Before getting into the details of the Now Cast formula, it is worth taking one step back and looking at Ozone pollution as a global issue. The total atmospheric ozone concentration simulation, done by the NOAA gives a good overview of the areas with high/low Ozone concentrations, but also shows how global climatic conditions impact trans-boundary pollution.

The above simulation shows the total atmospheric concentration, from surface up to 20 KM height, and thus contains both the good Ozone (the "Ozone layer" in the upper stratosphere) as well as the Bad Ozone in the troposphere.

When one talks about Ozone pollution, it is the tropospheric ozone, and more particularly the surface concentration which matters in order to quantify the health impact. Moreover, the tropospheric Ozone is having a diurnal cycle, with pollution peaking in the afternoon when the temperature reaches its maximum, and almost no pollution during the night. This cycle can be clearly seen in the below animation, which is based on the forecasted ground concentration.

Without surprise, Ozone pollution is still a global problem even for surface concentration (countries in the Southern hemisphere would have higher concentration during their summer).

Looking at China in particular, Ozone is actually becoming more often the primary pollutant, surpassing PM_2.5 pollution. The main explaination is that all the recent efforts done for reducing air pollution in China have been really efficient and PM_2.5 concentration significantly reduced ^[5].

Moreover, PM pollution is actually acting as a global warming cooler (i.e. decreasing the global warming by reflecting sun light). So, when PM pollution gets reduced, so are the number of blue sky days (that's good!), but as a side effect the peak daily temperature increases. And since temperature is one of the element which impacts Ozone production, increased temperature also means increased Ozone

Now that it is known that Ozone pollution tends to become the primary pollutant during the hot summery day, it is even more important to fully understand the impact of the Ozone AQI calculation on the overall AQI.

The two graphs below show the AQI for 5 different cities in China. The graph on the top shows the AQI as it is published today on aqicn.org (using the instant cast formula), and the graph below shows the AQI as it it would be using the AirNow NowCast AQI formula.

The most striking problem with the NowCast AQI is that it tends to overshoot and report extremely high AQIs when the actual (past hourly data) AQI is low. For instance:

• For Beijing on Thursday August 3rd as 11PM, the Now Cast AQI is reporting an AQI for 158 (Unhealthy) while the actual AQI is 45 (Good).
• For Shanghai on Friday 29th July at 20PM, the Now Cast AQI is reporting an AQI for 180 (Unhealthy) while the actual AQI is 50 (Good).

Trying to fine tune the NowCast formula, as it has been done for the PM_2.5 Now Cast in Asia does not work well for Ozone (see graph below for N=3 hours, and min weight=.1). The reason is that Ozone is having a diurnal cycle with peaks in the afternoon (correlated with temperature peak), so any kind of averaging would just offset the peak by few hours. That would be wrong, as what is needed is to report the Air Quality as it is now, and not as it was few hours ago.

There is however a problem with the current Instant Cast formula for Ozone which is in use on waqi.info and aqicn.org: It is that the Instant Cast breakpoints are based on the US EPA 1 hour air quality standard, which is still among the least safe Ozone standards compared to other EPAs standards (see graph on the right) for AQIs below 200.

There are quite a few reasons for this low standard from the US EPA: First the USEPA focused first on improving the 8-hours standard, and that's something good - since the 8 hours standard is used for annual reports. It is also the standard used on air now (together with the now cast formula), so it made sense to focus first on this standard, and this even despite all the problems linked to the now-cast formula. Most likely, the US EPA is now working on doing the same update to the 1 hour standard, but that might take quite some time to happen, and quite a lot of lobbying.

So, while waiting the the US EPA to improve the 1-hour standard, and setup a standard which is not left behind compared to other Ozone standards in the world, the world Air Quality Index project has been looking at existing standards which could better reflect the Health Impact of Air Quality, and this without having to use the irrelevant NowCast formula based on 8-hours breakpoints.

One of the options which has been considered, and immediately dropped was to use an Instant Cast based on the 8 hours breakpoints instead of the 1 hour breakpoints (like it is actually done for PM_2.5, where the AQI is reported assuming a 24-hour exposure). The reason for not using this solution for Ozone is the diurnal cycle, which makes it not straightforward to assess what an 8-hour exposure can be. For instance, if an AQI is reported to be 35 at 10AM, what does that mean for the AQI at 2PM? Will it be 80, 120, to 180? That's quite different from PM, where one can guess that the AQI is similar throughout the day, under similar visibility and humidity conditions^[6].

For the other options, the analysis covered existing standards for more than 20 EPAs worldwide, the recommendations from the WHO, the preliminary findings from world-wide researchers from the TOAR project and the research papers from the from the UNEP CCA Coalition catalog. An additional constraint for selecting the standard was to use one which is officially approved. as well as one close to the median standard, as it would not over exaggerate the Health Impact, and neither underestimated it.

The best option appeared to be the CN MEP 1-hour standard, which is much more safe than the US-EPA 1-hour standard for AQI below 200 (above 200, the two standards are the same). The two graphs below show the Ozone Instant Cast AQI, based on the old breakpoints (top) and new breakpoints (bottom). Note that since the breakpoints are updated for AQI between 50 and 200 only, the only 'visible' difference on the color graphs is for Moderate, USG and Unhealthy AQIs.