LGBTQ+ gayborhoods and gay-friendly U.S. cities, compared on air quality (current + 2050s), wildfire-smoke season, ND-GAIN climate risk & readiness, and LGBTQ+ healthcare infrastructure.
Real EPA county-level moderate-or-worse air days per year (AQI ≥ 51, 2019–2023 average), split by cause. Regular blue = days not driven by wildfire smoke (ozone, traffic, industry). Light blue = days driven by smoke. Green = projected additional smoke days by the 2050s. Sorted cleanest (top) to worst.
Total = EPA AQS county "Days with AQI" minus "Good Days" (AQI ≥ 51), 2019–2023 average. The smoke vs non-smoke split is modeled, constrained to sum to the measured total.
Assumes: the county's monitor represents the gayborhood inside it (large counties hide big intra-county swings); 2019–2023 — which included the historically severe 2020–2021 Western smoke years — is a stable "current" baseline; and 2050s smoke grows by a single flat regional multiplier (≈2.7× West / 2.5× NE-Midwest / 2.2× South) that ignores elevation, prevailing wind and local fuel load. The green 2050s bars are the most heuristic element on this chart; treat them as scenario, not forecast.
The same EPA signal, broken out by calendar month: average days per month with AQI ≥ 51 (moderate or worse), from daily EPA AQS data. Each cell is the long-run monthly average for that city's metro. Rows sorted by annual total (worst air at top). This is measured data — it shows when in the year each place's air turns, before any 2050s projection.
Average days with AQI ≥ 51 per month, EPA AQS daily data (multi-year average, 2015–2024). City rows map to their metro CBSA; same-metro cities share values. Measured, not modeled.
Assumes: every city in a metro has identical air (e.g. all LA-basin cities carry one shared row, erasing coastal-vs-inland differences); and the 2015–2024 average is treated as stationary — it smooths over the very smoke-season lengthening flagged below, and a month averaging "3 bad days" may be 0 in most years and 30 in one fire year. A different window (2015–2024) than the chart above (2019–2023), so the two air sections aren't on the same baseline.
Smoke risk rolls across the calendar by region: late-winter grass fires in the southern plains, spring dry season in the Southeast, the pre-monsoon Southwest peak, summer Canadian-smoke transport into the Northeast/Midwest, late-summer interior-West fire, and fall Santa-Ana fires in Southern California. Darker = higher smoke probability.
Generalized fire-season climatology (NIFC seasonal outlooks). Climate change is lengthening these windows at both ends.
Assumes: a single hand-assigned 0–4 intensity per region-month — qualitative expert judgment, not derived from (or cross-checked against) the EPA smoke data in §1/§1a. Every city in a region shares one curve, so there is no sub-regional resolution, and the "windows are lengthening" note is asserted from the literature, not shown in these cells.
Real ND-GAIN Urban Adaptation Assessment scores: Risk (x — lower risk to the right) and Readiness (y — higher up). Top-right is best positioned: low risk, high capacity to adapt. Hover or tap a point — or use the menus.
Dashed lines = cohort medians. Scores are ND-GAIN UAA Overall Risk & Readiness indices ×100 (2018 release). ~85 of the cities here have a UAA record.
Assumes: 2018 scores still describe a 2026+ decision (readiness shifts with funding and politics); and that the smaller havens with no UAA record — Fire Island, Guerneville, Ogunquit and others — can simply drop out, which biases this chart toward larger metros. Quadrant labels are relative to this curated cohort's medians, not an absolute or national standard: change the city list and a city's quadrant can change.
A 0–100 score: 50% estimated local LGBTQ+ clinical infrastructure + 50% real state policy from the Movement Advancement Project (gender-affirming-care access, insurance protections, bans). Cities in hostile-policy states score down even with strong clinics. Best at top.
Assumes: a 50/50 weight between clinics and state policy — an editorial choice that is what makes hostile-state cities score down, and the score moves if you reweight it. The clinical half is an unpublished local heuristic (no clinic count or named provider is exposed), so roughly half of every score is a hand estimate. State policy is applied uniformly statewide (Austin and rural Texas score alike, ignoring municipal/sanctuary protections), is a point-in-time MAP snapshot, and is scored as compensatory — strong clinics can offset a legal ban, when in reality a ban is often a hard floor, not a tradeable quantity.
Approximate typical warm-season feels-like temperature (°F) by 2050 — air temperature plus humidity. A moderate (RCP4.5-ish, ~+2.5°F over current normals) planning estimate from regional climatology, not a formal downscaled model output.
Each line is a city's seasonal curve, colored by region. Hover a line to read it, pick a city to highlight it, or highlight a whole region.
Feels-like = NWS heat index (temperature + relative humidity). Peak is the Jul/Aug single-number estimate; the curve interpolates May–Sep with a regional seasonal amplitude. Coastal and high-altitude cities swing less; interior and desert cities swing more.
Assumes: a uniform +2.5°F applied everywhere, though real warming is faster in interiors and at high latitudes; a single RCP4.5-ish scenario with no uncertainty band or high-emissions case shown; humidity that scales with temperature in a fixed regional way (future humidity is itself uncertain and can move the other way); and an afternoon peak that shifts by the same amount as the seasonal mean. The five monthly values are generated from one Jul/Aug peak plus a synthetic amplitude — the curve shape is interpolated, not observed.
The two big external exposures on one chart, for the ~85 cities with both scores. x = projected 2050s moderate+ air days (cleaner to the right). y = ND-GAIN climate risk (lower risk higher up). Top-right is the sweet spot: clean future air and low climate risk. Hover or tap a point.
Dashed lines = cohort medians. Points are hollow because the x-axis (2050s air) is a modeled projection. Air totals: measured today (EPA) + modeled 2050s smoke. Risk: real ND-GAIN UAA.
Assumes: everything inherited from §1 (air baseline + flat smoke multipliers) and §2 (2018 readiness, dropped small havens). Vintages are mixed on one chart — 2019–2023 air, a 2050s air projection, and 2018 risk plotted as if same-era — and the ~85-city intersection again excludes the smaller havens. Quadrants are cohort-relative medians, not absolute thresholds.
The cities with all four of air, ND-GAIN risk, healthcare and 2050 heat data, scored on a 0–100 balance score — the average of four min-max-normalized sub-scores (2050s air, climate risk, LGBTQ+ healthcare, Jul/Aug heat). Unlike a rank-average, this preserves magnitude: a runaway lead on an axis counts more than a hair's-breadth one. Higher = better all-round. Top 12 shown. Cities projected to exceed 200 moderate+ (AQI≥51) air days/yr are removed. A handful of smaller havens with no ND-GAIN record are added on a clearly-flagged ◇ estimated climate risk, so they aren't excluded just for being small.
Balance score = mean of four 0–100 normalized sub-scores within the qualifying cohort. Heat = estimated 2050 Jul/Aug feels-like (cooler = better, ▦ modeled). Every upstream estimate (flat +2.5°F warming, smoke multipliers, the ½-estimated healthcare score) compounds into this one number — treat it as a starting point, not a verdict.
What this scoring does and doesn't fix. Two earlier flaws are now addressed: magnitude is preserved (normalized values, not averaged ranks), and a hostile-policy state is a hard floor — disqualified, not averaged back up. What remains a judgment call: the equal default weighting is still an editorial choice (the "if healthcare counted double" line on each card shows how much it moves things), and normalization is cohort-relative — a "100" means best within this qualifying set (the cities that clear the floor and the air cap), not in absolute terms. The "all four data points" gate still favors larger metros; the cities it can't score are now named above rather than silently dropped.