Some observations and thoughts on NUCAPS Soundings and the Experimental Soundings...

New to the HWT this year is the addition of Experimental Suomi NPP NUCAPS soundings.  Data accuracy in the boundary layer of the NUCAPS soundings is generally lower than that in the mid and upper levels of the troposphere, since these are satellite-derived retrievals.  The Experimental version of the NUCAPS soundings is supposed to represent an improvement in these lower level conditions.  In the past, forecasters have typically modified the lower portions of the soundings manually by using a simple temperature and dew point adjustment based on nearby surface observations.  The Experimental data, however, automatically makes this adjustment for forecasters, by using a mixed layer growth model with Real-time Mesoscale Analysis (RTMA) data for surface winds and temperatures and GOES-16 data for skin temperatures (thanks for the explanation Jack Dostalek!).  Not only is this method more efficient for forecasters, but it represents a more robust method for sounding modification based on an increased number of observations.  Below are a couple of soundings from this afternoon in western Kansas, near the town of Syracuse.  The first sounding represents the regular NUCAPS sounding, while the second image is the experimental or modified NUCAPS sounding.

Image 1.  Soumi NPP NUCAPS Sounding near Syracuse, KS, ~19Z 21 June 2017

    

Image 2.  Experimental Soumi NPP NUCAPS Sounding near Syracuse, KS, ~19Z 21 June 2017

The Experimental sounding included here doesn't necessarily represent a drastic change to the original sounding.  But, you can clearly see the removal of the rather large superadiabatic layer in the lowest layer of the sounding (up to nearly 700mb), and its replacement with a dry adiabatic layer in the experimental sounding.  There have been a number of potential improvements noted in the soundings during the HWT so far, and forecasters are happy not to have to make these modifications themselves, although there is a bit of a delay with the Experimental soundings, amounting to another hour or two past the receipt of the original NUCAPS sounding data.  Sometimes though, the amount of "improvement" in the soundings is not always clear.

This brought up some discussion between myself and the other NUCAPS sounding representative at the HWT.  In the past, apparently forecasters have desired to make modifications to the surface data in the soundings and also to the boundary layer simply by using some nearby surface observations.  It should be understood by researchers that this is fairly normal practice...to take a forecast sounding and modify it with a more recent and perhaps accurate surface observation.  However, I think the inclusion of model data would perhaps be better in some cases.  If the model initialization/analysis or <6 hour forecast grid is ultimately the best representation of the state of the atmosphere at the time, over a simple point observation, then the model data should be utilized or at least considered.  It should probably be pointed out to forecasters that the NUCAPS soundings are volumetric soundings, not necessarily point data as represented by a point sfc ob or perhaps even RTMA data.  So, forecast data which also represent volume information may ultimately be a good choice for sounding modification in some or maybe even most cases.  Just some thoughts here.

Kris W