This morning I took a look at the meteorology associated with the Yarnell Hill fire in Arizona on Sunday, and the more I dug into it, the more disturbed I got. You will see why as I explain.
From what I can glean from news reports, the fire blew up around 4-5 PM Sunday (June 30th). A nearby observation site (RAWS station) was located about 5 miles away. The observations, shown below, indicates a profound wind shift from south to north around 5 PM associated with a sudden increase of wind gusts to just over 40 mph. Solar radiation dropped rapidly at the same time, indicating a sudden increase in cloudiness.
The origin of this sudden increase in wind is clear: outflow from a line of convection (thunderstorms) that had developed during the preceding hours and which was moving to the southwest. Here are some satellite images for the hours preceding and during the terrible accident (the circle indicates the location of the fire). First image (20 UTC, 1 PM MDT, no daylight savings time there) shows the convective line to the northeast.
By 2230 UTC (3:30 PM MST) the convective line was approaching the fire and clouds had spread over the location.
A little over an hour later (2345 UTC) one can clearly see the development of a cumulus tower directly over the fire. This is call pyrocumulus. The heat from the fire can cause a tall cumulus cloud to form directly over the fire.
The Flagstaff National Weather Service radar clearly showed the approaching convection. Here is the radar at 2:58 PM. You can see the arc of red/yellow/green colors approaching the fire from the NE.
There is often an outflow of cooler air moving away from convection...the leading edge is known as a gust front (see figure). Downdraft air from thunderstorms spread out as it hits the surface, producing strong winds. It appears that there was such strong outflow from this convection that caused the
winds to shift rapidly from southerly to northerly and to increase suddenly in speed (to 43 mph at the nearby station). The vertical sounding at Flagstaff, Arizona at 0000 UTC July 1 (5 PM on Sunday) showed the potential for strong, downdraft winds, with a moist layer at midlevels (the temperature and dew point close together between 650 and 300 hPa) and dry air (big separation between temperature and dew point) near the surface (see graphic). As rain falls into dry air, there is strong evaporation and cooling, that produces negatively buoyant (descending) air parcels that accelerate towards the surface. When they hit the surface they spread out, producing intense horizontal winds.
A measure of the potential for strong downdrafts and gust fronts is something called downdraft convective available potential energy (DCAPE). The sounding at Flagstaff has values of around 1600 J per Kg, which is very high (anything above 1000 can produce strong downdrafts).
The existence of the strong convective outflow winds is confirmed by an amazing video of the area from 4 to 4:20 PM (click on image to view, cam viewing north). You will see strong winds picking up, an explosion of the fire, and then smoke pushing down towards the cam. You can see a fire line explode along the crest.
Numerical model forecasts of this event were quite good. NOAA runs a High Resolution Rapid Refresh (HRRR) system that makes high-resolution (3-km grid spacing) forecasts every hour, using many types of observational data to initialize it. The forecasts from this system, started at 1900 UTC (noon MST) and was available by 3 PM (you can access them here). I put a red oval in the first picture to show the location of the fire, and the plot shows maximum wind speeds (knots) over the past hour.. The model simulated the convection fairly well as well as the winds it produced. Here is the forecast for 2 PM...you can see the strong winds (red/purple colors) to the northeast of the fire.
The forecast for 4 PM shows the winds reaching the fire site.
The University of Arizona WRF forecasting system also indicated the potential for strong convection-related winds. (see graphic, click to expand)
You can see why I find this disaster so unsettling. Hours before the incident it was clear there was a real threat...satellite and radar showed developing convection to the north that was moving south towards the fire. High-resolution numerical models showed a threat. Were there any meteorologists working the fire? If not, why not? This terrible tragedy needs to be reviewed carefully.
A number of media outlets called the strong winds unpredictable and random. This is not correct, as shown by the information I provided above.