blog / how-to-overlay-the-cdc-social-vulnerability-index-with-a-blo

How to overlay the CDC social vulnerability index with a block-level heat map

An empty outdoor paved basketball court in full sun, with no shade or greenery nearby.

If you've pulled the CDC/ATSDR Social Vulnerability Index for your county and a land surface temperature layer for the same summer, you've probably hit the same wall every planner hits: the two files don't talk to each other. SVI comes packaged at the census tract level. Heat data, especially anything with real resolution, comes as a raster at 10 to 30 meter pixels or finer. You can't just drop one on top of the other and call it a composite map. This is the workflow that gets you a usable overlay.

Start with the mismatch, not the merge

The SVI has four themes, socioeconomic status, household composition and disability, minority status and language, and housing type and transportation, rolled up into an overall percentile ranking per tract. A tract can run a mile across and hold six or eight blocks with wildly different roof material, tree cover, and building age. Averaging heat across that whole tract erases the thing you're trying to find: the single block with the surface parking lot and the 1960s brick walkups that's running 8 to 10 degrees hotter than the leafy block two streets over, even though both sit in the same "high vulnerability" tract.

So the first real decision is which direction to resample. Most planning departments downsample the SVI, assigning the tract's percentile to every block group or block within it, rather than trying to upsample thermal pixels into some fake tract-wide average. You lose some nuance on the vulnerability side, but you keep the spatial precision on the heat side, which is the layer that varies block to block.

In practice:

  1. Pull the SVI shapefile or geodatabase from the CDC/ATSDR site for your state or county, at either tract or the newer county-adjusted tract level.
  2. Pull your surface temperature raster, converted to land surface temperature (LST) if your source gives you raw thermal bands.
  3. Reproject both layers to the same coordinate system. This step gets skipped more than it should and it's the source of half the "why doesn't this line up" tickets.
  4. Join the SVI attribute table to your block or block-group polygon layer using the tract FIPS code as the key, since the block inherits its parent tract's score.
  5. Run zonal statistics on the LST raster using those same block polygons, pulling mean or max temperature per block.
  6. Normalize both values, SVI percentile and block temperature, onto a common 0 to 1 or 0 to 100 scale so they can sit in the same symbology.

Building the overlay

Once both numbers live in the same attribute table, you've got a choice to make about how to show it. A bivariate choropleth, where color hue tracks temperature and color saturation or a second hue tracks vulnerability, is the standard approach and it reads well in a council presentation because the darkest, most saturated blocks are visually obvious without a legend essay. The alternative is a simple composite score, heat rank times vulnerability rank, which is easier to sort and rank in a spreadsheet but loses the ability to tell a reviewer "this block is hot but not vulnerable" versus "this block is vulnerable but not that hot."

Either way, keep the two source numbers in the table even after you build the composite. Someone on your team, or a council member, is going to ask why a specific block scored where it did, and "the SVI percentile was X, the mean LST was Y" is a much better answer than "the composite score was 0.74."

One more thing worth flagging before you present this: SVI updates every two years on a staggered release schedule, and heat data should be seasonal, pulled from the hottest weeks of the target summer rather than an annual average. If your two layers are three years apart in vintage, say so in the footnote. A reviewer who catches the gap themselves trusts the rest of the map less.

This is exactly the layering problem Heat Exposure Mapping is built around, pulling block-level thermal and demographic density together so the overlay is already done by the time it reaches your desk instead of being a GIS afternoon every budget cycle. If your team is doing this join by hand every season, it's worth seeing what a standing version of it looks like.

Start a pilot