3 Models: Which is the best?
I compare three model forecasts to what happened this morning
Nashville’s horrific tornadoes are too strong a reminder that the spring severe weather season is upon all of us. We haven’t had a lot of severe weather in the last few months here in Southeast Texas, so looking at which model is performing best is challenging.
However, last night (Tuesday), I took snapshots of three of our models that allow us to zoom in to our area (mesoscale). I looked at what those models forecasted at 9 p.m. last night for 11 a.m. this morning. The three models are the IBM GRAF, the RPM and the HRRR.
The newest is from IBM which has its own supercomputer and owns Weather.com (they spun off The Weather Channel bought by Byron Allen and his Entertainment Studios). IBM also owns WSI which is a primary vendor for weather graphics (ours and every other TV station in Houston, in fact). The IBM GRAF computer is long-awaited and has excellent “resolution” (think how well your new smartphone takes pictures compared to five years ago). The downside is that it only forecasts out to 14 hours, although I expect with time, that will increase. Here’s the IBM forecast for 11 a.m. this morning:
The RPM, or Rapid Precision Mesoscale, also has excellent resolution, goes out 48 hours on a 4-kilometer scale, updates once an hour and is driven by the well-respected Weather Research and Forecast model (created by the Air Force, NOAA and the National Center for Atmospheric Research). The HRRR (High-Resolution Rapid Refresh) also updates each hour and is based on the same WRF model but until recently only forecasted out to 18 hours (now it goes to 36, but the final 18 tend to be bulky looking).
Here’s a look at the RPM forecast for 11 a.m. this morning:
And the HRRR looked like this last night for today:
So to compare the three model forecasts side by side:
The upside is that ALL of them have the strongest storms this morning north of I-10. IBM is faster with the front’s movement. The RPM and HRRR line up almost exactly, except the RPM has the front extending farther south and the RPM has a larger storm signature.
So what DID it look like at 11 a.m. this morning? Herewith:
A severe thunderstorm watch went into effect for our northern counties, not surprising given the model forecasts. We can cherry-pick any event to compare and give out trophies, of course, but it’s great news that all these models were very much in line with each other.
Which “won”? Interestingly, the IBM timing was the best, the RPM was correct with the extension of the front to the south, and the HRRR had the bullseye of severe weather in our northern counties. That is why it’s important to look at all the models. Which did I show you last night on the news? The RPM: it initialized well and was the most aggressive, and I would rather impress upon viewers the potential severity of the situation. The RPM did that best.
Copyright 2020 by KPRC Click2Houston - All rights reserved.