Only time will tell how Alaska’s 2010 fire season will go down in the record books. If spring rains bring enough moisture to keep fire danger low, Alaskans may be spared from the devastating destruction to life and property that can occur when wildfires blaze through vast swaths of the state. 

But chances are fires will flare, and if they do, along with the flames will come thick, billowing plumes of smoke. Alaskans can only hope that there will never be a fire season worse than summer 2004 when 6.5 million acres of land, about the size of Maryland, were consumed by wildfires. Smoke blanketed the state for months, extending from throughout Interior Alaska to Siberia. 

In 2004, the ability did not exist to provide precise forecasts of where the smoke would go, how long it would last and the level of dangerous fine particulates that were present in the smoke. Yet those same small smoke particles can penetrate deep into the lungs, causing serious damage, and smoke creates a severe restriction to air transport.

Since then, computational scientists at the Arctic Region Supercomputing Center have fine-tuned an Alaska-specific, high-resolution configuration of the Weather Research and Forecasting model with inline chemistry (WRF/Chem) to develop smoke forecasts.

The ARSC smoke forecasting model includes fire detection information from in situ and remote sensing sources, fire emission estimates, plume rise calculations, and calculation of the transport and chemistry
of smoke downwind from Alaska wildfires.

The smoke forecasts at ARSC also rely on the University of Alaska Fairbanks’ direct MODIS satellite data downlink and forest fire detection algorithms running at the Geographic Information Network of Alaska, as well as ARSC’s ability to compute and run operationally complex systems that crunch trillions of calculations a second. The power of these supercomputers provides the capability to run the model down to a 5-kilometer resolution for Alaska smoke forecasts. 

The WRF/Chem smoke model provides smoke prediction directly to the public and to forecasters at fire agencies and the National Weather Service Fairbanks office. Each run provides 72 hours of numerical forecasts.

Daily runs of the wildfire smoke forecast system are available during the summer fire season, which typically starts in May and ends in September. During summer 2010, runs of twice a day are expected. Smoke forecasts and model output are posted online at smoke.arsc.edu.

Using near real-time information from Earth-orbiting satellites and massive amounts of data from sophisticated weather modeling programs, computational scientists at ARSC are continually striving to improve predications of smoke behavior by running the numbers on supercomputers.

Better predictions help provide information to make flying safer and to help the public avoid dangerous smoke exposure.