Numerical Models and Convective Forecasting

[Brian Lejeck]

I know that every day is different and that there are so many parameters that if there is a little change in the ingredients of the atmosphere, the end result could be a difference between widespread convection or a bust. I was wondering if there is a numerical model that seems to be more accurate for predicting severe weather from 0-24 hrs out?

 

[rdale]

No. It depends on the synoptic situation, the model, the input and about 100 other variables

 

[MClarkson]

Ya, you could have a discussion on this topic with some of the best scientists in the field for about the next 5 decades. Not sure that belongs in "introductory..."! But if you really wanted to look at just one model field, and one model field only, I would suggest finding the highest resolution WRF you can (probably run on a small regional grid by a nearby university.) The higher resolution probably allows for resolution high enough to avoid convective parametrization, which probably makes the forecast "better". Those do an OK job at displaying convection directly, say by looking at a simulated radar reflectivity or precipitation field. But really, theres lots of other things I'd want to look at as well.

 

[Greg McLaughlin]

You are in the same boat as the rest of us in the world of chasing and meteorology. You just don't know until things start to happen. Every now and then (like May 25, 2012) you have a really good idea, but nothing is certain. Successful chasing isn't as much about trying to be perfect in forecasting as it is recognizing how things are unfolding and having the experience to react promptly and accurately.

 

[Paul Knightley]

The best thing is to look for the ingredients for severe weather on numerical weather output and then see whether they break out at least some precip. But the main advice from me would be: don't put all your eggs in one basket! Use a blend of models to make a poor-man's ensemble output. Then, head to a location within your risk area.

 

[Bill Tabor]

Yep, in agreement with the rest. I'd add, that generally you look at the overall set up. Look at your surface chart, dewpoints / pressures in advance along with water vapor for mid level waves. Check out different models - within 24 hours I usually focus on NAM, GFS, RUC, or HRRR. Lot of details, but in particular where will the fronts and low pressure generally be set up? Where will the mid level low, along with dryline, or fronts set up and interact via convergence, diffluence, various jet streaks at different levels. Pay attention to the forecast amount of CINH or cap predicted over the regions, temperature inversions at different levels, and the precipitation forecasts of these different models. If you don't have rain, you can't have a storm. HRRR offers the convective probability forecast as well which is sort of handy as if the HRRR model shows convection the probability forecast gives you the chances of that actually coming to pass. http://ruc.noaa.gov/hcpf/hcpf.cgi Of course the day of the chase you also want to watch mesoanalysis, and visible satellite and eventually radar as echoes begin to take shape. In other words, while one particular model may not shine, you want an overall mental view of what a few of them say may transpire, how, and where. That is usually enough to give a clue what level of shot you have to risk the drive for a chase. At that time you can monitor the shorter term products (I previously mentioned) as it unfolds.

 

[Greg McLaughlin]

On days when visible satellite is useful, I do most of my forecasting based on visible satellite. I like to use surface obs too. Its hard to beat a good visible sat loop. You can see instability, low-level shear (helicity), mid-level ascent and cooling, boundaries, etc... I strongly suggest going to a site like this: http://locust.mmm.ucar.edu/ and looking up vis sat loops from past tornado events. See if you can find patterns to how convection forms, evolves....and also look at the cloud formations around the tornadic storms. There are patterns to be disvcovered. Hope this helps.