The air we breathe
CO₂ at concentrations not seen in 3–5 million years.
The atmosphere is the ledger.
Earth Energy Imbalance · CERES EBAF-TOA
CURRENT
No hedging is possible here. This is a measurement, not a model.
The long-term trend is near the top of the 25-year satellite record.
The long-term trend is near the top of the 25-year satellite record.
Earth Energy Imbalance (EEI)
1.946 W/m²
12-month running mean
CURRENT POSITION IN RECORD
96th percentile · 25-year satellite record
7 consecutive months ≥ 1.5 W/m²
≈ 15.8 Hiroshima bombs / sec
of excess energy absorbed
992.5 TW
absorbed · 12-month running mean
Satellite record peak
2.219 W/m² · Jul 2023
ⓘ Positive values indicate energy gain by Earth.
Earth Energy Imbalance (EEI)
0.986 W/m²
12-month running mean
CURRENT POSITION IN RECORD
93th percentile · 25-year satellite record
≈ 4.0 Hiroshima bombs / sec
of excess energy absorbed by hemisphere
251.4 TW
absorbed · 12-month running mean
Satellite record peak
2.219 W/m² · Jul 2023
ⓘ Positive values indicate energy gain by Earth.
Earth Energy Imbalance (EEI)
2.905 W/m²
12-month running mean
CURRENT POSITION IN RECORD
97th percentile · 25-year satellite record
21 consecutive months ≥ 1.5 W/m²
≈ 11.8 Hiroshima bombs / sec
of excess energy absorbed by hemisphere
740.8 TW
absorbed · 12-month running mean
Satellite record peak
2.219 W/m² · Jul 2023
ⓘ Positive values indicate energy gain by Earth.
ⓘ methodology & sources
12-month running mean of global-mean top-of-atmosphere net flux (toa_net_all_mon) — the rate the planet is accumulating heat, ~90% of it into the oceans. The trend is the signal: the 12-mo mean has risen roughly +0.4 W/m² per decade since 2001, the trend line nearly tripling across the record — but the line is noisy and NOT monotonic (it swings year to year, and the single-month-window record high was mid-2023, above the current value). NOTE: EBAF's absolute level is constrained to in-situ ocean heat content (Loeb et al.); the satellite fixes the TREND well but not the absolute magnitude independently — which is why the climb, not the height, is the honest reading. The Hiroshima framing is an editorial transform of the W/m² value, computed live.
SOURCE · ceres_ebaf_eei
Daily Global Surface Temperature
· ERA5
CURRENT
16.50°C
Global AVG
+1.48°C
2026 Anomaly
35/249
calendar-day records in 2026
249/249
days above 1991–2020 daily mean
#2 / 87
today's rank among all years
189/249
days hotter than 2023
17.0°C
August 2026 avg — hottest August on record (87 yrs)
ⓘ methodology & sources
Cos-latitude-weighted global mean 2m air temperature from ERA5 reanalysis, published directly by ECMWF/C3S via Climate Pulse. ~2-3 day publication lag is intrinsic to the product, not staleness (most recent day is typically PRELIMINARY, settling to FINAL a few days later). The absolute annual peak is a Northern-Hemisphere-summer (July) value — ~17.15 °C in 2024, the warmest day in the record. The headline above tracks a different, honest superlative: the warmest reading on record for its specific calendar date, which 2026 has been setting through spring — weeks before that summer peak.
SOURCE · era5_daily_global_t
Atmospheric CO₂ · Mauna Loa
STALE
Pre-industrial baseline 280 ppm (1750)
427.13ppm
to CO₂ doubling
132.9 ppm remaining · 560 ppm target
year high (vs. current reading, above)
432.34 ppm
May 15, 2026 · seasonal cycle: CO₂ falls each NH summer
rate of rise
+2.23 ppm/yr
NOAA annual mean growth rate
above pre-industrial
+52.5%
Last seen at this level ~4 million years ago · Pliocene epoch
ⓘ methodology & sources
The breath of the planet: CO₂ falls each Northern-Hemisphere summer as plants take up carbon, then rises through winter. The full Keeling Curve — unbroken since 1958 — shows the relentless upward trend and the accelerating pace of rise.
SOURCE · noaa_co2_daily
Global Precipitable Water
CURRENT
Every +1°C, atmospheric moisture rises ~7%.More water vapor intensifies rainfall and drought extremes.
25.258 kg/m²
vs 1950–1980 baseline
+6.4%
baseline mean
23.7331 kg/m²
all-time record
25.2576 · 2026-08
N. mid-latitudes (30–60°N)
17.2142 kg/m²
vs 1950–1980 baseline
+9.6%
26.7457 kg/m²
vs 1950–1980 baseline
+7.3%
baseline mean
24.9252 kg/m²
all-time record
26.7457 · 2026-08
24.1365 kg/m²
vs 1950–1980 baseline
+5.4%
baseline mean
22.8988 kg/m²
all-time record
24.1365 · 2026-08
ⓘ methodology & sources
36-month running average of total-column precipitable water (water vapour), global mean. Atmospheric moisture rises ~7% per °C of warming (Clausius-Clapeyron), and the trend is upward — more water vapor means more intense precipitation, compounding flood and drought extremes. Per v2 exec summary Sec 1, which highlights the 30°N–60°N mid-latitude band (where over half the world's population lives); that exact band isn't available as a stable automated series, so this shows the global mean. Source: ERA5 total column water vapour (Copernicus C3S, 1940–present), area-weighted to a cos-latitude global mean via the CDS API.
SOURCE · era5_pwat
Global Mean Temperature · 146-Year Record
CURRENT
1.418°C above pre-industrial
Three independent methods agree within 0.10°C
ⓘ methodology & sources
Hero value: NASA GISTEMP annual mean for the latest complete year, rebased from the 1951–1980 baseline to a pre-industrial reference (1880–1899 mean = 0). The +0.29°C offset is computed live from the GISTEMP CSV — not hardcoded — so it tracks any future reprocessing. NOAA GlobalTemp and Berkeley Earth place current warming within 0.06°C of GISTEMP. The dashed line marks the Paris Agreement 1.5°C threshold (defined on the same pre-industrial reference). All three methods use independent station compilations and infilling; their convergence confirms the signal is real, not an artifact of method choice.
