I disappointed by a NOAA presentation this morning regarding testing between the two global model finalists: the NOAA/GFDL FV3 and the NCAR MPAS. I will blog further about this, but a few major points:
1. NCAR has pulled out because they feel the testing is inappropriate, and I have to agree.
2. All test models had to use the old GFS (current model) physics which are completely inappropriate at high resolution. In fact, GFS physics doesn't work well at any resolution. Like testing new racing cars on a muddy road--you can't do it.
3. The future of global prediction is at convection-allowing resolution (4 km or less grid spacing). But these resolutions were hardly tested (48 out of the 50 tests were at 13 km grid spacing or more).
4. Some of the results were clearly bogus, like the radically poor results of a 13-km forecast run and a hurricane simulation that had rain in the eye of the MPAS hurricane). Something was clearly wrong with the tests.
5. The testing had no vision of testing a configuration that might be used operationally in ten years (e.g., convection allowing over the globe). It was all about testing a configuration nearly identical to the current GFS.
