Surface ozone air pollution modeling and forecast seems idk broken

keithhoffman22

Hi,

As an ozone sensitivity person living on the Front Range of Colorado (sunny, hot, lots of auto traffic, and quite a bit of oil and gas development), I monitor the limited number of ozone monitoring stations the state provides to avoid ozone exposure.

I was really excited to discover the surface ozone layer in Windy. I'm currently unaware of a forward forecast model or a geographic now model published by the State of Colorado so I hoped to use this layer to determine whether outdoor recreation on the followings day(s) and locations would be safe.

However, the data seems not to translate well and I'm guessing its the model.

Incongruity with available real-time monitoring data:
For example, today, 06-06-25 in Boulder, CO, the monitor station said the peak ozone concentration was 62 ppb at 5 pm.
Latitude: 40.070014
Longitude: -105.220231

Meanwhile, windy said the ozone concentration at 1700 was ~95 ppb. By 2200, the monitoring station said 55 ppb while Windy was stating 68 ppb.

That's 8 hr average

Incongruity with personal experience with altitude
To address another possible point of poor modeling, I was looking at the ozone forecast for Ward, CO, a high elevation mountain town not particularly downstream of anything. In my personal experience, ozone is likely to be quite low up at Ward, CO tomorrow (again, I am hypersensitive to either ozone or ozone with PM pollution) but windy is forecasting a peak in the 80 ppb range. Meanwhile, the Boulder forecast is 85 ppb. I can believe 85 ppb in Boulder (sounds a bit high TBH) but the idea that Ward, CO will see 85 ppb is crazy. I have no idea how the model should account for various elements but surely these very different places and very different elevations will experience different ozone levels tomorrow.

Does anyone have ideas for me? Otherwise, I think the surface ozone layer element is not especially useful.

idefix37

@keithhoffman22
The model resolution (40km) probably doesn’t allow to get detailed and reliable values at ground level.