SOURCE · gistemp_preindustrial
Aerosol masking trap
STRUCTURAL
Aerosol masking trap
Fossil fuel combustion masks 0.5–1.5°C of warming via SO₂ aerosols.
Rapid decarbonization removes masking within weeks; CO₂ persists centuries.
No emissions pathway avoids the asymmetry between aerosol lifetime and CO₂ persistence.
Masking range
0.5–1.5°C
Removal timescale
WEEKS–MONTHS
CO₂ persistence
CENTURIES
Human control
CONSTRAINED
ⓘ methodology & sources
Sources: IPCC AR6 WGI Ch. 7 (aerosol ERF −1.06 W/m², 90% range −1.84 to −0.46); Samset et al. 2018 (aerosol removal timescale). Observed warming: NASA GISTEMP v4 annual anomaly (1880–1899 baseline). Aerosol cooling from IPCC AR6 best estimate. Total = observed + |aerosol forcing|.
1,939 ppb
↑ 6.3 ppb/yr
CURRENT RATE · 2025–26
Pre-industrial
722 ppb
Atmospheric lifetime
~12 yrs
Atmospheric concentration (ppb)
Annual growth rate (ppb / yr)
1999–2006 AVG
~1.2 ppb/yr
STABILIZATION
→
×8
2014–PRESENT AVG
~9.6 ppb/yr
RE-ACCELERATION
The cause remains scientifically contested — fossil fuel leaks, tropical wetland expansion, and shifting OH sink capacity are all implicated. No single driver has achieved consensus.
HUMAN CONTROL
DIRECT
CAUSE OF RE-ACCELERATION
CONTESTED
80-YR WARMING POTENTIAL
80× CO₂
ⓘ methodology & sources
NOAA GML global marine surface network. Pre-industrial CH₄ ≈ 722 ppb (ice-core consensus). Current levels represent a 2.6–2.7× increase. CH₄ has a ~12-year atmospheric lifetime but an 80-year warming potential 80× that of CO₂ — near-term reductions have an outsized effect on near-term warming. Annual growth rate computed as December minus January monthly mean for each complete year. Growth rate eras: stabilization 1999–2006 (near-zero net growth); re-acceleration 2014–present (cause contested — fossil fuel leaks, tropical wetland expansion, and shifting OH sink capacity all implicated). Ice core toggle: EPICA Dome C composite (Loulergue et al. 2008, NOAA/NCEI Paleoclimatology), spliced with this same NOAA GML record where the ice core itself stops (~1937).
Atmospheric Nitrous Oxide · Global Mean
CURRENT
339.82
ppb 1.26× pre-industrial
273× the warming potential of CO₂, and it lingers for over a century
The third greenhouse gas — mostly from fertilizer and the soils we farm.
Pre-industrial
270 ppb
Current
339.82
ppb
Atmospheric lifetime
~115 yrs
Human control
CONSTRAINED
ⓘ methodology & sources
NOAA GML global marine surface network. Pre-industrial N₂O ≈ 270 ppb (ice-core consensus). Synthetic nitrogen agriculture is the dominant source; the ~115-year atmospheric lifetime means every tonne emitted today is still warming in 2141. Unlike CH₄ reductions, N₂O mitigation offers no near-term shortcut — persistence is the story. The curve is nearly linear: no plateaus, no policy inflections, no visible ceiling.
SOURCE · noaa_n2o_global
The Godzilla El Niño
The equatorial Pacific is the planet's primary weather engine. ENSO shifts rainfall, drought, storm tracks, and global mean temperature across seasonal-to-interannual timescales. Four indicators read the current state: the canonical Niño 3.4 index, the CFSv2 ensemble forecast, the Southern Oscillation Index (atmospheric confirmation), and a real-time SST map.
Historical 5th–95th
Observed (OISST daily)
Forecast 5th–95th (120 members)
Ensemble mean
Historical mean
1997
2015
2023
2026
Historical range 1981–2025 (45 yrs), 5th–95th pct
OISST daily · provisional
1991–2020 mean
1988
1997
1999
2015
2023
2026
All years 1981–2025
Each line = 1 year, 1981–2025
2026 (Climate Reanalyzer, daily)
+1.5°C
+3.0°C+
Blank = below +1.5°C that month
Dashed = provisional (daily OISST, not yet confirmed)
Niño 3.4 · Sep 7 (daily)
+2.86°C
1991–2020 baseline · El Niño
Same month, older baseline
+1.74°C
1971–2000 baseline
Absolute SST · Niño 3.4
29.6°C
physical sea surface temperature
Among all Seps (76 yr)
#1/77
100.0th percentile — near record
Niño 3.4 · Sep 6
+3.85 SD
vs 1991–2020 mean
Highest on Sep 6 (46 yrs)
2026
+3.85 SD — 2026 leads
Days ≥ +2 SD this year
100/249
strong-anomaly threshold
Rank, this calendar day (46 yr)
#1/46
100th percentile
This episode, running
5mo
≥+0.5°C since 2026-05 · latest month provisional
Months ≥+2.0°C, this episode
3
record: 5 (2015–16)
Months ≥+2.5°C, this episode
2
record: 2 (2015–16)
El Niño episodes since 1950
19
≥5 consecutive months ≥+0.5°C
The Temperature Ratchet
STALE
+1.41°C
12-month mean · vs pre-industrial
LATEST MONTH
+1.46°C
July 2026
CURRENT FLOOR · 2023–24
+1.50°C
2023–24
FLOOR STEP-UP
+0.32°C
above prior floor
RATE OF RISE
+0.30°C / decade
MONTHS ≥1.5°C BY YEAR · 13 since Sep 2023, 2 in the decade before
Floor = 12-month mean starting 6 months after ONI peak · verified from NASA GISTEMP
Statistical projection (trend + ENSO regression), not an official forecast: ENSO-adjusted trend fit, 3-month lag, R²=0.80 vs 379 months of history; Nino 3.4 path from the live CFSv2 ensemble.
ⓘ methodology & sources
Floor = 12-month mean starting 6 months after ONI peak · verified from NASA GISTEMP
Source: NASA GISTEMP GLB.Ts+dSST · pre-industrial baseline 1880–1899
FORECAST ZONE METHODOLOGY — a monthly-resolution ENSO adjustment of the GISTEMP trend itself, the same technique used in published climate literature to separate short-term ENSO variability from the secular warming trend (see Foster & Rahmstorf, 'Global temperature evolution 1979–2010', Environ. Res. Lett. 2011, for the canonical version of this method).
