Wednesday, January 18, 2017

GEFS Plumes Technical Details

Pat Holbrook asked: Bob...what's the lat/lon/elevation for the GFS point that the EMC GFS is using?

Pat

--------------------------------

Good question and like most things NWS, very hard to find a precise answer online. The statement below is at the plumes site:


Click on text box for more readable font size.

The QPF plumes (above from 06 UTC last night) are interpolated from 0.5 degree grid point forecasts to the airport location. However, this could be done using just the four closest grid points, or for more nearby points using a distance-weighting function of some sort. I could not find an answer regarding the exact procedure employed.


Based on the statement above about the GEFS procedures, I assume that both of these plume products (above - total snow accumulation and below - probability of snow) are derived from the forecast for the nearest 0.5 degree grid point (i.e., the crux of Pat's question), which could be located anywhere within about a 25 km radius of the airport. No big deal for the flatlands but very significant here. Where is the nearest grid point wrt TUS? Very important question and, again, I could find no precise answer online.

So, if any reader knows the answer, please let us know.


1 comment:

  1. There's another potentially important issue at play, and that is whether or not these are plumes from BUFR data or a GFS grid and, if the latter, which grid.

    The GFS precipitation from the high-resolution "flux" grid is simply an abomination in complex terrain. For example, we see gross overprediction of mountain precipitation from that grid on a regular basis. One can see this at http://weather.utah.edu/index.php?t=gfssflux&r=IM&d=SF. Just look for the big, persistent precipitation maxima over high terrain. Yes, such maxima are observed climatologically, but they are far stronger and persistent than they should be. Depending on location, some very large biases might be observed if these plumes are coming from the BUFR data or flux grid. Smoothed data is better.

    ReplyDelete