METEOROLOGICAL CRYSTAL BALLS: HOW TO GET ACCURATE FORECASTS FOR BACKCOUNTRY LOCATIONS
Words & Photos from Michael DeYoung (@michaeldeyoung)
Spencer Glacier, like most Alaska glaciers, is disappearing. I've seen dramatic shrinking in my lifetime (I've been photographing the same glaciers for 30 years) just as I have of other nearby, accessible glaciers such as Portage. This gorgeous tucked away gem's hallmark is the ever-present, huge icebergs grounded in the shallow end of the namesake lake. These towering ice monoliths are surrounded by fields of wildflowers in July and August beneath the verdant green and snowcapped peaks of the Chugach Mountains. The reason the glaciers are here is that this place is cold and wet. More often than not, conditions are challenging here, even in summer, and many mid-latitude visitors are shocked to be in a hypothermic environment when much of the country they come from is locked in brutal August heat.
I watch the changing weather patterns closely. When a favorable fair-weather window appears in an otherwise wet place, we drop everything and go. We know this opportunity will be short-lived and rare. Thankfully, advances in meteorology, technology, and access to weather information far from where people live in a remote corner of North America allow opportunities like this to be seen days in advance. In our lifetime, we know this lake, like others around it, will no longer be filled with the mystique and majesty of ever-changing icebergs calved from a once-massive glacier.
My main goal today is to provide a basic overview of where forecasts come from and where backcountry adventurers should get them from. Knowing how easily I can go off the rails with technical meteorological geek-speak, I will try my best not to do that. This is just a 30,000-foot overview. I believe webinars and workshops are the places where interested backcountry travelers can learn more about how atmospheric model forecasts are made and how useful they can be in their trip planning and decision making.
Operational forecasts come from atmospheric models that rely on a global network of an unimaginable amount of data collected multiple times a day. Millions of meteorological data points from surface and upper-air observations and derived from ever-improving satellite analysis from all over the world are collected multiple times a day, 24/7/365. These data are run through supercomputers at NOAA's National Center for Environmental Prediction (NCEP) and by the European Center for Medium-Range Weather Forecasting. The World Meteorological Organization (WMO) founded in 1950 and based in Geneva, Switzerland, has forged and maintained international scientific cooperation and data sharing second to none. No private forecasting firm has done this.
The US contributes to and greatly benefits from the global sharing of meteorological observation and forecast data. The earth has but one atmosphere that knows no political boundaries. Forecasts for our own country depend upon modeling the entire atmosphere first. What's happening in the polar regions and tropics has profound effects on the upper air patterns flowing over the US and the weather we see. It makes no sense to begin modeling right off the Pacific Coast and end it as soon as systems enter the Atlantic.
Paddle on a tranquil glacier lake amongst icebergs, keeping a safe distance of course.
The two principal global models used in US forecasting are the Global Forecast System model (GFS) run by NOAA and NPEC, and the European Center for Medium-Range Weather Forecasting (ECMWF), a public/private entity from the EU, based in the UK, that US forecasters use as much as they do our models.
A regional model we use in addition to the global models is the North American Mesoscale (NAM), also run by NOAA and NPEC. The global models run forecasts out to 10 days while the NAM only goes out to five. The Swiss global model, MeteoBlue, is an ensemble model taking other models into consideration and applying artificial intelligence. The Meteoblue is thought to be especially accurate in remote mountainous regions. From my unscientific observations and simply as a result of pattern recognition, I agree that the Meteoblue is more accurate for temperature and wind forecasts for remote canyons and mountains here in the western US. I use the Meteoblue frequently in my own trip planning. The ECMWF is updated twice a day, and the GFS is updated four times a day.
Figure 1: From windy.com I was able to zoom in to a fairly detailed topographical map which included trails, roads, and landmarks. Using the weather picker feature, I selected the summit of Mt. Whitney, the highest mountain in the CONUS.
How does this help those who are serious about understanding weather more for their remote adventures? We'll get there. First, here are a couple of important points to make.
As mentioned above, virtually all observational data, including meteorological satellite and radar observations that go into operational model forecasts, are all run by government agencies overseen by the WMO. No private industry, not the Weather Channel, AccuWeather, Weather Underground, or any other entity, operates its own global network of meteorological observations or produces its own weather models. They all use government-produced models (the US GFS model is free) in their operations.