Model: GISTEMP_anomaly(t) = intercept + trend·t + β·Niño3.4(t − lag). Fit by ordinary least squares against NASA GISTEMP's full monthly series (1995–present, ~380 points) and NOAA's ERSST-based Niño 3.4 monthly index. The lag (months between an ENSO reading and its effect on global mean temperature) is not assumed — it's chosen live by testing every lag from 0–12 months and keeping whichever gives the best fit (highest R²).
Forecast values project that fitted trend forward, plus β times NOAA CFSv2's live 80-member Niño 3.4 ensemble forecast at the same lag. Months close to 'now' use CFSv2's own recent observed SST (little uncertainty); months further out use the ensemble's actual forecast spread. The shaded band combines that ENSO-forecast uncertainty with the regression's own residual scatter, so it doesn't understate near-term noise.
This is a statistically-derived projection built from live public data (NASA GISTEMP + NOAA ERSST/CFSv2), not an official forecast product from any single agency — the fit (lag, R², and trend) recomputes on every refresh as new months of data arrive, so it updates itself rather than going stale like the two retired citations did.
El Niño shading (background bands on the chart) marks moderate-or-stronger months (Niño 3.4 ≥ 1.0°C) from the same live NOAA ERSST series — El Niño only, not La Niña, since this card's subject is specifically El Niño-driven floors.
Source: NASA GISTEMP GLB.Ts+dSST · pre-industrial baseline 1880–1899
FORECAST ZONE METHODOLOGY — a monthly-resolution ENSO adjustment of the GISTEMP trend itself, the same technique used in published climate literature to separate short-term ENSO variability from the secular warming trend (see Foster & Rahmstorf, 'Global temperature evolution 1979–2010', Environ. Res. Lett. 2011, for the canonical version of this method).
Model: GISTEMP_anomaly(t) = intercept + trend·t + β·Niño3.4(t − lag). Fit by ordinary least squares against NASA GISTEMP's full monthly series (1995–present, ~380 points) and NOAA's ERSST-based Niño 3.4 monthly index. The lag (months between an ENSO reading and its effect on global mean temperature) is not assumed — it's chosen live by testing every lag from 0–12 months and keeping whichever gives the best fit (highest R²).
Forecast values project that fitted trend forward, plus β times NOAA CFSv2's live 80-member Niño 3.4 ensemble forecast at the same lag. Months close to 'now' use CFSv2's own recent observed SST (little uncertainty); months further out use the ensemble's actual forecast spread. The shaded band combines that ENSO-forecast uncertainty with the regression's own residual scatter, so it doesn't understate near-term noise.
This is a statistically-derived projection built from live public data (NASA GISTEMP + NOAA ERSST/CFSv2), not an official forecast product from any single agency — the fit (lag, R², and trend) recomputes on every refresh as new months of data arrive, so it updates itself rather than going stale like the two retired citations did.
El Niño shading (background bands on the chart) marks moderate-or-stronger months (Niño 3.4 ≥ 1.0°C) from the same live NOAA ERSST series — El Niño only, not La Niña, since this card's subject is specifically El Niño-driven floors.
SOURCE · crw_sst_anomaly_tropics
Historical mean
1997
2015
2023
2026
Historical range 1951–2025 (75 yrs), 5th–95th pct
Walker Circulation Coupling
El Niño Coupled
Niño 3.4 +1.8°C · SOI -1.1
Percentile Rank · Last 3 Months
Aug
8th
←
Jul
1st
←
Jun
3rd
The committed warming reservoir
Oceans absorb ~90% of excess planetary heat. This heat represents warming committed regardless of future emissions decisions.
Ocean Surface Temperature · Argo Float Network
Mean anomaly
+0.90°C
vs WOA climatology by location
Above / below baseline
77%
above ·
23%
below
Coverage
3,610
floats reporting
Mean anomaly
+0.19°C
vs WOA climatology by location
Above / below baseline
59%
above ·
41%
below
Coverage
3,542
of 3,610
floats reporting
Mean anomaly
+0.01°C
vs WOA climatology by location
Above / below baseline
48%
above ·
52%
below
Coverage
3,292
of 3,610
floats reporting
Mean anomaly
+0.04°C
vs WOA climatology by location
Above / below baseline
57%
above ·
43%
below
Coverage
3,104
of 3,610
floats reporting
EL NIÑO SIGNAL
+2.73°C
248 floats in region
+0.35°C
246 floats in region
+0.02°C
226 floats in region
+0.04°C
218 floats in region
AMOC FINGERPRINT
+1.16°C
139 floats in region
+0.06°C
138 floats in region
-0.19°C
132 floats in region
-0.07°C
120 floats in region
ARCTIC
ARCTIC AMPLIFICATION
FASTEST WARMING OCEAN REGION
+1.16°C
103 floats in region
+0.13°C
91 floats in region
+0.07°C
72 floats in region
+0.04°C
61 floats in region
NORTH PACIFIC
MARINE HEATWAVE REGION
+1.70°C
91 floats in region
+0.07°C
89 floats in region
+0.02°C
82 floats in region
+0.01°C
80 floats in region
SOUTHERN OCEAN
PRIMARY HEAT SINK
+0.15°C
326 floats in region
+0.15°C
323 floats in region
+0.00°C
295 floats in region
+0.04°C
281 floats in region
INDIAN OCEAN
FASTEST WARMING BASIN
+0.61°C
156 floats in region
+0.09°C
143 floats in region
+0.09°C
145 floats in region
-0.04°C
134 floats in region
ATLANTIC MDR
HURRICANE MAIN DEVELOPMENT REGION
+1.08°C
136 floats in region
+0.10°C
136 floats in region
+0.06°C
129 floats in region
+0.01°C
117 floats in region
GULF OF MEXICO
HURRICANE INTENSIFICATION ZONE
+1.52°C
38 floats in region
+0.54°C
37 floats in region
+0.01°C
33 floats in region
+0.01°C
21 floats in region
MEDITERRANEAN
FASTEST WARMING SEA
+2.85°C
81 floats in region
+0.53°C
75 floats in region
+0.13°C
58 floats in region
+0.03°C
47 floats in region
GLOBAL ANOMALIES
OFF THE NORMAL SCALE
285
257 warm · 28 cold
479
286 warm · 193 cold
290
163 warm · 127 cold
261
184 warm · 77 cold
Hover over Regions
- LAST 12 MONTHS
- 2025
- 2024
- 2023
- 2022
- 2021
- 2020
- 2019
- 2018
- 2017
- 2016
- 2015
- 2014
- 2013
- 2012
- 2011
- 2010
- 2009
- 2008
- 2007
- 2006
- 2005
- SELECT MONTH
- January
- February
- March
- April
- May
- June
- July
- August
- September
- October
- November
- December
Atmospheric forcing signal — what's happening now
Thermocline — heat committed regardless of surface conditions
Permanent reservoir · 70-year warming record below
Bathyal signal — centennial inertia, written in fractions of a degree
· Argo float anomalies vs WOA climatology
The ocean doesn't care about the political calendar.
