Georgia’s COVID-19 Maps: Bad Faith or Bad Design?

In How to Lie with Maps, Mark Monmonier warns that map readers “must watch out for statistical maps carefully contrived to prove the points of self-promoting scientists, manipulating politicians, misleading advertisers, and other propagandists. Meanwhile, this is an area in which the widespread use of mapping software has made unintentional cartographic self-deception inevitable.”1

So which of these two scenarios—careful contrivance or unintentional self-deception—is at play on the Georgia Department of Public Health’s COVID-19 daily status report page?

Twitter user @andishehnouraee notes the difference in scale between two county-by-county COVID-19 maps of Georgia. The earlier map maxes out at 4,661 cases per 100,000, the later (and as of this writing, current) map maxes out at 5,165 cases per 100,000. As they point out, there has been a 49 percent rise in total COVID-19 cases between the two maps, but you wouldn’t know it at a glance, because the scales have changed in the meantime.

Is this, as @andishehnouraee suggests, a concerted attempt to hide the severity of the outbreak in Georgia—or, as T. J. Jankun-Kelly thinks might be the case, something that happens when you max out the old scale. In other words: bad faith or bad design? (Or both: it can be both.)

Update 19 Jul: See Twitter threads from Darrell Fuhriman and Jon Schwabish disagreeing with critiques of the Georgia Public Health maps. It’s worth clarifying that only one map is ever viewable at the website: the map’s scale has changed over time, but it’s not like they’re side-by-side except in @andishehnouraee’s tweet.

Update #2: See Jon Schwabish’s blog post critiquing the data visualization critique in more detail.

Mapping Mask Wearing in the United States

The New York Times (screenshot)

Wearing a mask in public is increasingly being encouraged or required as a measure to slow the spread of COVID-19. The New York Times maps the rate of mask wearing in the United States. The county-level map is based on more than 250,000 responses to a survey conducted in early July, in which interviewees were asked how often they wore a mask in public.

The map shows broad regional patterns: Mask use is high in the Northeast and the West, and lower in the Plains and parts of the South. But it also shows many fine-grained local differences. Masks are widely worn in the District of Columbia, but there are sections of the suburbs in both Maryland and Virginia where norms seem to be different. In St. Louis and its western suburbs, mask use seems to be high. But across the Missouri River, it falls.

[MAPS-L]

Mapping COVID-19 Exposure Risk at Events

Screenshot

The COVID-19 Event Risk Assessment Planning Tool is a county-by-county map of the U.S. that shows the risk of coming into contact with a COVID-positive individual at an event. “This site provides interactive context to assess the risk that one or more individuals infected with COVID-19 are present in an event of various sizes. The model is simple, intentionally so, and provided some context for the rationale to halt large gatherings in early-mid March and newly relevant context for considering when and how to re-open.” A slider changes the size of the event; risk goes up dramatically with bigger events, of course. Which you’d think would be intuitively obvious. You’d really think so, wouldn’t you. [Cartophilia]

A County-by-County COVID-19 Map

Screenshot

COVD-19 is hitting the United States very hard right now. This interactive map from the Harvard Global Health Institute measures COVID-19 risk at the county level. The four colour-coded risk levels are based on a seven-day rolling average of new COVID-19 cases per 100,000 people: less than one means green (“on track for containment”); more than 25 means red (“tipping point”). It’s explained here. [Matthew Edney]

Pandemic Mapping and Posterity

The flurry of COVID-19 maps that have emerged in the first half of this year will be something that future cartographers and librarians will look back on, both in terms of historical records that need preserving, which is the subject of this CityLab interview with Library of Congress map librarian John Hessler, and in terms of best practices for disease mapping—what to do and what not to do when mapping a pandemic—which is the subject of this Financial Times video interview with Kenneth Field. (Both from early May; I’m playing catchup right now.)

Library of Congress Livestream on the History of Pandemic Maps

Tomorrow (23 April 2020), the Library of Congress will be livestreaming No One Was Immune: Mapping the Great Pandemics from Columbus to COVID-19, in which John Hessler and Marie Arana will “discuss the sweep of history from the 1500s smallpox pandemic that decimated the indigenous population of the Americas to the meticulous work that is being done now to map COVID-19.” To be streamed on the Library’s Facebook page and YouTube channel at 7 PM EDT. [WMS]

CityLab Wants Your Hand-Drawn Quarantine Maps

CityLab is asking readers to send them hand-made maps of their life under quarantine.

We’re inviting readers to draw a map of your life, community, or broader world as you experience it under coronavirus. Your map can be as straightforward or subjective as you wish. You might show key destinations, beloved neighbors, a new daily routine, the people or restaurants you miss, the future city you hope to see, or anything else that’s become important to you right now. It might even be a map of your indoor life. For an added challenge, try drawing from memory.

Deadline is 20 April, with a selection of submissions to be featured in a future article.

Prior art would include Fuller’s quarantine maps and Kera Till’s “Commuting in Corona Times” (previously).

Behind the Scenes at the JHU Coronavirus Dashboard

JHU coronavirus dashboard screenshot
JHU CSSE (screenshot)

ArcGIS-based dashboards tracking the spread of the novel coronavirus are now reasonably common, but the first was produced by Johns Hopkins University’s Center for Systems Science and Engineering. As Nature Index reports in this behind-the-scenes look at the JHU coronavirus dashboard, the decision to launch was spur of the moment, but now the dashboard and its underlying data get more than a billion hits every single day, and it is now managed by a team that numbers nearly two dozen. [GIS Lounge]

Mapping the Lockdown-Related Drop in Emissions

ESA

The European Space Agency maps the drop in nitrogen dioxide concentrations in the atmosphere in the wake of coronavirus lockdowns in many countries (see above). [GIS Lounge]

Meanwhile, CESBIO researcher Simon Gascoin built a map that compares NO2 concentrations over the last 30 days with the same period in 2019.

