary conditions (LBC) observations which lie in the future are already incorporated, so the erros in the LBC are unrealistically to small. But everybody is doing so. Same thing with the Reanalysis. Why? So I have to use a forecast/reforecast, right?
Thanks a lot for any advise!