earthflux
Open the lab

MULTI-HAZARD CLIMATE INDICES · OPEN SCIENCE

Follow drought as it becomes fire, flood and energy stress.

Drought, wildfire, flood-potential and energy indices. We are working toward one shared calendar and one probability scale, so the cross-hazard figures are synthetic for now. Each value traces back to its source data, its method and its validation evidence.

Open the Scenario LabSee the evidence

Built on climate_indices, a BSD-3-Clause library forked from code that NIDIS and NOAA NCEI developed.

DROUGHT · SPEI · 9-MONTH

Standardized Precipitation Evapotranspiration Index, contiguous U.S., September 2026

Preview — Domain Review pending

September 2026preliminary

Map of the contiguous United States showing 9-month SPEI for September 2026 on a grid of half-degree cells. Drought classes in gold cover most of the West, the Plains, the Gulf states and the Southeast. Wet classes in blue are limited to a few patches around the Great Lakes and the Ohio Valley. The Midwest and the central Appalachians are mostly near normal.

Until the interactive map has loaded, its controls stay off and the map shows the latest month.

Drier · extremely dry ≤ −2.0Near normalWetter · extremely wet ≥ 2.0

Each cell is weighted by the cosine of its latitude. SPEI is computed from inputs averaged to 0.5° cells. September 2026 is a preliminary month, so NOAA may still revise it. Source: NOAA nClimGrid-Monthly. Modified data; not endorsed by NOAA.

THE RECIPE A PUBLISHED MAP CARRIESdata: nclimgrid-monthly · method: spei · log-logistic · thornthwaite · 9mo · calibration: 1991-2020 · engine: climate_indices 3.2.0 · evidence: implementation parity

Hazards connect. Drought is often where it starts.

Drought dries fuels, sets up the swing to flood potential when it breaks, and arrives with the heat that raises cooling demand. We start from drought and follow it outward, using one set of conventions for every hazard so the links can be measured.

THE HUB

Drought

SPI, SPEI (gamma, Pearson III, log-logistic), EDDI, the Palmer family (PDSI, PHDI, PMDI, Z-Index) and self-calibrated PDSI, with potential evapotranspiration by Thornthwaite, Hargreaves or FAO-56 Penman–Monteith.

  • ILLUSTRATIVEIllustrative map of fire-weather percentile, highest across the Southwest and Plains
    FWI percentilelowhigh

    DROUGHT →

    Fire weather

    Dry fuels raise fire danger long before ignition. Tracked with KBDI, the Canadian Fire Weather Index system, Fosberg FFWI, Hot-Dry-Windy and Haines.

  • ILLUSTRATIVEIllustrative map of flood-potential percentile, highest across the Southeast and Northeast
    API percentilelowhigh

    DROUGHT →

    Flood potential

    Droughts often end abruptly. Effective precipitation, the Effective Drought Index, the Flood Index and the Antecedent Precipitation Index track how fast flood potential returns. Potential, not observed flooding.

  • ILLUSTRATIVEIllustrative map of cooling-degree-day percentile, highest across the South and Southwest
    Cooling degree days percentilelowhigh

    DROUGHT →

    Energy stress

    Hot, dry seasons raise cooling demand. Heating and cooling degree days, hub-height wind speed, wind and solar capacity factors, and standardized production, demand and residual-load indices.

Illustrative, synthetic values, not observations.

Drought month by month, and the hazards beside it.

With the interactive map loaded, selecting a cell on it shows that cell's real SPEI and SPI here, month by month. Fire weather, flood potential and energy stress have no published data yet, so the cascade below it is an invented place, synthetic for every hazard.

An invented place over five years

Illustrative, synthetic values, not observations.

  1. In this illustration, a drought builds over months. A short time scale shows it first, and a long one shows it later and holds it longer.
  2. In this illustration, fire-weather danger rises in the warm seasons inside the drought, as fuels dry.
  3. In this illustration, the drought ends abruptly, and flood potential returns soon after it breaks.

The evidence, stated plainly.

The four hazard families are not equally mature, and we say so. The library keeps a validation record for each index. The Registry reads it as one Evidence Kind per specification, at its weakest link, once a named reviewer has approved the reading. Where none is approved, we say that instead.

HazardRegistry readingWhat it rests on
DroughtRegression covered to implementation parityThe Scenario Lab's nine SPI and SPEI members, from SPI at regression covered to SPEI log-logistic at implementation parity. The library also compares EDDI with NOAA PSL reference series at one location, standard Palmer with NOAA nClimDiv for all 344 climate divisions (qualified), and SPI with NOAA NCEI climate-division SPI (Pearson III only).
Fire weatherNot yet reviewedThe Canadian FWI system matches the Natural Resources Canada reference implementation on one calibration sample. KBDI reproduces the Keetch and Byram (1968) worked example and is not independently validated. Hot-Dry-Windy reproduces one published case. Fosberg and Haines have no validation row yet.
Flood potentialNot yet reviewedTested against the published method definitions only. No external reference dataset has been adopted.
Energy stressExperimental to implementation parityCooling degree days reproduce NOAA CPC's published example, and heating and energy degree days are specification-level. Hub-height wind speed matches two independent implementations on synthetic inputs. No product-level external reference has been adopted.

Full record: VALIDATION.md in climate_indices

Questions we are working on.

Commitments that don’t change.

  1. 01

    climate_indices stays permissively open (BSD-3-Clause) and never requires Earthflux to compute.

  2. 02

    Every number we publish can be reproduced from climate_indices, public data and its recipe.

  3. 03

    Provenance and validation conventions are published as open specifications.

  4. 04

    Software releases and key derived datasets get DOIs and citation workflows.

  5. 05

    Uncertainty and validation status are shown, never collapsed into one confidence badge.

BUILT ON

climate_indices

Forked from code that NIDIS and NOAA NCEI developed. On GitHub since 2017 under BSD-3-Clause, and now covering drought, fire weather, flood potential and energy.

Public data

NOAA nClimGrid-Monthly for the Scenario Lab and the hero map, gridMET for the Lab's Hargreaves and Penman–Monteith members, and your own data through the library.

Review the science with us.

We are looking for drought, fire-weather, hydrology and energy-meteorology scientists to review methods and evidence, and for research and agency partners to shape what comes next.

Become a reviewerContribute to climate_indices