- May 4, 2020 2:30 pm
While there still can be snow in May, I think it’s time to wrap up the forecast for year. Let’s first recap the season, and then let me share my lessons learned from the forecast this season. Recap Setting aside the elephant in the room of COVID-19, it was an amazing season. October, March, and April all had big storms. But the true winner was February, where a powerful jet streak gave us a one-two punch of measuring snow in many feet as opposed to many inches. I think the below graph says it all, which shows the total SWE (snow water equivalent) at the Lake Eldora Snotel site. Not once in the entire year were we ever below average. That’s pretty great. Blue is this year. Purple is median. Lessons Learned So, what did I learn from this year of forecasting? Here are my three big takeaways. First, it’s not to underestimate jet streaks as a snow producer. Second, if there’s a model that’s an outlier, I should put that into my forecast discussion – I’m still embarrassed only calling for 60+ mph winds in the storm that went well north of 100 mph. I should have noted that one model was showing winds off the chart, as at least everyone would have known of the possibility. Third, as some of you have heard at some of our patrol classes, I realized that a good forecast discussion is 75% reporting. Generally you don’t, and shouldn’t, care what I think. I’m just trying to report to you what the models are saying so you can reach your own conclusions. And, 25% of the forecast discussion should be an op-ed – where I share my opinions and thoughts of what might happen – though never losing sight of my primary role as a reporter. How Did the Models Do Now, what did I learn about the various numerical models’ forecasting ability for our patrol zone? It’s easy to say that the best model in the world is the European Model – but does its amazing hurricane prediction in the Gulf of Mexico really show that it’s better at forecasting snow in the Indian Peaks? Spoiler alert, the answer is no. As I’m sure was obvious from reading the forecasts this winter, I wanted to get a sense of how each of the major numerical weather models performed throughout the season. As you’ll have noted in my recaps after each forecast, I’ve called out models for having performed particularly well or particularly poorly as to snowfall predictions. Below, I put pencil to paper in figuring out how the models performed, comparing total references to models with the number of compliments or insults I doled out in my retrospective discussions. This analysis is, admittedly, rather subjective. It’s based upon my own take in calling models out in my retrospective discussions. Perhaps to put it more bluntly, this isn’t real science, rather its running numbers based upon somewhat subjective and varying comments on how well or poorly models did. Also, it’s worth noting that for the first month of forecasts I wasn’t doing as good of a job comparing how the models predicted snow versus what occurred. And, in mid-March I lost the ability to judge results on the Eldora Snowstake Webcam due to COVID-19 shutting down Eldora – so I had to compare the forecasts with the less reliable snotel data of what actually occurred. Nevertheless, I still think this analysis is quite interesting. Let’s first dive into the numbers, and then let’s look at what conclusions, if any, we can draw. I’m only going to analyze the models I referred to by name in more than ten forecasts as there just isn’t sufficient data on the other models. I referenced the American and Canadian models 44 times, the European Model 31 times, the WRF model 30 times, and the NAM model 21 times. All other models (e.g., HRRR, RDPS) I referenced fewer than 10 times. Good calls – First, let’s compare the total number of references in the forecasts, to the number of times I remarked in the recap discussion how accurate any particular model turned out to be in snowfall prediction. Here are the percentages of times I complimented a model for doing a nice job. NAM – 29%; WRF – 27%; Canadian – 23%; European – 23%; American – 18%. It’s not surprising that the regional shorter term models (i.e., the NAM and WRF) outperformed the global medium term models – as among other things I didn’t break out the numbers by how far away the forecasts were. The one surprise from all of this is how poorly the American model performed. But wait, things are about to get both more interesting and more confusing. Bad calls overall – Second, I compared the number of times I criticized each of the models in a recap to the total references in forecasts. The numbers were interesting, and to some extent an unexpected flip of the good call numbers. I criticized the models the following percent of forecasts: NAM – 43%; Canadian – 36%; European – 35%; WRF – 33%; American – 27%. So, just looking at good calls overall and bad calls overall – its hard to reach meaningful conclusions except the WRF seems to be above average in good calls and below average in bad calls. Too optimistic – Okay, let’s now look at how often the models overpredicted snowfall. Overall I called out the Canadian Model the most number of times for being too optimistic, but it was never ridiculously over optimistic. Only four times the whole season did I call out a model for being embarrassingly too optimistic, twice with the American and twice with the WRF. So, when the Canadian, European, or NAM models are predicted gigantic snowfalls, not once were there not at least significant snowfalls. Weighted numbers of how often a model was too optimistic is as follows (labeled as a percent, though not …
