Tuesday, July 26, 2011

How Moist Is It?

We've already discussed some of the differences in precipitable water measured by soundings and ground-based GPS, but today provides a particularly aggravating example.

A two day loop of IR satellite imagery, Dynamic Tropopause pressure, and precipitable water shows quite nicely the monsoon surge that brought us showers last night, but also the advection of drier air from the west early this morning.

Two day loop (1400 UTC 24 Jul – 1400 UTC 26 Jul 2011) of
dynamic tropopause pressure (shaded), precipitable
water (contours), 925-mb wind, and IR imagery.
These changes in moisture are important for assessing the potential for convection today.  Knowing the importance of looking at real data, I decided to see what the ground-based GPS shows.  Consistent with the loop above, it shows a pronounced peak in precipitable water at over 3 cm just after 0000 UTC yesterday afternoon, followed by a gradual decrease to about 2.6 cm today.
The NWS sounding, however, gives an entirely different impression.  In particular, note the massive dry bias in the 0000 UTC sounding yesterday afternoon (I am assuming the GPS is closer to ground truth).

Comparing yesterdays 0000 UTC sounding with this mornings 1200 UTC sounding shows substantial moistening of the lower troposphere.  

0000 UTC (purple) and 1200 UTC (red, green) 26 Jul 2011
soundings from KSLC.
In fact, from 0000–1200 UTC the measured precipitable water increases  from 2.31 to 2.75 cm.  Although it is unreasonable to expect the 12-h soundings to pick up on the peak after 0000 UTC, the dry bias at 0000 UTC gives a false impression that the PW has increased during this 12-h period, whereas it has actually declined slightly (see GPS meteogram above).

In 2007, Bob Maddox discussed some of the problems with the RRS Sippican Sondes used by the National Weather Service at that time.   I don't know if these sondes are still being used, but it certainly appears that problems persist, with significant implications for weather forecasting and long-term climate records.  

Some things are a shame, but the decline in the quality of our upper-air observing system is a damn shame.   


  1. I think the sondes you mention are the same ones being used now...or at minimum, a slightly updated version. And yes, the RH sensors are terrible (to use a polite word).

  2. Indeed, the Sippican sondes are now in use at almost all of the conus NWS upper-air sites. The data from these instruments exhibit, at times, a plethora of various problems. The first change to be made by the NWS will likely be a switch to a new RH sensor, sometime during the coming year or so. Eventually, the entire sonde will be quietly scrapped for a replacement instrument!

  3. Good Riddance!

    We had some action here today, with urban flooding. Thus, the erroneous trend inferred from the sonds may have proved helpful :-).