DEEP OCEAN WARMING RECORD
Argo float network 2005–present · 1000–2000 dbar depth band
Anomaly distribution — the whole ocean's shift, not just the mean
2005
Warming by latitude — where the heat is concentrating, 2005–2026
Hover a cell for the exact year, band, and anomaly
ⓘ methodology & sources
Argovis API (argovis-api.colorado.edu); Argo Program (Argo, 2024). 3,800+ autonomous profiling floats operated by ~30 nations. Each float descends to 2,000 m and measures temperature and salinity as it rises, transmitting via satellite every ~10 days. Surface temperature shown at 0–30 dbar depth. Raw degrees Celsius — no anomaly baseline applied. Blue = polar cold, orange/red = tropical warm. Data window: rolling 10-day, de-duplicated to one reading per float.
Global SST
21.18°C
+0.79°C
vs 1991–2020 mean · Sep 7
Upper Edge of Envelope
2026 above the 1981–2025 mean
250 of 250 days YTD
250 of 250 days YTD
Calendar-Day Records
110 of 250 days set
an all-time calendar-day record
an all-time calendar-day record
Hotter Than 2024
110 of 250 days
above the 2024 trace
above the 2024 trace
Today in Context
Sep 7 ·
2026 ranks #1 of 46 years
Historical mean
1982
1997
2015
2023
2024
2025
2026
Historical range 1981–2025 (45 yrs)
Surface Warming
+0.75°C
1980 → 2026 at the surface (1m)
Signal Depth
2000m
warming exceeds 2× the 1940–1969 natural variability down to this depth
warming exceeds 2× the 1940–1969 natural variability down to this depth
1km Turning Point
1996
first year the 1km layer turned anomaly-positive and stayed there
first year the 1km layer turned anomaly-positive and stayed there
OCEAN HEAT CONTENT
Source
Depth
Hemisphere
Upper-ocean heat anomaly above the 1955–2006 baseline
Heat that has penetrated below the thermocline
Deep-ocean heat recirculates on century timescales
Upper-ocean heat anomaly, IAP/CAS estimate
Deep-ocean heat anomaly, IAP/CAS estimate
Full-column heat anomaly, IAP/CAS estimate
Ocean Heat Anomaly
242.2ZJ
≈ 404
yrs of global energy use
zettajoules (10²¹ J) above 1955–2006 baseline
+11.1 ZJ / YR
past decade · 1955–2006 baseline
NH / SH DIVERGENCE
+1.6 ZJ
Northern Hemisphere accumulating faster
Southern Hemisphere is ~81% ocean — it buffers heat more efficiently, so divergence signals asymmetric planetary warming.
Ocean Heat Anomaly
121.9ZJ
≈ 203
yrs of global energy use
zettajoules (10²¹ J) above 1955–2006 baseline
+5.6 ZJ / YR
past decade · 1955–2006 baseline
Ocean Heat Anomaly
120.3ZJ
≈ 200
yrs of global energy use
zettajoules (10²¹ J) above 1955–2006 baseline
+5.6 ZJ / YR
past decade · 1955–2006 baseline
Ocean Heat Anomaly
97.5ZJ
≈ 162
yrs of global energy use
zettajoules (10²¹ J) above 1955–2006 baseline
+3.1 ZJ / YR
past decade · 1955–2006 baseline
NH / SH DIVERGENCE
-24.1 ZJ
Southern Ocean absorbing more heat
Southern Hemisphere is ~81% ocean — it buffers heat more efficiently, so divergence signals asymmetric planetary warming.
Ocean Heat Anomaly
36.7ZJ
≈ 61
yrs of global energy use
zettajoules (10²¹ J) above 1955–2006 baseline
+0.8 ZJ / YR
past decade · 1955–2006 baseline
Ocean Heat Anomaly
60.8ZJ
≈ 101
yrs of global energy use
zettajoules (10²¹ J) above 1955–2006 baseline
+2.3 ZJ / YR
past decade · 1955–2006 baseline
Ocean Heat Anomaly
339.7ZJ
≈ 566
yrs of global energy use
zettajoules (10²¹ J) above 1955–2006 baseline
+14.2 ZJ / YR
past decade · 1955–2006 baseline
NH / SH DIVERGENCE
-22.5 ZJ
Southern Ocean absorbing more heat
Southern Hemisphere is ~81% ocean — it buffers heat more efficiently, so divergence signals asymmetric planetary warming.
The 340 ZJ banked in the full ocean column is the accumulated integral of the EEI signal — heat absorbed faster than it can be radiated to space.
