Cool tool

I don't know how long this little thingy has been around, but I found it really sweet.

I was cruising the NCAR site and was looking at MM5 models and came across the "Simulated Radar - -dBZ".

I don't usually use the NCAR MM5 model, how long has this trick been around?

Simulated radar images? Now THAT kicks a$$.

how long has that been there?
 
RE: UCAR MM5

I played around with the UCAR MM5 data output a while back, and then UCAR stopped updating the plots on a regular basis. Yes, the simulated radar is "interesting", and the few times I looked at the model, it did seem to be more or less consistent with what actually happened. Supposedly, one can download the MM5 source and run it on their own machine. Beyond that, I don't know anything about it: what platforms are supported? Who, if anyone, uses it operationally? Bill

http://www.mmm.ucar.edu/mm5/mm5-home.html
http://www.mmm.ucar.edu/mm5/mm5v3/wherev3IBM.html
 
Originally posted by Kyle Masters

I don't usually use the NCAR MM5 model, how long has this trick been around?

It's been around a long time, at least since the mid-70's. Any model that has precipitation processes in it can take those fields at a specific moment and calculate what a radar might see. This is almost never done exactly as it should be - but nor is the way precipitation is done in models. The reason you won't find it in operational model products very often is that it only tells you the instantaneous precipitation at the time the model data is valid. More often, the accumulated precipitation since the last forecast product shown is given, such as six hour acculumated precip, since most folks are more interested in if it is going to rain than how intense it might be at say 1 PM. Now that model forecast products are being generated at higher frequencies, such as hourly, I suspect these sort of products will become more common. To me, these aren't very useful for making a forecast, but I think they've been used for some time the broadcast field.

Glen
 
Originally posted by Glen Romine
To me, these aren't very useful for making a forecast...

Glen

Not only that, but I haven't found the output to be very accurate at all... I seriously don't think we're approaching this idea of storm-scale modeling very efficiently, mainly because it seems that the limiting element is an accurate, high-resolution analysis of the atmosphere. With the paucity of the surface ob network (even the OK Mesonet), not to mention the upper-air network, I don't think modeling to a smaller and smaller scale (e.g. individual forecast radar returns of storms) without a better analysis going very productive... Perhaps if there's an increase in the spatial and temporal resolution of the observational networks we'll see better results. So far, however, I haven't been impressed with any 'simulated dbz' forecasts...
 
Originally posted by Jeff Snyder
I seriously don't think we're approaching this idea of storm-scale modeling very efficiently, mainly because it seems that the limiting element is an accurate, high-resolution analysis of the atmosphere.

It's tough business - this is actually what I'm working on for my research (more specifically polarimetric radar data assimilation). Storm-scale data assimilation is still in its infancy - don't expect major improvements in real-time forecasts on the convective scale for at least 5 or more years, more likely 10. Very little radar and sat data is currently used in model forecasts - but a number of folks are working on ways to do this and should lead to improved short-term model forecasts - say the 30 minute to 2 hour range. Computing power is still grossly inadequate to make real-time probablistic forecasts - but expect to see these in the future. Those of you planning to go to the TESSA conference will get a glimpse at this from Lou Wicker's presentation.

Back to the issue at hand - the main value of the simulated dBZ products is it can give you a glimpse at a possible convective mode, for instance if your model resolution is sufficient you might note mesoscale convective systems, squall lines or discrete cell development within the model simulated reflectivity field - and sometimes this actually pans out. Folks are investigating the quality of these kinds of solutions, particularly from WRF - but I've not seen any specific numbers with regards to how often the model 'gets it right'.

Glen
 
Originally posted by Chris Rozoff
....do you think accurate quantitative information (like precipitation accumulation at surface) will be restricted so long as bulk microphysics schemes are used by the models? I guess there are more accurate microphysics methods which are quite expensive. But one obvious application is usefully predicting the size of the hail at a given location, which is not an easy task in our current condition.

Well, that is the state of the art stuff that I and many others are working on. The microphysics get extremely complex - but the vision is the we will be able to predict hail swaths (including max size), mesocyclone tracks, wind damage, etc... as probabilistic forecasts. This is an old idea that has been around for some time - but has in the past not proved very successful. Newer observations and techniques to optimally use them combined with greatly improved computing power are making these types of forecasts from ensemble data assimilation systems at least theoretically possible. Probably won't help with convective initiation much - but will hopefully pick up storm character in advance and offer hope of identifying which storms will become most severe, when and where it is most likely to be along with the envelope of uncertainty.

Glen
 
Re: MM5

This is all fascinating stuff, however I think you will always reach a point where you're limited by the temporal and spatial resolution of the atmospheric sampling - which is rather coarse today and probably will not improve significantly for several more years.

- bill
 
Runs of the WRF at a 4 km resolution were done in support of project BAMEX in 2003, and some of the reflectivity output was pretty darn good. It was generally quite good and delineating storm type, although location was not always great.
 
During the summer I like looking at the RUC moisture convergence fields on the UCAR site. Those purple blobs seem to pinpoint future convection pretty well.
 
Originally posted by Joe Nield
Runs of the WRF at a 4 km resolution were done in support of project BAMEX in 2003, and some of the reflectivity output was pretty darn good. It was generally quite good and delineating storm type, although location was not always great.

Could see 2 km WRF products this spring - could be pretty interesting how well it performs at delineating storm types.

Glen
 
Back
Top