It's been a long, hot summer with frequent periods of poor visibility and air quality from fires in California, Oregon, and other regions.
A lot of people have asked me about smoke forecasts, so here's a quick primer.
The primary tool that I use for smoke forecasts is the High Resolution Rapid Refresh (HRRR) forecast model run by the National Weather Service. Sometimes referred to as HRRR-smoke, to my knowledge, this is the only current modeling system that predicts smoke transport across the contiguous United States. This requires estimates of smoke emissions from fires and techniques to transport and remove smoke aerosols within the HRRR modeling framework. I believe it also has the capacity to simulate the effects of smoke aerosols on cloud and precipitation processes, which is sometimes referred to as aerosol-aware microphysics.
It's important to recognize that the injection, transport, mixing, and removal of smoke aerosols varies vertically. Below is an example of a shallow smoke layer over Little Cottonwood Canyon this past Saturday. A variety or processes must be accurately simulated to account for these effects.
The HRRR is run hourly to 18 hours in the future and every six hours (at 0, 6, 12, and 18 UTC) to 48 hours. I tend to favor the latter runs with their longer forecast periods. I don't know of many sites that provide HRRR-smoke forecasts, but will discuss two here. The first are the free forecast graphics from NOAA that are available at https://rapidrefresh.noaa.gov/hrrr/HRRRsmoke/Welcome.cgi. A number of products are available including near-surface smoke, 1000 ft above ground level (AGL) smoke, 6000 ft AGL smoke, and vertically integrated smoke. I typically use the near-surface smoke for anticipating ground-level air quality and the vertically integrated smoke for anticipating impacts on sunsets and other effects.
For example, in the forecasts below from this morning's 1200 UTC initialized HRRR, the near-surface smoke concentrations along the trough forecast to extend from Oklahoma to the Great Lakes are relatively low, but the vertically integrated concentrations are much higher. In that region, air quality impacts will be small, but one would expect a deep orange or red sunset due to the existence of smoke aloft.
The HRRR-smoke graphics are conus scale and don't provide a lot of detail. If you want more regional or local graphics, you could subscribe to opensummit.com or opensnow.com.
One thing to keep in mind is that these are model forecasts so the George Box quote "All models are wrong, but some are useful" applies. Adjust your plans as needed and consider a higher error tolerance if you are someone who is more sensitive to pollution. I typically use the HRRR-smoke forecasts for planning, but adjust outdoor pursuits based on air quality observations and visual evidence. I haven't been living in a cave this summer, but I shifted the timing of my activities and shortened their duration depending on the conditions.