WindNinja: A Weather Station In Your Pocket

Over the years, wind shifts have caused untold damage to communities and claimed the lives of firefighters and civilians alike. Although the National Weather Service (NWS) provides regular forecasts of the weather across the country, it could never drill down to a small enough area to be useful for firefighters.

So something more was needed, something which could sharpen the focus of those NWS forecasts to make them truly useful to fire crews. Thanks to the U.S. Forest Service, that ‘something’ is now available to anyone with a computer or a smartphone. And it’s called ‘WindNinja.’

Jason Forthofer, a Mechanical Engineer who works as a Division Supervisor for the U.S. Forest Service during fire season, oversees a number of projects at the Missoula, Montana, Fire Lab. He reflected on the origins of WindNinja. “I’ve been working here at the Fire Lab for about 19 years, and a lot of my time here has been developing WindNinja. This is a model which a team here developed, pretty much all written in house. It started out as my Master’s thesis around 2005.”

WindNinja simulates high-resolution wind in mountainous terrain and was specifically designed for operational firefighting use – meaning that it works under time and computational constraints of firefighters and fire managers.

“It was built to simulate winds quickly on computers that would be common for firefighters and fire managers to have,” Forthofer explained.

“The models the National Weather Service might run are typically at a coarser resolution because they are more concerned with winds for western Montana than they are in winds just in the Missoula Valley, for instance. They also have supercomputers to run them on.”

Three Ways to Use WindNinja

The WindNinja model is more site-specific than the NWS models, so the user can define the area where a fire is located, usually in a 50-mile by 50-mile area, roughly the size of a larger fire. The user then provides some simple inputs into a GUI (Graphical User Interface on the screen) on their computer, hit ‘Go’, then choose one of the three different ways to drive the flow in WindNinja.

Weather Model Initialization, the first mode of operation, is where WindNinja accesses an NWS server on the Internet, from which the computer downloads their latest coarse model run, then uses that information to simulate the wind at a higher resolution.

The WindNinja model is basically a spatial down-scaling of the NWS model. Whereas the NWS models might be at four-to-twelve kilometer resolution, WindNinja converts this data into 100-meter resolution.

“The advantage of our model is that we can simulate the effect a particular ridge or drainage is going to have on the wind flow, whereas the NWS can only simulate the effect terrain is going to have on wind in a major valley, like the Bitterroot Valley,” Forthofer confirmed.

“We take shortcuts with our computations and the physics we simulate are less than the NWS, so we capture the first order effect the terrain has on the wind, the mechanical influence terrain has on wind. Some of the more complex effects caused by moisture and heat, like thunderstorms, are not captured by WindNinja.”

But what data they can use in the app, they do. “Any information that came from the coarser model is included in our model, so if there was a thunderstorm that was predicted with the NWS model, that information shows through into our model, too, so it can take some of those effects into account. This is probably the best way for firefighters to run WindNinja.”

Another way to run WindNinja is to specify measurement points using data from Remote Automatic Weather Stations (RAWS) at the airports, which is available over the Internet, specifying one or more of those in the domain area. That information can be used to drive the wind flow. This information, which was taken in the past – either an hour ago, or five years ago – is used instead of the NWS information, which is a forecast of the future. It can be used in reconstruction mode, if the user is trying to reconstruct fire behavior and wind in the past.

The third method in which to use WindNinja, which is probably the simplest way, is called a ‘domain-average wind,’ where the user specifies an input wind speed and direction at 20-foot height, which is the general upper-level wind that would prevail over the whole area. WindNinja uses that to show how wind gets channelled and sped up over and around ridges. Although this is mostly just the mechanical flow of the wind, there is also a useful sub-model which simulates diurnal winds, that is upslope and downslope wind flow due to heating and cooling of the earth’s surface.

Even though these models are first-order (mechanical) models, not as sophisticated as the NWS models that take into account more weather-related phenomena, WindNinja models are orders of magnitude faster. They can also be run on smaller computers – no supercomputer required!

“There are now two solvers in WindNinja,” Forthofer explained. “The original solver is called the ‘mass-conserving solver,’ and the other one is called the ‘mass and momentum-conserving solver.’ The difference between the two is that the latter has a little better accounting of the physical processes going on by conserving the momentum of the flow, whereas the first one only conserves mass. The real difference that firefighters will see between the two models is that the mass and momentum-conserving model will simulate the wake flow you might see on the lee side of a ridge or mountain as the wind is blowing over it, whereas the mass-conserving solver won’t do a good job of simulating that in the flow.”