Data for these analyses generally come from the Copernicus Programme’s Sentinel-5P satellite. The Copernicus Atmosphere Monitoring Service issued a warning last week about using the data improperly.

Concentrations of NO2 in the atmosphere are highly variable in space and time: they typically vary by one order of magnitude within each day and quite substantially from one day to another because of the variations in emissions (for example the impacts of commuter traffic, weekdays and weekend days) as well as changes in the weather conditions. This is why, even if observations are available on a daily (currently available from satellites) or even hourly (ground-based observations) basis, it is necessary to acquire data for a substantial period of time in order to check that a statistically robust departure from normal conditions has emerged.

Cloud cover is a factor that needs to be taken into account as well.

Previously: Emissions Drop Due to Coronavirus Outbreak.

Still More Coronavirus Maps

Kera Till

Kera Till’s “Commuting in Corona Times” is a transit map of the new normal. More at Untapped New York.

On a personal level, the coronavirus map I stare at the most is the one closest to home: a dashboard that shows the regional incidence of COVID-19 in Quebec. Maintained by two geographers at Laval University, it’s extremely helpful in that it shows the per capita rate as well as the raw numbers, which highlights (for example) just how many cases there are in the Eastern Townships, and how few there are here in the Outaouais, as a percentage of the population. [Le Droit]

New York City COVID-19 mapLess helpful is New York City’s map showing the percentage of patients testing positive for COVID-19, because its neighbourhood detail is so difficult to interpret, as Patch’s Kathleen Culliton points out. “Neighborhoods are designated by numbers instead of name—408 is Jamaica, Queens, by the way—and the percentages are not connected to population data but to those tested. The number of people tested per zone? Not included. The population [per] zone? Not included.” [Kenneth Field]

It’s hard to maintain social distancing in a dense urban environment like New York, but that doesn’t mean that rural areas are inherently safer. Identifying areas that would be hit harder by the coronavirus can be a factor of age and various social vulnerability factors (such as poverty and vehicle access); John Nelson looks at the intersection of age and social vulnerability in this StoryMap and this blog post. The Washington Post’s maps of vulnerability are based on age and flu rates. A third example is Jvion’s COVID Community Vulnerability Map, which is based on anonymized health data from some 30 million Americans [ZDNet].

The New York Times maps the number of cases at the global level and for the United States. It’s also making available county-level coronavirus data assembled from various states and counties, since there seems to be no single agency tracking this at the national level.

Failing to observe social distancing makes the pandemic worse. You might have seeen Tectonix’s video on Twitter, drawn from the location data of mobile devices that were active at a single beach in Florida over spring break, and followed them home. As CTV News reports, the video has drawn fire from privacy advocates, though Tectonix asserts that the data was anonymized and collected with user consent. Meanwhile, the New York Times explores several scenarios of coronavirus spread, comparing what might happen with some control measures, more severe control measures, and no action taken at all.

For mapmakers: Matthew Edney on how and how not to map the COVID-19 pandemic. Kenneth Field on using coxcomb charts (memory-intenstive example here) and waffle grids to map the pandemic.

Fuller’s Quarantine Maps

Gareth Fuller: Quarantine Maps: Day 4
Gareth Fuller

Gareth Fuller, whom we first heard about thanks to his masterpiece London Town, is in the news again. Now based in Beijing, he found himself forced to self-quarantine for 14 days after returning from a trip to Kuala Lumpur. Fuller has mapped every place he has lived, so he spent his two weeks of isolation creating quarantine maps—one for each day. The maps are claustrophobic—his apartment is 590 square feet—metaphorical, even fantastical. They’re very much on-point in this age of self-isolation and social distancing. They’re available as a set of postcards for £14, which is considerably cheaper than his other limited edition prints.

Last fall Fuller released a tourist map of Pyongyang (BBC News). I seem to have missed his map of Beijing when it came out.

Previously: Fuller: London Town; Fuller Update.

Some More Coronavirus Maps

Washington Post

The Washington Post maps COVID-19 cases by U.S. state and country (above).

Animated maps from HealthMap and NBC New York trace the spread of the coronavirus over time. [Maps Mania]

Maps Mania also has a list of official government coronavirus maps.

Via Geography Realm, a collection of “responsible” live visualizations, some of which are maps, that can be embedded on other websites.

(See my other posts about COVID-19 for maps I’ve already linked to.)

Emissions Drop Due to Coronavirus Outbreak

Map of mean tropospheric NO2 density over China, January-February 2020
NASA Earth Observatory/Joshua Stevens

As you may have seen elsewhere, the coronavirus pandemic is having an impact on air pollution, as countries shut down human and economic activity in an attempt to deal with the outbreak. Take nitrogen dioxide. Tropospheric NO2 density decreased significantly over China between January and February, and the same seems to be happening in northern Italy, which normally has some of the most severe air pollution in Europe. See the ESA’s animation:

More broadly, try this online map, which compares NO2 emissions before and after 20 February 2020 anywhere on the planet. [Maps Mania]

Previously: Mapping Nitrogen Dioxide Pollution.