Visualize This: Mapping Regions Polluted by Rigged VW Diesels

I’ve written about the challenges and opportunities in visualizing diffuse public health threats in new ways. Here’s a new effort, created by Kevala Analyticsfor the California Clean Energy Fund. Kevala’s blog post is invaluable.*

This interactive map, using car registration information and other data, provides Californians with a statewide view of the hidden and illegal air pollution from the 193,000 Volkswagen diesels sold in the state between 2009 and this year with rigged software masking high emissions. The Volkswagens are represented by orange dots. You can click particular zip codes to see the busiest commuter routes. With another click you can add a layer showing an index of air pollution and poverty, which shows the outsize impact on poor communities, as the Greenlining Institute, a California nonprofit group focused on social and environmental equity, pointed out on its blog.*

Of course the problem is not restricted to California, with an estimated 482,000 such vehicles sold in the United States.
To get caught up on how this scandal was caught, read The science behind the Volkswagen emissions scandal in Nature. The public health impact has been estimated several ways, including in “Impact of the Volkswagen emissions control defeat device on U.S. public health,” a paper in Environmental Research Letters, and by the Upshot unit at The Times.

The Center for American Progress, examined the extent of the problem last month:

The EPA alleges that VW installed these defeat devices in certain model year 2009 to 2015 VW and Audi light-duty vehicles equipped with two-liter diesel engines, which amounts to approximately 480,000 vehicles in the United States alone. VW had marketed these cars as using clean diesel technology that allowed owners to meet strict pollution limits while also achieving a higher fuel economy. Understandably, many consumers feelcheated and duped.

While VW has announced that it will recall 8.5 million cars in Europe, it has not yet announced a plan in the United States to either buy back the defective cars or fix them so that they meet emissions standards on the road. However, it is clear that—as a matter of justice—consumers in the United States must be made whole when penalties and other remedies are determined.

But, hey, the company is “Working to Make Things Right,” as it says on its website offering packages of gift cards to vehicle owners.

It’ll take some time before residents of areas experiencing that extra pollution pulse from those “clean” diesels will see their compensation.

Postscript, Nov. 19, 4:00 p.m. | Noting some of the constructive questions posed in comments, Aram Shumavon, the chief executive and co-founder of Kevala, Inc., sent this note:

Thanks for the interest in our mapping work. We noticed some substantive questions in the comments section and wanted to address them directly, but first, many of these answered might become clearer if people read the accompanying blog post released at the same time as the tool in question.

1) Does the VW diesel distribution just reflect population density?

The penetration of non-compliant VWs does generally reflect population density. This should not be surprising – it’s hard to sell a lot of cars where there are not a lot of people. However, there is geographic variation in the density of VW diesel sales that doesn’t track directly with population density. Perhaps somewhat unscientifically it’s been our experience that a disproportionately large percentage of VW diesel owners are heavy researchers of vehicle performance and total benefit cost analysis as part of their car purchase decision making process. They trend toward the hard sciences, are well educated, relatively wealthy, often care about the environment, and have a tendency to have longer commutes that favor the high MPG of the VW diesels in the spreadsheets they use to evaluate car purchase options.

2) Are there a large number of VW diesel owners in the Channel islands?

No. This is an artifact of the ZIP Code Tabulation Area (ZCTA) the covers Santa Cruz and Anacapa islands along with portions of Ventura on the mainland (93001). Because we have represented diesel VWs by randomly scattering dots across each ZCTA, the relatively unpopulated areas are covered as well. We actually had fixing the Channel islands on our punch list before the tool’s release but didn’t get to it in time. We hope to fix it in the coming days.

3) Aren’t VW diesels mostly located in areas with a low pollution and poverty index?

They are, generally. The purpose of including commute patterns (revealed by mouse click/tap on the map) was expressly to emphasize that the effects of NOx and PM emissions are not limited to the areas where the cars are purchased, but can be localized to where the cars are driven. We limit the number of destination ZIP codes to the top five for visual clarity and to improve server response time. The long tails of these commutes patterns are worth looking at, but not well suited for a public facing, web-based tool.

4) Does the pollution and poverty index show who is harmed by VW diesels?

Not directly, but it is part of the story. The pollution and poverty index is a ranking of each census tract as scored by the CalEnviroScreen (CES 2.0) – it’s a combination of poverty, language isolation, and exposure to toxics in both air and water that was generated by the California Office of Environmental Health Hazard Assessment ( for more details). CES 2.0 a very well researched and documented environmental justice score that is designed to quantify risk of exposure to health hazards based on a number of factors. We simply ranked the applicable census tracts into top, middle, and bottom thirds based on the CES 2.0 score.

When combined with the distribution of VW diesels and probable commute patterns, we hope users will see that not all areas are equally affected. We were simply hoping to localize the areas and communities that have a higher probability of negative effects from exposure to these vehicles because that becomes a tool for measuring how to geographically target potential remedies. If one considers the a 25X multiplier on emissions (as has been suggested in some places) it would be as if VW sold almost 5 million cars in those affected communities (that would be more than 25% of the state’s passenger vehicles). Any explicit quantification of that harm is something we’re interested in looking at, but beyond the scope of this public facing tool.