North Is a Choice | Zhanchao Yang

Plate I · A copy, on purpose

Heat Islands in Philadelphia

The Times mapped Baltimore in 2019. I rebuilt it for Philadelphia to see how much of what works is craft and how much is data. Mostly craft.

Shows: Surface heat, one summer afternoon
Data: ACS 2022, TIGER/Line
Color: Teal to rust, neutral at 93°F
Made in: R

Map of Philadelphia showing estimated land surface temperature on a summer afternoon, with cool teal areas near Wissahickon Valley and hot rust-colored areas in South and West Philadelphia. — Same scale, same two annotations, same cooler and hotter bar.

Reference graphic: The New York Times 2019 map of heat islands in Baltimore, showing cooler areas near parks and hotter areas east of downtown. — *“Summer in the City Is Hot, but Some Neighborhoods Suffer More”*, The New York Times, August 2019, by Nadja Popovich and Christopher Flavelle. Shown here as the reference for a study exercise. All rights remain with the publisher.

What I kept

A diverging ramp with a real middle.
Two annotations, each ending in a temperature.
Neighborhood names on the map, small, in caps.
No basemap, no north arrow, no scale bar.

What I could not match

Theirs is satellite heat. Mine is a density proxy.
They draw the street grid inside the surface.
Their labels were placed by hand, and it shows.
Theirs sits in a city. Mine floats on the page.

My first try used a plain ramp and the whole city looked warm. True, and empty. A diverging scale makes you say what counts as normal.

Then the parks and the row homes pulled apart on their own. Twenty degrees, one afternoon, two blocks.

Plate II · Where the breaks go

Inland California bears the heat

County heat risk in five equal tiers. Two choices carry it: a projection that keeps areas true, and breaks that follow the data.

Data: FEMA National Risk Index
Projection: California Albers, EPSG:3310
Classes: Quantile, five
Color: Magma, light to dark

Choropleth map of California counties shaded by heat wave risk in five quantile tiers, with the Central Valley and Inland Empire in the darkest highest-risk class. — FEMA National Risk Index. California Albers Equal Area, five quantile classes.

San Bernardino alone could swallow a few states, so in Mercator the desert wins every map it turns up on. Albers keeps areas honest.

Which means the dark band down the Central Valley is real. Thin vegetation, low ground, trapped heat.

Plate III · Turning the world on its pole

Three expeditions. Three fates. One continent.

Amundsen got there first. Scott arrived thirty-three days later and died walking home. Shackleton never reached the pole and lost nobody.

Projection: Polar stereographic, EPSG:3031
Data: Natural Earth, expedition records
Color: One route, one outcome
Ground: Dark, so the ice reads first

Polar stereographic map of Antarctica on a dark navy background, showing the Amundsen, Scott, and Shackleton expedition routes in cyan, red, and orange, with research stations marked. — Antarctic Polar Stereographic. Natural Earth and historic expedition records.

In Mercator the South Pole is not a place, it is the whole bottom edge of the world. Turn the globe onto its pole and the smear becomes a continent with a middle. Only then do three routes converge.

Ten nations, one continent, no owner. A century on, nobody owns the ice.

Plate IV · Live

Two variables, one surface

Two distributions in one color, so you can see where they disagree. Also the map most likely to be misread, which is why it helps to hover.

Open the map