COMMITTED THERMOSTERIC SLR
~2.8 cm
from banked ocean heat · independent of ice melt · 0.11 mm/ZJ (IPCC AR6 WG1 Ch9 Table 9.5)
Ocean Heat Anomaly
158.6ZJ
≈ 264
yrs of global energy use
zettajoules (10²¹ J) above 1955–2006 baseline
+6.4 ZJ / YR
past decade · 1955–2006 baseline
Ocean Heat Anomaly
181.1ZJ
≈ 302
yrs of global energy use
zettajoules (10²¹ J) above 1955–2006 baseline
+7.9 ZJ / YR
past decade · 1955–2006 baseline
Ocean Heat Anomaly
290.7ZJ
≈ 484
yrs of global energy use
zettajoules (10²¹ J) above 1955–2006 baseline (history through 2025–12 rebased from IAP's native 2006–2015; 2026–01 onward computed from IAP's raw gridded anomaly data, rebased from its 1981–2010 baseline)
+4.3 ZJ / YR
past decade · 1955–2006 baseline · no hemisphere breakdown available from this source
Ocean Heat Anomaly
119.94ZJ
≈ 200
yrs of global energy use
zettajoules (10²¹ J) above 1955–2006 baseline (history through 2025–12 rebased from IAP's native 2006–2015; 2026–01 onward computed from IAP's raw gridded anomaly data, rebased from its 1981–2010 baseline)
+1.9 ZJ / YR
past decade · 1955–2006 baseline · no hemisphere breakdown available from this source
Ocean Heat Anomaly
410.64ZJ
≈ 684
yrs of global energy use
zettajoules (10²¹ J) above 1955–2006 baseline (history through 2025–12 rebased from IAP's native 2006–2015; 2026–01 onward computed from IAP's raw gridded anomaly data, rebased from its 1981–2010 baseline)
+6.2 ZJ / YR
past decade · 1955–2006 baseline · no hemisphere breakdown available from this source
RATE OF ACCUMULATION · ZJ PER YEAR BY DECADE
+1.5 ZJ/yr1955–1980
+1.7 ZJ/yr1980–2000
+6.9 ZJ/yr2000–2015
+7.6 ZJ/yr2015–present
RATE OF ACCUMULATION · ZJ PER YEAR BY DECADE
+1.3 ZJ/yr1955–1980
+2.1 ZJ/yr1980–2000
+1.1 ZJ/yr2000–2015
+5.4 ZJ/yr2015–present
RATE OF ACCUMULATION · ZJ PER YEAR BY DECADE
+0.2 ZJ/yr1955–1980
+-0.5 ZJ/yr1980–2000
+5.8 ZJ/yr2000–2015
+2.2 ZJ/yr2015–present
ⓘ methodology & sources
NCEI/NOAA quarterly mean heat content anomaly relative to the 1955–2006 baseline. Upper ocean (0–700 m): available 1955–present; stores roughly half of all excess planetary heat. Deep ocean (700–2000 m): derived series (0–2000 m minus 0–700 m); available from 2005 when the Argo float array reached full coverage. The 700–2000 m layer is a slow, near-irreversible store representing committed warming regardless of future emissions. 1 ZJ = 10²¹ J. Source: NOAA/NCEI · Levitus et al.
ⓘ methodology & sources
IAP/CAS (Institute of Atmospheric Physics, Chinese Academy of Sciences) IAPv4.2 ocean heat content estimate (Cheng et al., Earth Syst. Sci. Data 2024). A second, monthly-updating source alongside NCEI's quarterly product -- not a replacement, NCEI remains the primary series. History through 2025-12 comes from IAP's own OHC summary text product (native baseline 2006-2015, rebased onto NCEI's 1955-2006 baseline the same way -- IAP's own mean over that period, subtracted). That text product stopped updating after 2025-12, so from 2026-01 onward this collector computes OHC directly from IAP's raw gridded temperature-anomaly NetCDF files (still updating monthly), rebased onto the same 1955-2006 baseline via a fixed offset calibrated against the overlap between the two IAP products. No hemisphere breakdown is available from this source. 1 ZJ = 10²¹ J.
SOURCE · NCEI · NOAA/NESDIS · LEVITUS ET AL.
SOURCE · IAP/CAS · CHENG ET AL.
Heat available at depth to fuel hurricane rapid intensification — not just surface temperature
Gulf of Mexico
80.0 kJ/cm²
D26 (26°C isotherm): 58.3 m
Above rapid-intensification threshold
Atlantic MDR
53.3 kJ/cm²
D26 (26°C isotherm): 57.3 m
Above rapid-intensification threshold
Rapid-Intensification Threshold
50 kJ/cm² · Shay, Goni & Black (2000)
2026-09-03 · Gulf 80.0 · MDR 53.3 kJ/cm²
Gulf of Mexico
Atlantic MDR
RI threshold (50 kJ/cm²)
ⓘ methodology & sources
NOAA/NESDIS Satellite Ocean Heat Content Suite (SOHCS), NCEI THREDDS (no auth). TCHP is the integral of ocean heat content from the surface to the depth of the 26°C isotherm (D26) — the layer of warm water a hurricane can churn up before reaching cooler water that saps its intensity. AOML's own experimentally-labeled "TCHP" product is unmaintained (confirmed stalled since 2026-01); this card instead uses NESDIS's operationally-maintained `ohc` field, which shares the identical physical definition and units (kJ/cm²) — same metric, actively updated source. Rapid-intensification threshold (~50 kJ/cm²) is a widely cited operational rule of thumb (Shay, Goni & Black 2000), not a hard physical cutoff. Weekly-sampled since Jan 2024 from rolling 14-day satellite-altimetry + Argo composites, ~5-6 day publication lag. Gulf of Mexico and Atlantic MDR boxes match the Argo card's own regional callouts exactly.
The vanishing ice
Arctic sea ice is tracking below the historical floor on an exceptional number of days in 2026. Each km² of missing ice flips reflectivity from ~85% to ~6% — turning the Arctic from a planetary heat mirror into a heat sink.
Arctic Albedo Deficit · Additional Solar Absorption
CURRENT
THE SIGNAL
Ice extent · today
NSIDC
4.60 Mkm²
JAXA AMSR2
4.52 Mkm²
Sensor divergence
+0.08 Mkm²
Effective albedo
Today
32%
Historical mean
38%
Deficit
-5.9 ppts
vs historical mean
Albedo rank · this date
10th lowest
since 1978 · 44 years
Season absorption anomaly
+1776 W/m²·days
excess heat since Apr 1 · 160d · 16.2% above historical
Additional absorption · today
+5.14 W/m²
vs historical mean · per m² Arctic
Energy impact · total
72.3 TW
additional absorption · Arctic-wide
CERES measured albedo
31.6%
planetary · TOA · 2026-05
THE ENERGY
Cumulative absorption · since Apr 1
12,767 W/m²·days
+1776 W/m²·days
· 16.2% above historical
Historical mean to date
10,990 W/m²·days
Days into season
160d since Apr 1
Rank vs history
10th lowest since 1978 · 44 yrs
Today's instant rate
72.3 TW vs 12-mo mean 73.7 TW
TW = terawatts, trillions of watts of power
THE SYSTEM
An area the size of the Arctic accounts for roughly
7.4%
of the planet's entire energy imbalance — both figures as trailing 12-month means.