- May 1, 2020 2:51 pm
Forecast summary: Warm weather continues. We’re gazing into the crystal ball to figure out the next real snow. Forecast discussion: Over the next few days, a number of weak systems pass through Wyoming. While that may mean times of gusty winds and clouds, none of these systems are forecasted to get far enough south to produce meaningful snow in our patrol zone. Sadly, the Cinco de Mayo storm that looked possible in my Tuesday forecast will likely remain too far north for any meaningful precipitation. Very warm weather will continue for almost a week. Tragically, there was a wet avalanche fatality on Tuesday near Crested Butte, and as I said in my Tuesday forecast, wet slabs will continue to be a threat in our patrol zone for the next week. When will we get snow next? Spoiler alert. Who knows? The models disagree and the predictions for real snow are out in forecast fairyland. According to the American Model, the next snowstorm will be next Friday (5/8) and dump 7.5” of snow. The Canadian Model tells a completely different story. It’s predicting a dusting of snow on Tuesday (5/5), Wednesday (5/6), and Thursday (5/7) and is not calling for any snow after that. The European Model is calling for 1” of snow on Tuesday (5/5) and 3” of snow between Friday (5/8) and Saturday (5/9). Don’t think that the safe guess is the European Model because it splits the difference. The safer guess is not to guess at all. 😊 These forecasts are too far out and the models are all over the map. K.B.O. everyone. -Jordan (Friday morning) Note: Unless otherwise noted, all forecasts are for 10,000’ in exposed areas. References to American Model are the American (GFS) Model. References to the Canadian Model are the Canadian (GDPS) Model. References to the WRF Model are the CAIC WRF Hi-Res Model. References to the European Model are the European (ECMWF) Model.
- April 28, 2020 3:24 pm
Forecast summary: Sunny and warm this week. Beware of wet avalanches. Colder weather may return in a week. Forecast discussion: Another short and boring forecast. We’ll be under a ridge of high pressure to Friday. And it will be warm. Highs are forecasted per the American Model to reach the low 60s at 10,000’ by the end of the work week. While there are a few nights that the low temperatures are not predicted to go below freezing – you should still expect that the surface of the snowpack will freeze every night, even if it’s only a light freeze. As this is our first big warm-up of the Spring, odds are that the biggest wet avalanche cycle of the year will hit our backcountry patrol zone this week. So, if you are out in the backcountry, be very warry of steep terrain this week. A system passes to our north around Friday or Saturday. Perhaps we’ll ring a little moisture out of it, though it could possibly fall as rain as opposed to snow. The American Model is calling for 0.3” of rain on Saturday. The Canadian Model isn’t calling for any precipitation. Sunday and perhaps Monday will remain very warm, but temperatures are supposed to drop early next week. Although out in forecast fairyland, all three major models are calling for a shift in weather patterns starting next Tuesday, Cinco de Mayo. While out in forecast fairyland, if all three major models are in agreement there will be a change, we can have some confidence that there will be a change. The question, of course, is how cold and how much snow. The European Model is the most pessimistic, with 1” of snow between next Tuesday and Thursday. The Canadian Model is the most optimistic, calling for 13” of snow between next Tuesday and Thursday. The American Model falls between them, but is leaning closer to the European Model, calling for 2.5” of snow next Tuesday and Wednesday. I wouldn’t pay attention to the numbers this far out – just know there’s a good chance for a return to colder weather and some snow a week out. Retrospective Discussion: The Lake Eldora Snotel site recorded 0.2” of SWE (Snow Water Equivalent) on Saturday and 0.3” of SWE on Sunday. So, for Saturday the WRF and HRRR models did the worst jobs predicting, while the Canadian, NAM, and RDPS Models called it spot-on. As for Sunday, only the American model got it right, with the Canadian model being too pessimistic, and the NAM and RDPS models completely missing the snow. K.B.O. everyone. -Jordan (Tuesday morning) Note: Unless otherwise noted, all forecasts are for 10,000’ in exposed areas. References to American Model are the American (GFS) Model. References to the Canadian Model are the Canadian (GDPS) Model. References to the WRF Model are the CAIC WRF Hi-Res Model. References to the European Model are the European (ECMWF) Model.
