Pacific Sampling of Shortwaves

[am19psu]

There has been some recent back and forth in the NYC subforum regarding the sampling or lack of sampling of important shortwave features over the Pacific. On this board, it's long been argued that the models won't get a clue until the responsible shortwaves have been properly sampled. This is somewhat true, but how much does it matter? Also, does the sampling issue rear its head during the off hour GFS/NAM runs, as some people also claim?

I know the answers to these questions notionally, but there are others (*cough* dtk *cough*) who can answer them with real science. Let's clear up this old wives' tale once and for all.

[dtk]

There has been some recent back and forth in the NYC subforum regarding the sampling or lack of sampling of important shortwave features over the Pacific. On this board, it's long been argued that the models won't get a clue until the responsible shortwaves have been properly sampled. This is somewhat true, but how much does it matter? Also, does the sampling issue rear its head during the off hour GFS/NAM runs, as some people also claim?

I know the answers to these questions notionally, but there are others (*cough* dtk *cough*) who can answer them with real science. Let's clear up this old wives' tale once and for all.

Ha. No pressure or anything. I probably won't have time to put together a detailed, thoughtful reply to this until later in the week.

I'll add one thing to your comments though....even if we had all of these shortwave features sampled perfectly with perfect observations, the models could still very well screw up a forecast (either through imperfect assimilation or through model error).

I started to discuss the issues relevant to 00/12z versus 06/18z in another thread. In summary, I suspect the NAM is much more suspect to issues during the so-called off hour runs (due to the partial cycling procedure and less reliance on satellite data).

[baroclinic_instability]

I think it could be argued that the recon flights can make a huge difference in the moist Pacific during winter. Targeted sampling would be nice since errors grow fast in such an environment. I am interested in what dtk has to say.

[Tropopause_Fold]

want to bump this thread. 

[ohleary]

The latest studies show little improvement in forecast accuracy with WSR obs. This may be due to many things...much better satellite coverage and assimilation...improved assimilation period or better modeling skill in general.

[dtk]

The latest studies show little improvement in forecast accuracy with WSR obs. This may be due to many things...much better satellite coverage and assimilation...improved assimilation period or better modeling skill in general.

I assume you are referring to this:

http://journals.ametsoc.org/doi/abs/10.1175/MWR-D-12-00309.1

 

However, that is only a single data denial study, performed with a particular modeling system (I'm not at all surprised to see that they had little impact on the ECMWF skill).  I know that there were some in-house experiments run with the GDAS/GFS, but those did not even perform proper DA cycling so the results are nearly impossible to interpret (at best) or perhaps meaningless altogether (at worst).

 

Despite the above caveats, my hunch is that these observations no longer have the impact that they once did (for some of the reasons you mentioned, in addition to the fact that they simply do not cover a large enough spatial extent).