Surface area vs energy share — proportional
~4%
Arctic share of
Earth's surface
Earth's surface
→
7.4%
Arctic share of
planetary EEI
planetary EEI
Why the Arctic punches above its weight
Ice loss
Albedo deficit · ↑ solar absorption
→
Insulation removed
Ocean heat + moisture enter atmosphere
→
Arctic warms faster
3–4× global mean rate
→
Jet stream disrupted
Blocking highs · drought further south
Ice removal exposes the relatively warm Arctic ocean to cold atmosphere —
amplifying warming beyond albedo alone and disrupting mid-latitude weather patterns.
ⓘ methodology & sources SOURCE · seaice_heat_absorbed
Physics: ice reflects ~85% of incident sunlight back to space; open ocean absorbs ~94%. Each million km² of ice lost to the historical per-DOY mean absorbs an additional Δα × S_eff watts, where Δα = 0.79 and S_eff is the all-sky daily mean insolation at 75°N (clear-sky astronomical formula × 0.50 Arctic cloud-transmission factor). Value is 0 during polar night — no additional absorption when there is no sun. Conservative: only counts deficits relative to the 1979–2025 per-DOY mean; a day where 2026 extent exceeds the mean is treated as 0. Note: the '2026 Year to Date · Days Below Historical Floor' card uses a stricter reference — the per-DOY historical minimum — so both cards can be nonzero simultaneously (deficit vs mean, but not yet below the all-time floor for that day). Energy impact W/m² derived from TW ÷ Arctic Ocean area (14.056 million km²). CERES measured albedo (stat strip below) is the global mean planetary albedo including tropics, land, and clouds — not directly comparable to the Arctic regional effective albedo derived from sea ice extent above. Both are shown because they tell different parts of the same story: the Arctic is losing its mirror, and Earth’s total reflectivity is declining as a consequence.
Arctic Sea Ice
PIOMAS SUSPENDED
NEXTSIM CURRENT
Arctic sea ice volume
41%
less ice than 02/1979
Rank · this month
2nd lowest
since 1979 · 48 years
Vs historical average
-25.1%
September trend
-3.0 ×10³ km³/decade
neXtSIM · current
5.4 × 10³ km³
Hover the chart to compare PIOMAS and neXtSIM
PIOMAS 2026
neXtSIM 2026
2012
2020
2023
Historical range 1979–2025
Historical mean
neXtSIM and PIOMAS are independent models with different domains and physics — they don't track in lockstep at any given moment. Treat each as a trend indicator in its own right rather than a spliceable continuous series.
ⓘ methodology & sources
Volume integrates thickness × area — a more complete measure of ice loss than extent alone. Extent shows how far the ice reaches; volume shows how much is actually there. VOLUME view: PIOMAS (Schweiger et al. 2011) has been suspended since March 2026 (NCEP/NCAR R1 forcing terminated) — Copernicus Marine's neXtSIM (NERSC, ECMWF-forced) continues the live story, but the two models do not track in lockstep (validated offset swings -26% to +18% by season) so they are shown as separate segments, not a spliced series. EXTENT view: NSIDC v4 daily extent 1978–2025, reference years 2012 (all-time record minimum) and 2020 (second-lowest). VALIDATION view: NSIDC (NASA Team algorithm, DMSP SSMI/S) vs JAXA (Bootstrap algorithm, AMSR2) — independent instrument cross-check, more important than usual with NSIDC on Basic Level of Service due to federal funding cuts.
SOURCE · PIOMAS v2.1 + Copernicus Marine neXtSIM + NSIDC v4 + JAXA/NIPR ADS
2026 Year to Date · Days Below Historical Floor
CURRENT
40days
Days in 2026 where Arctic extent was below every prior year on that calendar date
days below floor
within envelope
not yet observed
ⓘ methodology & sources
Cumulative count of calendar days in 2026 where the NSIDC daily extent reading fell below the per-day historical minimum across the full 1978–2025 record. Unlike a streak, this total does not reset when extent recovers — a day back inside the envelope stops contributing, but previously counted days remain. Source: NSIDC Sea Ice Index v4.
SOURCE · seaice_days_below_range
▲ Data Infrastructure
NSIDC Sea Ice Index — Basic Level of Service. Primary source moved to Basic Level of Service due to funding limitations. Reduced QC, slower issue response, no algorithm updates going forward. Data continues but support degrades. JAXA cross-check elevated in importance.
NSIDC service-level notice ↗
What We Eat
World corn stocks at 76 days of supply — below the 90-day buffer floor where one bad harvest becomes a price shock.
Data as of September 2026
Breadbasket Stress
CURRENT
Heat · Moisture · Vegetation · 9 major regions
← scroll →
| Region | Heat | Moisture | Vegetation | Trend | Risk |
|---|---|---|---|---|---|
Normal
Watch
Stress
Extreme
|
Normal
Watch
Stress
Extreme
|
Normal
Watch
Stress
Extreme
|
Improving
Stable
Worsening
|
||
| US Midwest Corn / Soy | +1.4 °C Watch | 94% Normal | — — | Stable | Watch |
| Ukraine Wheat / Sunflower | -0.8 °C Normal | 99% Normal | — — | Stable | Normal |
| India (IGP) Wheat / Rice | +0.5 °C Normal | 114% Normal | — — | Improving | Normal |
| Brazil (Cerrado) Soy / Corn | +0.7 °C Normal | 80% Normal | — — | Stable | Normal |
| Morocco / N. Africa Wheat / Barley | +1.7 °C Watch | 106% Normal | — — | Stable | Watch |
| Australia Wheat | +0.1 °C Normal | 94% Normal | — — | Improving | Normal |
| Argentina (Pampas) Soy / Corn / Wheat | +1.1 °C Watch | 102% Normal | — — | Stable | Watch |
| Russia (Black Earth) Wheat / Sunflower | +0.2 °C Normal | 117% Normal | — — | Stable | Normal |
| EU Wheat Belt (Fr/De/Pl) Wheat / Barley | +2.0 °C Watch | 96% Normal | — — | Stable | Watch |
SOURCE · NCEI CDR NDVI · NOAA PSL CPC · ERA5/CDS
30.8% of land
6 POPULATED CONTINENTS COMBINED · ANTARCTICA EXCLUDED
1991–2020 era mean
17.2% of land
Threshold
SPEI3 ≤ −1.0 · moderately dry or worse (McKee 1993/2011 classification)
Global % of land in drought · 1991–present
┊ El Niño peak (ONI), reference only
Regional breakdown
2026-07-01
Hover the trend line to see any month's regional breakdown
ⓘ methodology & sources
Standardised Precipitation-Evapotranspiration Index (SPEI), 3-month accumulation window, from Copernicus CDS's ERA5-Drought product (ECMWF; peer-reviewed methodology: Keune et al. 2025, Scientific Data). Unlike a raw precipitation deficit, SPEI factors in temperature via potential evapotranspiration — so it captures drought intensified by heat, not just by low rainfall. The global figure and the 6-continent breakdown both exclude Antarctica: a probe found it makes up 29.4% of all land grid cells in this product yet behaves nothing like the rest of the world (mean SPEI3 +0.21 vs −0.28 elsewhere) — blending a permanent ice sheet into a "land in drought" figure diluted the real number by ~2 points and has no food/water/ecosystem meaning. Continent boundaries are approximate lat/lon boxes (same convention as this site's other illustrative regions), not authoritative political borders — North Africa's Mediterranean coast is grouped with Europe, Russia east of ~45°E with Asia, Greenland with North America. "Moderately dry or worse" = SPEI3 ≤ −1.0, the standard WMO-referenced classification (McKee et al. 1993/2011). 0.25° global grid, monthly, 1991–2020 reference period. This is a global measure — it is not a substitute for the Southwest U.S. Palmer Drought Severity Index tracked on the American West panel, which remains its own regional card.
