On October 29th, 2012, Hurricane Sandy's storm surge advanced mercilessly on New York City. Subway lines we inundated. City streets turned fluvial. Gawker's server host did its best impression of Niagara Falls. Shit got — as they say — real.

Here's the thing: risk analysts had accounted for a hurricane like Sandy (to the best of their abilities, at least). So how did their projections stack up against the real thing? John Nelson — a data visualization maven who specializes in portraying existential risk — has some answers in the form of jaw-dropping maps.

Disaster modeling is not an exact science. It doesn't matter if it's an earthquake, a tornado, a volcanic eruption or a tsunami — every natural disaster will come with its own set of ruinous quirks.


Those charged with predicting the effect of a hurricane on New York City must therefore assume what Nelson calls an impossibly generic storm. "Reality," he says, will inevitably "look somewhat different than the generalized risk zones." In his latest visualization, Nelson has created a map that highlights not just the variability of NYC's storm surge projections, but the instances where these models were more-or-less consistent with Sandy.

Here's how reality stacks up (click here for a ginormous version, or see below for interactive zooming):


Writes Nelson:

Following Hurricane Sandy, teams of FEMA investigators from their Modeling Task Force scoured New York City for signs of high water marks and combined those records with data collected by an array of USGS gauges to render a geographic delineation of the inundation zones of the Hurricane.

This map paints the surface footprint of every building in the city of New York (pre-Sandy) by where it stands in relation to modeled flood risk from a general storm surge of hurricanes, categories 1 - 4 (dark to bright blues), and by actual Hurricane Sandy highwater (red) painted atop.

Click and scroll on the image below to zoom in on specific regions of the city:

Nelson describes his design process in an email to io9 (featured below is Nelson's visualization separated out into two separate maps. The first illustrates risk projections, the second buildings actually affected by Sandy. Click here or here to see either map enlarged):

The process in this case was the discovery of really cool data that started leading me around to other cool data, which makes data visualization especially fun.

More broadly, it has been a meandering evolution of one project to another. I'd just worked on a set of pointillist maps showing the biking and walking commuters in the "top 10" bike-friendly US cities, and the New York section was especially interesting because of the sheer density of all the data.

So I got to wondering about other sets of NY data that might also be really data-heavy. I was inspired by Bill Rankin's Manhattan map of building heights as a proxy for land value, so I hunted around for the amazing set of building footprints in New York City, and found them at NYC Open Data.

We mentioned NYC Open Data yesterday in this post about what New Yorker's complain about. It's a positively massive repository of public data, generated by a variety of the city's agencies and organizations and made available for public use. It's ripe beyond ripe for data mining. Nelson continues:

I knew that the FEMA modeling team had recently released what they found to be the highwater lines of Hurricane Sandy's storm surge around the city (based on field checks and USGS gauges). I supposed that the building footprints would provide a salient canvas to pain in this inundation data because seeing the physical intersection of those things, rather than their layering in, would give readers a more direct means of finding the fate of a specific building in that storm, and for non-locals it might make a more firm connection between an incident on the news and actual places people live and work in.[Emphasis ours]

See more of Nelson's analysis and visualizations on his blog.