Is West Coast Air Getting Dirtier?

Analyzing particulate matter trends in three major Pacific metros

Photo taken by bit.io employee in San Francisco, on Sept. 20, 2020 at 9:26AM

The data for this analysis was analyzed with the help of bit.io, a standards-compliant cloud Postgres database. bit.io is the fastest way to get your data into a private, hosted Postgres database. Follow bit.io on Twitter at @bitdotioinc.

Wildfire woes

Wildfires are getting worse by nearly every metric. Fires in the West are increasingly becoming large conflagrations that burn millions of acres of forest and destroy thousands of structures each year. The pervasive smell of smoke and occasional orange sky are now familiar to West Coast residents, yet national reports show continued air quality improvement. What does the data reveal about these seemingly-conflicting trends?

The recent return of smoke to Seattle and the Bay Area inspired us to look more closely at the impact of wildfire smoke on West Coast air quality and if the air there is, despite national trends, getting dirtier.

The data

National reports of continued air quality improvement are dissonant with the frequent smoke in Seattle, San Francisco, and Los Angeles in the summer and fall. An EPA-computed national average shows a 19% decrease in the mean annual concentration of fine particulate matter, the most hazardous part of what is commonly called “soot”, from 2010 to 2020. In the same period, local monitoring showed only a 6% reduction in Seattle, no change in San Francisco, and a 10% increase in downtown LA.

To be fair, that comparison reveals nothing about the absolute position of those particular locations, or whether 2010 and 2020 are representative years. It’s true that 2020 was a record-breaking year for wildfires out west. An estimated 10.1 million acres of land burned nationwide, compared to the five-year average of 7.8 million acres, and 94% of those 10.1 million acres were in the West. But, that five-year average is more than doubling the annual average from the nineties, so even though 2020 was particularly high, it shouldn’t be considered an aberration.

The average annual U.S. acreage burned in wildfires from 2016 to 2020 was more than double the annual average from the 90’s, and 94% of the acres burned in 2020 were in the West.

Using the EPA’s public data, we’ve gathered 20 years of particulate matter monitoring data for Seattle, San Francisco, and LA to investigate if the air in the West Coast is actually getting dirtier.

Air Quality Index (AQI)

Under the Clean Air Act of 1970, the EPA defines air quality standards and regulates pollutant emissions to protect public health. Of those pollutants, particulate matter is one of two pollutants, along with ground-level ozone, of greatest public health concern. Particulate matter concentration is generally the critical driver of alerts related to wildfire smoke.

Studies associate short-term particulate matter exposure with health effects including increased blood pressure and risk of heart attack for sensitive populations, while long-term exposure has been linked to reduced life expectancy. However, most non-specialists can’t stay on top of the latest science around health hazards at varying levels of particulate matter, which is where the Air Quality Index comes in.

Not all particulates are created equal. Particulate matter varies in size, so particulate matter concentrations are reported with a maximum particle diameter in microns, or millionths of a meter. For example, PM₂.₅ concentration is reported as the mass of particles smaller than 2.5 microns per unit volume of air. Over time, studies have increased our understanding of the hazardous effects of smaller particles. This evolving understanding has driven EPA regulations to focus more on fine PM₂.₅ particles and lower exposure limits.

EPA air pollution limits track with evolving scientific findings which have driven the particulate matter limits for public health protection towards focusing on lower levels of increasingly small particles.

To keep the public informed amidst science and exposure limits that change over time, the EPA maps each individual pollutant to evolving Air Quality Indices (AQI) that maintain easy-to-remember thresholds corresponding to comparable levels of hazards.

The AQI function is defined separately for each pollutant so that a particular AQI number is representative of similar levels of hazard across different pollutants. The AQI function is also updated over time as we learn more about the hazards of different pollutants.

For example, the physical concentrations of PM₂.₅ and PM10 have different mappings to AQI values. The AQI for PM₂.₅ escalates more quickly than PM₁₀ at lower concentrations based on research showing greater health hazards associated with fine particles. These thresholds are useful in determining what behaviors are safe or unsafe and for whom in a 24-hour period. The EPA also defines pollution compliance limits for local governments, with an AQI of 50 corresponding to the mean annual concentration limit while an AQI of 100 aligns with the 24-hour limit.

National and local trends

The EPA requires metro areas with 350,000 residents or more to measure and report air quality daily. This reporting generates high-quality, though geographically sparse, public data for tracking air quality in more-populated areas.

To start, we selected three monitoring sites from each of three major West Coast metros — Seattle, the Bay Area, and LA — and compared mean annual PM₂.₅ to both the EPA limit and the EPA-computed national average.

Seattle, the Bay Area, and Los Angeles all have different levels of PM2.5 compared to the national average. All three west coast metros appear to have non-improving, and in some cases worsening, PM2.5 concentrations over the past five years.