World Grain Stocks · Days of Supply
Corn
76.2 d
2026 · days of supply
-7.4 vs prior year
below 90d buffer floor
Wheat
121.4 d
2026 · days of supply
-3.7 vs prior year
Rice
130.1 d
2026 · days of supply
-5.0 vs prior year
ⓘ methodology & sources
World ending stocks expressed as days of domestic consumption remaining — the standard food-security buffer metric. Corn has fallen from ~95 days in 2022 to ~77 days in 2026, already below the 90-day floor where one bad harvest triggers a price shock. Wheat and rice remain comfortable; the spread between them is the story. Source: USDA FAS Production, Supply, and Distribution (PSD) database — world totals from all-country sum of ending stocks ÷ domestic consumption. Updated monthly alongside the WASDE.
1990–last 12 months
Trailing 12 months
Elevated-price period (≥120 pts, 2+ months)
US Wheat Production · WASDE
CURRENT
1,531million bushels
ⓘ methodology & sources
Headline value is the USDA WASDE 2026/2027 projection — a forward estimate, since the 2026 crop is not yet harvested. The chart below compares two eras with the intervening decades omitted (axis break): the early 1970s, the last time US production sat this low, against recent years ending in the 2026 projection (dashed red bar). If this projection holds, it would be the smallest US wheat crop in over 50 years. Bars are NASS realized production; the dashed bar is the WASDE forecast. Headline source: WASDE via ESMIS, migrated May 2026 off the Akamai-blocked www.usda.gov; history: USDA NASS Quick Stats.
SOURCE · usda_wasde_csv
The seas we’ve already locked in
Satellite altimetry shows the rate of sea level rise has increased by 40% within the observational record — from 2.77 mm/yr in 1993–2005 to 3.88 mm/yr today. Direct measurement by two independent constellations, not a model output.
Data as of February 2026
Global mean sea level
Current rate
3.88 mm/yr
vs 2.77 mm/yr
1993–2005 baseline
1993–2005 baseline
Cumulative rise since 1993
112.0 mm
Rate acceleration
1.4× faster
Measurement
Satellite altimetry
TOPEX · Jason-1/2/3 · Sentinel-6
ⓘ methodology & sources
Data: CU Sea Level Research Group (sealevel.colorado.edu), seasonal signals removed. Satellites: TOPEX/Poseidon (1992–2005), Jason-1 (2001–2013), Jason-2 (2008–2019), Jason-3 (2016–present), Sentinel-6 (2020–present). Rate periods: 1993–2005 vs 2005–present. Two independent satellite constellations with overlapping missions confirm the trend. Citation: Nerem et al. 2018 (Sci. Adv.) for rate-acceleration methodology; CU 2026 release for current values.
Thermal expansion · sea level
Full column rate
1.20 mm/yr
0–2000 m · 2005–present
Deep layer addition
+0.42 mm/yr
700–2000 m · previously invisible
Share of total SLR
31%
of observed rise rate
Thermosteric committed floor
+28 mm
locked in from heat to date
Ocean heat penetration · 0–2000 m
0 m
300 m
700 m
2000 m
Heat below 700 m cannot escape on any human timescale.
What Argo revealed · 2005–present · same coefficient throughout
+0.42
mm/yr
from the deep ocean
Argo revealed 35% more expansion than upper-ocean instruments could see
Thermosteric committed floor · physics floor
+28 mm
From heat already absorbed · independent of future emissions · irreversible on civilizational timescales
Causal pipeline · EEI → expansion → sea level
EEI · 12-mo mean
+1.95 W/m²
→
Ocean heat 0–2000 m
250 ZJ
→
Thermosteric rise
+27.5 mm
→
Share of SLR
31%
ⓘ methodology & sources
Thermosteric sea level contribution derived from 0-2000m ocean heat content (NCEI/Argo, same sidecar as Oceans panel). Conversion: ~0.11 mm per ZJ of OHC change (IPCC AR6 approximation). Covers Argo float network depth range 0-2000m; deep-ocean expansion below 2000m is not captured and adds to the true thermosteric total. IPCC AR6 estimates thermosteric at 40-44% of total sea level rise; the ~30% shown here reflects the 0-2000m layer only.
Antarctic ice sheet
SL Contribution
+0.014 mm/yr
2025 (latest full year)
Cumulative since 2002
+7.359 mm
sea level equivalent
2002–2010 mean
+0.274 mm/yr
2011–present mean
+0.316 mm/yr
Peak rate
+0.905 mm/yr (2015)
ⓘ methodology & sources
NASA GRACE/GRACE-FO JPL Mascon RL06.3Mv04 (Tellus Level-4 regional mass anomaly time series); podaac.jpl.nasa.gov; C3206284786-POCLOUD. Coverage: 2002-present, monthly, ~3-month processing lag. Mass anomaly in Gt (relative to 2002–2021 mean) converted to mm sea level equivalent (1 mm SLE = 361.8 Gt); cumulative anchored to 2002. Updated monthly via collect_ice_sheets.py.
