US Extreme Weather and Climate Change Dashboard

A project by Roger Pielke Jr., author of The Honest Broker

Check out the Global Dashboard
What's New

What's New

Everything added to this dashboard since it launched on July 29, 2026, newest first. Weekly data refreshes and the daily satellite image run on a schedule, so this page skips them.

    • The Winter Storms page now links to Ryan Maue's Storm Atlas, where he maps every cyclone on Earth since 1979 with its central pressure — a measure of storm strength, where the page's snowfall index covers impact only. His counts span the whole globe and come from ERA5 reanalysis, so they stay out of the US trend charts.
  2. Headline change
    • Meteorological drought now holds the drought tile on the homepage. That tile used to read the US Drought Monitor's D2+ area, and the USDM record starts in 2000 — under the 30-year minimum this site requires before it will fit a trend at all. So the one drought verdict on the top page rendered no verdict. The SPI series, a precipitation deficit measured against the 1895–present station record, takes its place.
    • The tile now states a finding: no detected change over 131 years, p=0.289. Nothing about the measurement moved — that verdict already sat on the Drought page. The top page now leads with a measure whose record supports the test.
    • SPI holds the tile rather than the SPEI index beside it for one reason: SPI needs no potential-evapotranspiration formula, and Xu et al. (2026) report that formulas without a land-atmosphere coupling constraint inflate drying trends at least sixfold. The headline drought verdict now rests on none of them. Both panels stay on the Drought page, along with the USDM composite and its record-too-short note.
  3. New page
    • A new page separates two questions this site had answered as one. Detection and Time of Emergence sets out what a "detected change" verdict here does and does not say, and how it differs from the IPCC's emergence assessments. It quotes all four AR6 glossary definitions in full, and one of them settles how the terms relate: AR6 defines time of emergence as the moment a signal "is statistically detected to emerge from the background noise of natural climate variability". The time of emergence is the date on which detection succeeds against natural variability.
    • AR6's emergence table now sits on the site, and almost every cell in it stands empty. Table 12.12 assesses 33 climatic impact-drivers across three time windows. Nine show an assessed emergence in the historical period, and of the 18 that name a discrete extreme event rather than a mean or a background state, two have emerged — extreme heat and cold spells. The table is global, so a coloured cell means a signal has emerged somewhere, not over the United States, and the page says so. The reproduction downloads as a PNG.
    • The site now publishes an emergence timescale for TXx. AR6 Figure 11.3(b) prints how far the annual hottest daily maximum rises per degree of global warming for each AR6 region: 1.5 °C per °C in Western and Eastern North America, 1.6 in Central North America. At the 0.2 °C per decade rate SR1.5 assesses, detecting that trend would take 26 years of the station record, which holds 130, and 21 years of the NOAA-homogenized record, which holds 74. Both records run several times longer than the trend would need, and the two series still render opposite verdicts — a detected decrease in the stations, a detected increase in the homogenized grid. Too little data cannot explain that, and AR6's own footnote puts Central North America among the regions with "low agreement across observation datasets".
    • SPI added: a drought measure that depends on no evaporative-demand formula. The Drought page now tracks the Standardized Precipitation Index from the same HCN station record the SPEI series uses, standardized the same way, with the evaporative-demand term set aside. It shows no detected change nationally, agreeing with the other three measures that have a long enough record.
    • What SPI adds that a fourth chart would not. PDSI builds in a Thornthwaite-family evapotranspiration formula, SPEI uses Hargreaves, and the CPC soil-moisture series comes from a model driven by one — so until today every meteorological and agricultural measure here rested on a formula of that kind. Xu et al. (2026, Communications Earth & Environment 7:726) report that formulas lacking a land-atmosphere coupling constraint inflate drying trends at least sixfold, with precipitation supplying 75% of the drought trend under the constraint. Only SPI answers that question.
    • The SPEI note now states the reach of its robustness check. That series tests PDSI's drying trend against Milly & Dunne's (2017) critique of PDSI's formula, and the note called Hargreaves "more defensible" for it. Hargreaves escapes PDSI's particular formula and not the wider critique, so the note marks the series an upper bound on the drying signal.
    • Cell area now weights the soil-moisture aggregation. CPC's grid runs regular in latitude and longitude, so a cell at 49°N covers 0.73 times the area of one at 25°N; the published unit already read "% of CONUS land area" while the aggregation averaged cells. The correction moves the D2+ period change from +3.675 to +3.818 points and changes no verdict; it closes the gap between the label and the arithmetic. The sibling global site found the same defect on its 5° grid; there the latitude span runs far wider, and the correction moved three verdicts.
    • A five-measure summary table now opens the Drought page, so the agreement or disagreement across measures reads at a glance rather than from four separate panels.
    • The Drought page gained an agricultural measure. It now tracks the share of the contiguous US in a soil-moisture deficit, from NOAA CPC's Monthly Soil Moisture record (1948–present, 3,308 land cells). It reproduces the canonical US droughts without tuning: 1956 and 1954 lead, then 2021, 2002, 1977, 2022 and 2012.
    • Four independent drought measures, one page. Soil moisture completed the set alongside the operational US Drought Monitor composite, the SPEI rainfall-minus-evaporation deficit, and hydrologic low flow. Each measures a different deficit; none separates a trend from the year-to-year noise.
    • Soil moisture shows the sharpest start-date sensitivity on the site — the slope measures +0.04 percentage points per decade from 1948 but +1.96 from 1990, because the two driest years in the record sit in the 1950s. The panel keeps the full record for exactly that reason, and the Methodology page and PDF now cover the new variable.
    • Corrections to the Detection & Attribution page. Checking it against the new variable turned up several claims the data no longer supported: the page described warm nights (p=0.102), heavy precipitation extremes and the MTBS large-fire count as detected increases where the site renders none, and two quoted p-values had moved with the weekly refresh. All now read correctly, and Cold Extremes gained the AR6 row missing since August.
    • Trend lines suppressed at 30 years or fewer. A record that short no longer gets a fitted trend line, matching the WMO's climate-normal period length. The Every Variable table marks each affected series "time series too short" rather than rendering a detection judgment.
    • The Methodology page gained a Data Sources — Attribution & Licensing section naming every provider, the citation each one asks for, and its license terms.
    • The Global Extreme Weather Dashboard launched, and the header of every page here now links to it. It applies this site's detection framework to 53 variables across every continent.
    • A link to this site now unfurls with a title, description and image on social platforms.
    • New Cold Extremes page: four raw US Historical Climatology Network metrics plus the homogenized nClimGrid cold-spell duration index and annual minimum temperature. Cold extremes also joined the Year in Progress box on the homepage.
    • Trend tests now correct for autocorrelation, and every variable reports how many years of record its noise level would require before any trend could emerge at all (the Weatherhead diagnostic).
    • The Every Variable table became sortable by any column, and the homepage gained a daily satellite view of the contiguous US.
    • The full variable table moved off the homepage and onto a dedicated Every Variable tab.
  10. Launch
    • The dashboard went live at extremeweather.thehonestbroker.org with seven phenomena pages — heat waves, severe convective storms, winter storms, flooding, drought, wildfire and tropical cyclones — each drawn from its official source and tested against the IPCC's detection threshold.
    • Alongside them at launch: Loss Normalization, setting hurricane, tornado and flood losses against growth in population and wealth; Detection & Attribution, placing the NCA4, AR6 and NCA5 conclusions side by side hazard by hazard; and the Methodology page and Bibliography.