Usefulness of High Resolution Weather Soundings?

[T Mestler]

Hi All,

There are only 91 weather balloon launch locations in the USA, and those balloons are only launched twice a day. Naively I would think more data would help increase the accuracy and precision of severe storm weather forecasts and perhaps push out the prediction window by a few days, and I'm curious as to why there isn't more coverage.

Satellites help fill in the coverage gaps, but this data has a certain amount of error as it is not acquired in situ. In situ sensors aboard airlines also help fill the gaps, but these are either done at constant altitude OR only near airports.

So I guess my questions are:

- Is there a need for high spatial (every 100 miles) and temporal resolution (every hour) in situ soundings (wind, temp, humidity, pressure) or has the point of diminishing returns already been reached?
- Would it help make severe storm forecasting more precise/accurate?
- Are there other needs for the data outside of storm chasing?

Thanks!

Troy

 

[Willoughby Owen]

The newer satellites these days are very well adapted to provide up to the hour sounding values. But for an accurate display, then yes -- an observed sonde is needed.
JMA's new Himawari satellite will be able to provide the Japanese domain soundings every 2.5 mins. The technology is amazing now at our disposal!

 

[rdale]

Studies showed in winter that one extra sounding site provided dramatic improvements in Lake Effect Snow. There was a proposal to have a "floating" sector where sites in the area of expected rough weather would automatically run 6/18Z soundings but I'm sure budgetary issues kept that from coming alive.

 

[Brian G]

Of some relevance to the thread here is that I've seen a couple of different data denial experiments that show for the short range models (RAP, HRRR, etc.) that AMDAR data from commercial aircraft bests RAOB soundings in many categories and often by significant margins. Linked below is one such study available on the AMDAR website which shows that for the midwest if you take away the aircraft data you are removing one of the best (if not the best) source of data. It certainly fills in the gaps for locations (like St. Louis) which do not do RAOB launches but have busy airports. The other nice thing about this dataset is that it is maximized during the part of the day that matters the most for severe weather. Apparently it's hard to get people on those red-eye flights

http://amdar.noaa.gov/docs/Benj-WMO-Impact-Sedona-12May2012.ppt

Now if only the general public had access to the aircraft soundings!

 

[Paul Knightley]

The simple answer is that more obs data is always beneficial. Sounding data would be useful but if you want to improve model performance you need more sounding data upstream, i.e. out over the vastness of the Pacific! So, yes, it would improve performance and is always useful, but there's never going to be enough data per se!

 

[Jeff Duda]

From a scientific standpoint, you can NEVER have too much data. The only issue is whether adding a s---ton of sounding sites is cost effective. The current network across the CONUS is fine for synoptic scale observations, but many times the network misses mesoscale events like MCVs because of poor resolution. I don't think we need to get down to soundings every 10 km or anything, but why can't every WFO launch a balloon? I think that would be a major improvement.

Regarding temporal resolution, every 6 hours is certainly better than every 12 hours, but of course, the more launches you add, the more cost. I don't think NCEP needs to run the GFS every 6 hours. No other entity in the world runs a global model out past two weeks every 6 hours, and the difference in day-10 forecasts initialized 6 hours apart is going to pale in comparison to any improvements that would be made by improving ICs. Since I think synoptic/mesoscale obs are more important for extending the synoptic scale predictability rather than storm scale predictability, soundings every 1-6 hours would not add enough skill to make it worth it. I think with the current state of storm-scale forecast models like the HRRR and 4 km NAM, assimilated obs such as METARs and radar provide most of the improvements in short-term forecasts.