Has anyone standardized a reproducible workflow for mapping urban heat inequity with open data? I’m prototyping in QGIS 3.34 using Google Earth Engine LST and the latest ACS 5-year estimates to support a cooling grant review next month, and I’d love examples, pitfalls, or trusted neighborhood-level validation sources. Happy to share our PostGIS schema and notebook if that helps.
In QGIS 3.34 with GEE LST, I’d take a July–Aug daytime median, cloud-mask, and run population‑weighted zonal stats at block‑group in PostGIS — , LST ≠ air temp, so don’t treat it like it is. > to support a cooling grant review next month, and I’d love examples, pitfalls, or trusted Agree on the timing — use NOAA Heat Watch/CAPA campaign data (https://heat.gov/pages/heat-island-mapping) for neighborhood validation against your LST percentiles. If no Heat Watch nearby, pair ASOS/Mesonet stations with LCZs (WUDAPT) to sanity‑check — what city are you targeting?
But i’ve gotten cleaner equity signals by masking LST to residential areas before any zonal stats — intersect block groups with building footprints or residential parcels, area‑weight those pixels, then apply your ACS 5‑year weights. For the “cooling grant review next month,” this dampens the pull of large parking/industrial heat; if parcels aren’t open, MS US Building Footprints are a solid stand‑in: GitHub - microsoft/USBuildingFootprints: Computer generated building footprints for the United States. Do you have local land‑use or parcel data to drive that mask?
One tweak that’s helped me: weight the GEE LST by daytime population using LEHD LODES before aggregating, and carve out freeways/water so the equity signal isn’t diluted; for the ACS 5‑yr I’ll blend in SVI components only after that… For neighborhood validation, a quick pass is state mesonet + a couple Weather Underground PWS on the same July–Aug afternoons — , QC is messy, but it flags obvious outliers. If useful, the LODES data are here: Data - Longitudinal Employer-Household Dynamics.