Thursday, February 13, 2020

Forecast Consistency (or lack thereof) for this weekend's winter storm

  You may have heard by now that a winter storm is coming to the Inland Northwest this weekend. While that is still true we are still trying to wrestle with what's been a particularly troubling forecast? Why is this forecast so difficult compared to others we do in the winter? The biggest reason has to do with forecast consistency or in this case a substantial lack thereof. When doing forecasts beyond a couple days we meteorologists are highly dependent on using weather models for guidance. We have a plethora of models to choose from and picking the best one isn't always an easy task. In fact in the winter this often isn't easy. But whats making this forecast especially tough is the models are offering us a multiple of solutions or forecast scenarios and the consistency from run to run has been unusually poor. When we see good consistency with the models this makes forecasting easier. Below is a great example of good forecast consistency. This map shows a ridge of high pressure forecast over the Upper Midwest for this afternoon. Each frame represents a subsequent model run from the GFS starting from last Sunday and ending with the model run from this morning. Notice that although the center of the high pressure (blue H) moves slowly westward with each model run, the overall pattern does not change much. Even the positioning of the light snow (blue shading) near the eastern Great Lakes does not show a huge amount of variability. So if a forecaster were sitting on the forecast desk for the past couple days and was following this evolution they could be fairly confident in forecasting a dry over most of the Upper Midwest.

GFS Forecast consistency over the Upper Midwest for this afternoon (2/13). Courtesy of Tropical Tidbits
OK so that forecast was an easy one for a ridge that isn't over our region. How about over the Inland Northwest what's the story? Let take a look at the cold front we expect to pass over the region this afternoon and evening. Although there are no definitive highs or lows on the chart below over the Pacific Northwest, there is a fairly consistent pattern with the precipitation. Notice over the Cascades, snow was depicted by the last 8 GFS model runs as well as rain off the coast and over the western interior. Meanwhile over eastern Washington into north Idaho there has generally been a consistent band of rain and snow extending from southeast Washington and into the Idaho Panhandle. This is the band associated with the cold front. For winter weather, this is very good consistency. Even the transition from rain to snow right around Spokane has been consistent.

GFS Forecast Consistency for the Pacific Northwest cold front as of 10 pm Thursday (2/13). Courtesy Tropical Tidbits

Now lets get to the system that started this whole discussion, the one for the upcoming weekend. When we posted our graphic yesterday here is what was shown. Notice each run was showing a low (red L) but look at how much variation there is from run to run. We know that the heaviest precipitation will be associated with atmospheric forcing or lifting from the low, but in order to put that information into our forecasts we need to have a good idea where the low is going to move.  Confidence levels when dealing with this type of model to model variability result in a very low confidence forecast. 

GFS Forecast Consistency for the Pacific Northwest storm system for 4am Sunday (2/16) Courtesy Tropical Tidbits
So far we have only looked at consistency using an operational model. This is not the best way to forecast for a weather scenario that is several days out. We have better tools at our disposal called ensemble forecasts. So what is an ensemble forecast? An ensemble forecast is a set or cluster of forecasts that present the range of weather possibilities in the future. To create the ensemble forecasts, multiple model simulations are run, each with a slight variation in its initial conditions. These variations begin to increase over time (in other words we will see more variations in the model forecast 144 hours out in the future compared to 12 hours out). These variations represent the uncertainty in the forecast and suggest a range of possible weather outcomes. If ensemble runs show little variation during portions of the forecast, this would infer a high probability of outcome and a high confidence forecast. If there are large variations, this would represent a low confidence forecast. For more on ensemble forecasting check out this link.

So what do the ensembles show for this weekends storm? First lets look at the position of the low pressure center as forecast by the GEFS or Global Ensemble Forecast System. The image below shows the positioning of the low forecast by each ensemble member of the model (there are 21 of them) as well as the average in all the solutions (black isobars). The red numbers correspond to the depth of the low and its center for each ensemble member (posted in millibars) while the colors in the background show the differences in all the runs. The brighter the colors the more variability in the runs. So on the image below the greatest variability is over the north Cascades (light green shading). This was the latest run of the GEFS.  It still does not give us a great amount of confidence as there are low centers scattered around much of area from the north Cascades to northwest Montana. The mean of all these runs places a broad (albeit weak) low across much of the Inland Northwest into northwest Montana.


GEFS Low pressure centers for 4 am Sunday. Courtesy of Tropical Tidbits
How did previous eight runs of the GEFS look? Although they all show some semblance of a low, the strength, positioning, and timing are all over the place. And the variability from run to run is substantial. Typically ensemble runs and their means tend to subdue the run-to-run variability. This isn't what we like to see when faced with the possibility of a large winter storm in our future.
GEFS Low pressure centers for the past 8 runs starting with the 10 pm Sunday run (2/9). Courtesy of Tropical Tidbits

So far we have only looked at weather maps which show the general positioning of the highs, lows,  precipitation, etc. But the devil is in the details. What does this translate into tangible weather on the ground? Since snow is generally the crux of this forecast and in the thoughts of many people, we can actually look at this element over Spokane again utilizing the power of ensemble forecasting. So what does this show? Below is the answer. Each grey line represents one of the 21 ensemble members. Some of the members show as much as 7 inches of snow in Spokane, while others barely have an inch. The blue line represents the GFS run while the black line is the mean of all the models. So as you can see the answer is all over the place. It all depends on the eventual track of the low. If it moves to our north (we think this is most likely based on other models), we would likely expect to see amounts on the lower end of these possibilities, while if it moves right over Spokane or slightly south we would see amounts on the higher end of the spectrum.
GEFS Plumes for snow over Spokane for this weekend.



While we don't have a lot of confidence for valley locations such as Spokane, we have much higher confidence it will snow (heavy at times) over the mountains as they are less dependent on the track of the low and more dependent on orographical lifting of moisture over the mountains due to upslope winds. Based on pattern recognition and the most likely low track our greatest threats will be near the Cascades, Central Idaho Panhandle Mountains, Blue Mountains and Camas Prairie. But that is subject to change as the models continue to offer a plethora of solutions. All we can say for now, is standby and wait for our latest updates for this daunting forecast.