Simplicity is the Name of the Game

The hallmark of this innovative app is its simplicity of use, as Forthofer explained. “It runs fast at high resolution. There was way more code put into the GUI than the actual physical equations because there was a huge effort put into making it easy to use by firefighters, automating a lot of the data fetching over the Internet.”

Once a user sets up their inputs, most of the operation is pretty automated. The user doesn’t need Geographic Information Systems (GIS) skills, and only rudimentary computer skills, according to Forthofer. “It has a built-in graphics window, with a Google Maps look that the user can zoom into the area they want to simulate. Click a button and swoop a box out to define your domain, then hit ‘Go’, select one of the three model choices, the type of output files you want it to write. The simulation will take a minute or two, depending on the type of computer you’ve got. When it’s done it spits out files, including a Google Earth file.”

If Google Earth is installed on the computer, it will activate when the user double-clicks on the file.

“What you’ll see is a little map of wind vectors, little wind arrows, that are different colors, with red being the highest speed and blue the lowest, draped over the terrain about 20 feet above the vegetation.”

Twenty-foot height above the vegetation, whether they be bushes or 50-foot-tall trees, is the standard in the U.S. This is the height that RAWS stations are set at and the height the information that comes from the NWS is usually set at, as well.

In addition to the Google Earth display mode, the app can also create a ‘shapefile,’ a simple, non-topological format for storing the geometric location and attribute information of geographic features, as Forthofer pointed out. “A shapefile allows you to overlay wind vectors on a map you might be trying to generate using ESRI’s ArcGIS program. There are also ASCII raster files generated, which can be fed into fire behavior modelling programs like FARSITE and FlamMap. Before this model was developed, users could only enter one wind speed and direction that prevailed across the whole terrain.”

Due to the disparity in processing power between desktop computers and smartphones, WindNinja’s phone app is less feature-rich, but still useful to firefighters dealing with winds during a wildfire. “There is the desktop version, which we’ve just discussed, and a mobile app now which you can run on your phone, which is a trimmed-down version of the desktop WindNinja with a lot less choices,” Forthofer explained.

“Phones are not powerful enough to do simulations on the phone. The app has a map and a graphic interface where you can set up a run by zooming into your area, swoop out the box, enter a couple more inputs about what kind of run you want to do, then hit ‘Go’ on the app, then it sends that information to our server, which does the simulation, packages up the output file and notifies you when it’s ready, then you download the file to your phone. The file stays on the phone, so you can go out of cell coverage and still view the wind vectors on your phone.”

Worldwide Coverage

Proud of his team’s creation, Forthofer has released WindNinja to firefighters across the globe.

“WindNinja is used around the world for fire management, which makes it unique because it’s the only program that is.”

And the number of users is growing every month. “We just released the free mobile app about two years ago and have about 2,000 registered users for it now.”

But there is one limitation to the smartphone app for international users. “The mobile app only works in the U.S., since it runs on an Amazon Web Service account that we pay for.”

But don’t despair if you are trying to track wind patterns outside of the U.S., Forthofer’s team has you covered. “One of the NWS models is called the GFS, the Global Forecast System, which simulates weather for the whole world. That’s one model WindNinja can use on a desktop computer. It’s a coarser model than the ones the NWS runs for the U.S.” The software is open source, so if other countries want to set up their own weather servers to run the phone app, they are welcome to do so.

“The Canadians use Prometheus [the Canadian Wildland Fire Growth Simulation Model] and can use WindNinja’s desktop app with that,” Forthofer said.

“Technosylva [which provides advanced GIS-enabled software solutions for wildfire protection planning, operational response & firefighter and public safety] also uses WindNinja embedded in their software, which is used around the world.”

Thinking Outside the Box

As the number of WindNinja users grows within the firefighting community, sub-groups are also experimenting with its application to their jobs. “Smokejumpers have used it a little, they’re expressing some interest in it,” Forthofer disclosed. “It’s been used to train cadaver dogs, to study wind power generation sites, finding the best places to put wind turbines, some people use it for windsurfing, and parachuting.”

And it may even have some out-of-this-world applications. “It has been used for wind calculations to simulate dust and wind erosion on Mars,” Forthofer added with a chuckle. It’s not likely the U.S. Forest Service will be dispatching any strike teams to Elysium Planitia anytime soon, but it’s nice to know they can tell them what the weather is like if they do.

Further Information

U.S. Forest Service has extensive information, including downloads and manuals, available at the WindNinja website.


Recommended Articles