- April 25, 2020 1:57 pm
Forecast summary: Warm. Sunny. Darn it! Forecast discussion: Short and boring forecast. While there may be a touch of snow today (Saturday) and a touch more tomorrow (Sunday), after that it will be warm and sunny for at least five days, and probably longer. Today, the Canadian, NAM, and RDPS Model are calling for 2” of snow. The American model is calling for 1” of snow. The HRRR and WRF models are not calling for any snow. Tomorrow (Sunday), the American Model is calling for 3” of snow, the Canadian Model is calling for 1” of snow, and the NAM and RDPS models are not calling for any snow. After that, sun and warm. This will be especially true from Tuesday through Saturday when the ridge looks to be the strongest. Highs may reach 60 degrees Fahrenheit at 10,000’ by the end of the week. On the bright side, this is the first time I can remember this winter that we had such a warm lull forecasted to last for so long. If we want to gaze way out into forecast fairyland, per the American Model a weak trough may get far enough south to get to us on Monday (5/4), and another one on Friday (5/8), but a Ouija board is probably as accurate as weather models that far out (i.e., 9 and 13 days out). Historically, our snowpack reaches its peak depth (as measured in SWE) at the beginning of May along the top of the Continental Divide. Sadly this year it may have peaked yesterday. But, snow is still common in May, so keep your hopes up. Retrospective Discussion: While there were certainly winners from the most recent storm (like Berthoud Pass), there was only an additional 0.2” of SWE (i.e., Snow Water Equivalend) at the Lake Eldora Snotel site, or roughly 2” of heavy snow. So, the Canadian model was sadly too optimistic, while the European Model probably did the best job on calling this storm when looking at the Lake Eldora Snotel site. K.B.O. everyone. -Jordan (Saturday morning) Note: Unless otherwise noted, all forecasts are for 10,000’ in exposed areas. References to American Model are the American (GFS) Model. References to the Canadian Model are the Canadian (GDPS) Model. References to the WRF Model are the CAIC WRF Hi-Res Model. References to the European Model are the European (ECMWF) Model.
- April 21, 2020 4:58 pm
Forecast summary: An inch of snow perhaps this afternoon, 3-6” of snow on Thursday through Saturday morning, then warm and pleasant. Forecast discussion: Today (4/21) and tomorrow morning (Wednesday) a weak trough (i.e., wet and unsettled weather) is over our region. The Canadian Model, the European Model, and the NAM Model are calling for 1” of snow this afternoon. The American Model is predicting 0.5” of snow this afternoon. The WRF Model is not calling for any snow. Then a weak ridge of high pressure arrives (i.e., warm and pleasant weather) Wednesday afternoon into Thursday morning. On Thursday afternoon a system comes in from the northwest and lasts until Saturday morning. The Canadian Model is calling for 6” of snow from Thursday afternoon to Saturday morning. The American Model is calling for 3.5” of snow during the same time period and the European Model is calling for 3” of snow. The WRF Model is predicting 1” of snow through Friday morning (at the boundary of the forecast). Sadly, there are no indications at this time that this will be a big storm like the ones last week. But, all snow helps building the base for the summer ski season. Starting Sunday, weather gets very warm and pleasant (boo!) for the foreseeable future. Fortunately, the foreseeable future isn’t too far out when it comes to weather forecasting. Retrospective Discussion: The snow kept falling on Friday, with an additional 0.3” of SWE at the Lake Eldora Snotel site beyond what I reported in my Friday forecast. That brought the Thursday / Friday storm total to 1.9” of SWE! So, to modify my winner and loser calls among the models, the WRF Model was a bit high, the Canadian Model was a bit low, the NAM Model was even lower, and the American and European Models were way too low. An inch or two fell on Saturday. And an inch or two fell on Sunday. No snow since then. So, it looks like the Canadian model was sadly too optimistic for the beginning of this week. K.B.O. everyone. -Jordan (Tuesday morning) Note: Unless otherwise noted, all forecasts are for 10,000’ in exposed areas. References to American Model are the American (GFS) Model. References to the Canadian Model are the Canadian (GDPS) Model. References to the WRF Model are the CAIC WRF Hi-Res Model. References to the European Model are the European (ECMWF) Model.