All of the weather models produce detailed forecasts in three-hour and even one-hour intervals for any point on their grid. That means any place - not just cities, towns, and popular tourist destinations. All that's needed are GPS coordinates (lat., lon., elevation), and away you go. The models take into account the elevation of the selected location as well as orientation to surrounding terrain as interpreted by the model's terrain profile. When you look up a forecast for a city anywhere in the world from your weather app on your smartphone, what you are basically seeing is the automatically generated Model Output Statistics forecast for that location presented with the company's own graphics display.
From the websites of private weather firms, I could not get a forecast from a simple search or their location databases for Mt. Whitney's summit in the Sierra or for Boucher Rapids and beach at the bottom of the Grand Canyon, or other remote places. Using forecasts from nearby urban locations, such as using a forecast for the town of Lone Pine to reflect what might happen on Mt. Whitney simply won't suffice.
Figure 2: This is the same meteogram as seen in Figure 1, with my chosen location of Mt. Whitney shown as a white dot with white circle.
Thankfully you can get useful forecasts for remote locations that factor in elevation directly from the models themselves and directly from the National Weather Service website. This can be a game-changer for those willing to invest the time and understand the capabilities and limitations of modern forecasting.
Backcountry travelers wanting serious and useful forecasts should move away from using oversimplified icon-driven forecasts seen in popular mobile weather apps on their smartphones. Don't even consider forecasts you see or hear from TV or radio for backcountry trips and planning. There is no way of knowing how current they are. The best strategy is to go straight to the horse's mouth, which is the National Weather Service and to raw weather model data. Until recently, understanding uninterpreted atmospheric model data and charts was something only trained meteorologists could do. This has changed, thankfully.
Figure 3: This is a comparison of forecasts for Mt. Whitney by 5 different weather models showing disparities in temperatures between the models.
Now, we have apps with fantastic graphics presentations that allow non-meteorologists to better understand forecasts produced by current model runs. My favorite app and the one I think is most useful for remote location forecasts, is the Windy app. You don't even need the app. You can go right to their website, windy.com, sign up for a free account, and start using it. In fact, I prefer using it on my desktop or 15" laptop over a mobile device because a bigger presentation is just easier to understand. Windy is widely used by people whose livelihoods and safety depend upon accurate and current forecasts, such as aviators and mariners. Backcountry travelers will find it useful too.
What I like most about Windy is it presents amazing weather forecast parameters right from current model forecasts that are easy for the non-meteorologist to understand and use with a little foundational knowledge and practice. Windy has integrated worldwide topographic maps complete with roads, trails, and geographic names so adventurers can pinpoint an exact location and get a site-specific forecast for that point presented in either a simple icon format or a more detailed Meteogram, which I will discuss in a future blog post.
See examples from the summit of Mt. Whitney, the highest point in the Continental US (Figures 1-4), and at the bottom of the Grand Canyon at Boucher Rapids (Fig. 5) Figs. 1 and 5 are topo maps taken straight off the windy.com site.
Windy does not make its own forecasts. They simply present the foundational tools produced by global weather models from which forecasts are derived from. They give you a choice of the main weather models used in global forecasting, including the ECMWF. You can easily verify that you are looking at a current model run and not an outdated forecast. Windy gives users free access to many useful products forecast by the ECMWF, even more so than their own website. The European model is in many ways more accurate than our own GFS, especially in areas with complex terrain. Windy also presents limited forecasts from Meteoblue in the form of basic Meteograms where you can compare forecasts with other models.
Figure 4: This is the NWS forecast for the summit of Mt. Whitney for the same time period as the ones we see from the windy.com site.
You can also get useful forecasts from NWS websites for any place in the US, including remote locations as well, but they are limited to plain text and only get fairly close to your exact location. For example, each regional website (see figure 4) allows you to select a point forecast, which is great. However, I couldn't get the exact elevation for the summit of Whitney, at 14,495'. I could only get close with an elevation of 12K. From there, I would adjust the temperature forecast for the difference in elevation by a few degrees. The advantage to using NWS is you get forecasts that are adjusted for model biases by knowledgeable forecasters that actually live in the area they forecast for. But the NWS does have specific forecasts factoring in elevation for many popular remote locations in their database - places like Phantom Ranch in Grand Canyon National Park, whose forecast would differ greatly from the rim of the Grand Canyon.