In general, we see that Seattle has had the lowest mean annual PM₂.₅ concentration, followed by the Bay Area. LA is the highest and tends to be far above the national average. The elevated air pollution in LA is well-known and mostly attributed to a mix of vehicle emissions, unfavorable geography, and warm climate.

We also see that the national average PM₂.₅ concentration has been steadily declining over the past twenty years, although the rate of decline may be slowing. In contrast, mean annual PM₂.₅ concentrations in the West Coast metros generally plateaued much earlier around 2010, with the Bay Area possibly even increasing in the last five years.

In the mean annual sense, the data indicates that the air may be getting dirtier in the Bay Area and possibly no longer improving in Seattle and LA. However, mean statistics don’t tell the whole story.

What the mean, means

Data analysts tend to think more in terms of distributions than single statistics like the mean. The mean provides limited information and is sensitive to extreme values. A higher mean can indicate many points being a little higher or a few points being much higher.

To dig into the recent high mean annual concentrations in the Bay Area, we looked at the daily PM₂.₅ concentrations across the most recent years.

Wildfire events drive particularly dramatic spikes in PM2.5 concentrations on the relatively small number of heavily affected days. Of the past three years, 2018 and 2020 were the worst for wildfire smoke in the Bay Area.

The mean annual values for 2018 and 2020, where we previously noted spikes, are influenced by clusters of particularly bad days that align with major wildfire events.

The scale makes it hard to judge the level of typical days, but most days appear to be roughly evenly distributed around the five-year daily averages shown in black. As an example, we can zoom in on the more “typical” range of days for 2018, the worst recent year, by truncating the scale at the 90th percentile day.

Focusing on the lower 90% of more “typical” days, the distribution of daily PM2.5 concentrations in the Bay Area in 2018 appears to be reasonably balanced around the five year daily mean.

We can’t precisely eyeball the data in this representation, but it does appear that most days below the 90th percentile are roughly evenly distributed around the daily mean from the last five years. This seems to indicate that the more typical days, with typical meaning not in the worst 10% of days, don’t appear to be higher in the worst recent year. The implication is that the spike in mean annual PM₂.₅ in 2018 likely is driven by the fewer wildfire affected days being particularly bad as opposed to a worsening of air quality on the majority of more typical days.

Digging into distributions

We’ve shown how mean annual PM₂.₅ is trending across sites and started digging into what is causing spikes in the Bay Area in recent years. We can use one further representation to try and better understand whether, and in what sense, air quality is getting worse.

Below, we show the median, quartiles, and 5th to 95th percentile ranges for each location over the last decade compared to the EPA annual limit.

Looking at percentiles over time, most of the movement in recent years is int he upper quartile and 95th percentile while the majority of days have remains fairly level in most locations.

While the patterns vary by location, in general, the lower 50% of days is holding level in most locations, while the upper quartiles are spiking in some locations, in some years. These observations further indicate that, in general, PM₂.₅ is not getting worse on the majority of days and that instead the worst days appear to be getting worse.

The Upshot

So is West Coast air getting dirtier? We only looked at a handful of locations, but we think that by some definitions, in some of the analyzed locations, the answer is yes.

Mean annual PM₂.₅ concentration appears to be level, and possibly increasing, across locations in Seattle, the Bay Area, and LA in recent years while the national average has been declining. And, it appears that the bad days associated with wildfires are likely contributing to worsening, or at least non-improving, mean annual air quality. However, we did not find evidence that the majority of days are getting worse.

There is a practical silver lining to these findings. Air quality is reported, and often forecasted, daily. It is more practical to reduce activity, and therefore PM₂.₅ exposure, on a limited number of very bad days compared to a large number of modestly bad days. Heeding daily AQI alerts may enable West Coast residents to mitigate the health risks of living with higher mean annual PM₂.₅.

This is an example of public data at its best: a coordinated effort across geographies to collect information that lets people make better decisions.

Your Turn

There are many questions that could be investigated using this public data:

• What about Portland, San Diego, or a broader sample of smaller West Coast metros?
• How does PM₂.₅ concentration vary with day of the week and hour of the day?
• How is 2021 year-to-date shaping up so far compared to other years?

For these, and other data projects, bit.io gives you the fastest way to get your data into a hosted database where you can keep it clean and up to date. Sign up, then check out our growing collection of guides on pipelining data into a database for your next project.

We also have a guide in the works, planned for publication on September 7, showing how to build your own home air-quality monitor and log measurements to a cloud database using a few lines of Python code. Please stay tuned to the Inner Join!

Interested in future Inner Join publications and related bit.io data content? Please consider subscribing to our weekly newsletter.