Observations of lower-lvl moistureRH in NUCAPS vs. GFS...

Ok, as promised, here is a view of lower level moisture in NUCAPS data vs GFS in the earlier pass this afternoon (Image 1).  Again, reasonably good agreement exists between the data sets.  It's important to mention here that NUCAPS model data are independent of the GFS data.  The NUCAPS data set does employ a statistical first-guess field to initialize the retrieval. This first guess is a regression analysis that correlates the measured radiance with ECMWF reanalysis climatologies.  All of this means that, the final NUCAPS soundings and gridded data observed in AWIPS (for example) are independent of numerical weather prediction models.

So, let's take a quick look now at some lower level observations of RH from an earlier pass of available NUCAPS data (Image 1).

Image 1.  GFS20 850-700 MB RH (left) and NUCAPS gridded 850-700 MB RH (right) at approx. 18 UTC 8 May 2018.  Notice the general agreement between the GFS and NUCAPS data, but of course some differences do exist in the details.  

 One of the advantages of the NUCAPS data, besides the fact that they're model independent as previously described, is that each columnar retrieval (or sounding) is spatially independent.  That is, one pixel value is not based on the value of an adjacent value, or there is no smoothing of the data.  This can be important for picking out all-important gradients that may exist in the atmosphere.  Let's take a closer look at the FL peninsula and adjacent areas to see an example of this.

Image 2.  GFS20 850-700 MB RH (left) and NUCAPS gridded 850-700 MB RH (right) at approx. 18 UTC 8 May 2018, zoomed over FL and surrounding areas. 

So, the NUCAPS data do indicate generally drier conditions at this layer than the GFS data.  But, also observe the very dry plume of air in the SW part of the image.  Notice that the data are smoothed in the GFS.  Likewise, notice individual pixels of lower RH just to the east of the FL peninsula.  The NUCAPS data indicate that this small area of relatively low moisture was probably smoothed and interpolated in the GFS model.  Granted, this may be of limited importance for operational meteorology considerations.  Nevertheless, this is just one observation today.  Important boundaries and strong gradients may exist regularly and models will tend to smooth these out.  The NUCAPS data may help to build better refine conceptual models for forecasters and better define the locations of these boundaries.