MPAS forecasts for 16 May 2016

Yesterday's forecasts were incredibly consistent between CAMs in developing a supercell or two in the vicinity of southern Colorado and the Oklahoma/Texas panhandles, before growing those storms upscale into an MCS that would then sweep across Oklahoma overnight. Due to the consistency of these CAMs, our full period outlooks showed high probabilities, with high certainty, over a relatively small area:

So today, I'd like to talk about one of the convection-allowing models that was showing a signal for severe weather: the Model for Prediction Across Scales (MPAS). MPAS is a model with variable grid spacing, and for the 2016 Spring Forecasting Experiment the finest, 3km grid spacing is centered over the United States. As the forecasts run out five days, we have five forecasts for each day to examine. The upscale growth of one or two supercells into a linear system was clear even at Day 4, in the 95 hour forecast below. Each of the forecasts below is valid for the same time (23Z on 16 May), but depicted for a different run of MPAS, indicated by the "Init:" date in each frame.

These forecasts get the general location of the supercells fairly well at 23Z, although there are additional storms over central Oklahoma in the 5 day forecast (hour 119, above). Storms are displaced too far east in the model forecasts, but the overall orientation and coverage of the storms is captured quite well when compared to observations.

While the evolution into a linear system evident at 5Z is beyond the range of the forecast initialized on 12 May, the other forecasts consistently show this upscale growth. 

None of the model forecasts extend the squall line as far south into Oklahoma as what occurred in actuality, although the structure of the MCS was well depicted as far out as 101 h, with the bowing segment in the reflectivity evident in the MPAS forecasts. The positioning also improved as the day of the event drew nearer, bringing the system further west with each run.

Every morning of the SFE except for Monday, the desks verify their forecasts made the previous day. At the total severe desk, when we were finished rating yesterday's forecasts we discussed the MPAS forecasts. Participants and facilitators agreed that yesterday was a very good forecast by MPAS, showing a long-range signal that other CAM guidance would echo once the event drew nearer. We discussed a few details that were incorrect, such as timing and placement errors, nitpicking the extent of the stratiform and the easterly bias caused by overmixing of the dryline. We also found a supercell that developed in the Texas Panhandle in the latest run of the model that did not occur in reality. This case was an excellent testimony to how far CAMs have come - for us to be looking at such small details means that the overall scenario portrayed by the model was excellent. 

MPAS, along with the other CAMs, gave us enough confidence to make a very specific forecast, which ended up being the best forecast we've made as of yet in the experiment. All participants on the total severe desk assigned ratings of 9 or 10 out of 10 for this particular full period outlook. While some operational models such as the NAM showed signs of discrete supercells growing upscale into an MCS, the signs were only evident to the experienced forecasters of our group (for the record, the signs were a relatively small area of intense updrafts, followed by a large area of downstream high QPF). Yesterday, the CAMs gave us the detail and confidence to issue an excellent forecast - one might even say it was practically perfect.