The National Weather Service has 122 regional offices called Weather Forecast Offices. What I like about their approach is that meteorologists live in the area they are forecasting for. This is important to me. It's good to know that the forecasts for the Grand Canyon are made by meteorologists who live in Flagstaff and understand model biases for their area of responsibility. For example, they might know how the GFS consistently under forecasts high temperatures at the bottom of the Grand Canyon. The GFS doesn't seem to factor in thermal mass heat retention. An NWS forecast specifically for Phantom Ranch will adjust temperature forecasts to correct for the model bias. Their primary mission is public safety, of course, so backcountry forecasts are not much of a priority, but from the local NWS site, you can also get forecasts and info for other things of interest to backcountry folks such as marine, aviation, avalanche, flash flood, wildfire and river forecasts, something you can't easily or always get from Windy.com.
Figure 5: This is also from windy.com and shows a specific location, Boucher Rapids along the Colorado River in Grand Canyon National Park and the meteogram produced by the European Center for Medium Range Weather Forecasting, from January 30, 2021.
While the NWS excels at the science of forecasting and model improvement, they are scientists, not graphic designers. So, their user interface and graphics presentations are a little lacking compared to some of the better apps like Windy. When feasible, my strategy is to look at both Windy and compare it with what the NWS is forecasting. I can get more useful cloud information from Windy to determine colorful sunrise or sunset potential than I can from plain language NWS forecasts, but I understand that the NWS is not forecasting for the photographer's interests. NWS sites and forecasts will also factor in things such as widespread wildfire smoke, ash from volcanic eruptions, etc., and how they affect the forecast.
Long ago, I came to terms with meteorology being a thankless job and accepting there are skeptics who don't see any accuracy or usefulness in weather forecasting. I believe the vast majority of fellow backpackers, backcountry skiers, river runners, and other outdoor enthusiasts, including guides and outfitters, would appreciate the science and capabilities of modern forecasting as much as aviators and mariners do. Avalanche conditions and river hydrology are directly connected to atmospheric changes. You can bet that snow professionals and winter backcountry adventurers are paying attention to snow and avalanche forecasts, just like river runners might pay attention to hydrology forecasts - all produced by the NWS.
The global weather models can accurately forecast the development, dissipation, position, trajectory, and strength of upper atmospheric features, including the jet stream up to five days and sometimes longer depending upon the pattern. Problems arise when upper-level systems and any associated fronts cross complex terrain like major mountain ranges, especially mountainous coastlines, and abrupt changes from mountain to desert areas. Complex terrain has a profound effect on weather systems. Each model has a terrain profile, and some are more detailed and accurate than others. Each model also has a minimum resolution in that any weather feature smaller than 22km for the GFS, or 9km for the ECMWF, won't be noticed or forecasted. A five-mile stretch of river in a remote canyon can be cloaked in fog and not be seen by models but can significantly impact anyone in that corridor. Very localized gap winds, the kind you might experience near Ogden Canyon in the northern Wasatch, might also go undetected or forecasted. But a knowledgeable person can recognize the surface pressure pattern that can produce strong local winds.
People who understand the complexities of mountain weather and understand weather model forecasts' weaknesses and biases can still get value from the products available from NWS, Windy, or other sites. As someone who has followed this closely, I know that forecasting is far more accurate today than it used to be and getting better as now. With AI, models can begin teaching themselves how to forecast better.
Not all useful forecast data comes from models, apps, and forecast offices. Useful information and short-term (a few hours) forecasts can come from paying attention to what's happening around you on location. Being able to identify cloud types and observing trends can tell you a lot about what the day has in store.
Michael DeYoung is a photographer, a wilderness traveler, and a weather guru who, together with his wife and adventure partner Lauri, has bucked corporate tradition and forged a life as a freelance adventure and travel photographers in the ’90s. They have shot campaigns for a wide variety of tourism clients and outdoor publications, including many assignments promoting Alaska as a visitor destination. When not out in the wilderness, they spend their time near Taos, New Mexico, in a 100% solar powered, sustainable straw bale home and office. Michael has turned more toward leading and guiding photography workshops and uses his meteorological expertise to teach others forecasting for wilderness travel and finding the best photographic light.
Portfolio: michaeldeyoung.com
Stock: mdphoto.com
Photo Tours: Active Photo Tours
Instagram: @michaeldeyoung
Sign up for our Newsletter