Greenland ice sheet
SL Contribution
+0.317 mm/yr
2025 (latest full year)
Cumulative since 2002
+15.684 mm
sea level equivalent
2002–2010 mean
+0.659 mm/yr
2011–present mean
+0.630 mm/yr
Peak rate
+1.285 mm/yr (2011)
ⓘ methodology & sources
NASA GRACE/GRACE-FO JPL Mascon RL06.3Mv04 (Tellus Level-4 regional mass anomaly time series); podaac.jpl.nasa.gov; C3206299308-POCLOUD. Coverage: 2002-present, monthly, ~3-month processing lag. Mass anomaly in Gt (relative to 2002–2021 mean) converted to mm sea level equivalent (1 mm SLE = 361.8 Gt); cumulative anchored to 2002. Updated monthly via collect_ice_sheets.py.
BY 2100
LOCKED IN
Committed floor
0.2–0.3 m
by 2100, any scenario
Low emissions
0.3–0.6 m
SSP1-2.6 · 2100
High emissions
0.6–1.0 m
SSP5-8.5 · 2100
Thermal expansion from heat already absorbed is irreversible on human timescales.
The floor is locked in regardless of future emissions decisions.
Greenland and Antarctic destabilization could raise this floor significantly.
ⓘ methodology & sources
IPCC AR6 WGI Chapter 9 and SPM Table SPM.2. Committed thermal expansion alone: ~0.1-0.2 m regardless of future emissions (Church et al. 2013; IPCC AR6). Scenario ranges shown as likely ranges (66th percentile) for 2100. Current sea level ~0.11 m above 1993 satellite altimetry baseline (CU Sea Level Research Group 2026).
Glacier mass balance
Loss rate · 2025
–1091 mm w.e./yr
IPCC AR6: ~1.0 mm/yr sea level (1993-2018)
Rate acceleration
+70%
faster than 1992–2010 avg
1992–2010 avg
–519 mm w.e./yr
2011–present avg
–880 mm w.e./yr
ⓘ methodology & sources
WGMS reference glacier synthesis (~90 long-record glaciers worldwide). Annual specific mass balance in mm w.e. (water equivalent). Not directly mm sea level — IPCC AR6 (Table 2.5) estimates global glacier sea level contribution at 0.92 ± 0.16 mm/yr for 1993-2018, rising to ~1.1 mm/yr in 2010-2019. Data updated annually each spring with prior season measurements. WGMS (2024): Fluctuations of Glaciers Database. DOI:10.5904/wgms-fog-2024-11.
The American West
Snow that doesn't fall, water that isn't stored, and the fire that follows.
One drying system, read three ways.
Snow · Fire · Drought — the American West drying system
Western Drought · Severity & Water Year Deficit
PDSI: Southwest U.S. (AZ, CO, NM, NV, UT) · Water Year: West of the Mississippi (17 states)
PALMER DROUGHT SEVERITY INDEX
-9.33
Extreme Drought
1st most negative of 1579 months since 1895
≤−4 Extreme
≤−3 Severe
≤−2 Moderate
0 Normal
PDSI = Palmer Drought Severity Index · 0 = normal · negative = drought · Southwest U.S. region
WATER YEAR DEFICIT · WEST OF THE MISSISSIPPI
-2.38"
cumulative precip departure · WY2026
Oct 2025 – present
44.31%
USDM · D2 (severe) or worse
week of 2026-09-01
WY2021
WY2022
WY2023
WY2024
WY2025
WY2026
USDM D2+ % (right axis)
Water year Oct 1–Sep 30 · area-weighted precipitation departure from 1991–2020 normals, 17 states west of the Mississippi (excludes river-straddling MN/IA/MO/AR/LA) · dashed line = USDM % of that footprint in D2 (severe) or worse drought, current water year
Snowpack · Upper Colorado Basin SWE
CURRENT
#1
of 46 years
Driest on record · 1981–present
PEAK SWE · 2026
9.24"
March 8
MEDIAN PEAK
17.77"
1981–2025
% OF MEDIAN
52%
of historical median
PRIOR RECORD
11.03"
2002
DEFICIT VS RECORD
-1.79"
below prior record
MELT-OUT · 2026
June 5
17d earlier than median
Historical range 1981–2025
Historical mean
2002 (prior record)
2012
2018
2023 (exceptional year)
2026 ★
2026 (purple) against every year 1981–2025. Band is the full
historical min–max per calendar day; dashed line is the historical mean. 2023 (red)
was an exceptional snow year — its contrast with 2026 illustrates the basin's full
drought range. During June–October, SWE drops to near-zero as melt-out completes —
trace readings are expected seasonal behavior.
ⓘ methodology & sources
Daily basin mean snow water equivalent (SWE) for the Upper Colorado watershed (HUC-14), computed from USDA NRCS SNOTEL stations. Filtered by HUC prefix 14, not state: 'CO:SNTL' includes stations in the Arkansas drainage that would contaminate the basin mean. 2026 peaked at 9.24" on March 7 — driest on record (1 of 46 years), 52% of the 17.77" historical median and 1.79" below the prior record set in 2002 (11.03"). Melt-out was the earliest on record. Percentile is Claude-computed from raw per-station data.
SOURCE · awdb_snowpack_upper_colorado
Wildfire · Acres Burned
CURRENT
8,397,760
acres YTD
≈161% of the 10-year average as of September 6th 2026
2026 leads all years in the comparison window
2026 leads all years in the comparison window
1,644,958
FIRMS detections · contiguous US · YTD 2026-09-06
¹ SNPP only ² SNPP + NOAA-20 ³ + NOAA-21 (2026-07-16+)
Good 374
· Moderate 121
· USG 8
· Unhealthy 2
505 western US stations · PM2.5 · (cached 2026-08-17)
64
large fires active
4
new today
15,025
personnel assigned
—
preparedness level
(as of 2026-09-04)
53,665
wildfires year-to-date
WA
most unhealthy+ stations · 1
unhealthy+ stations · 30d
82
ⓘ methodology & sources
National year-to-date acres burned, from the NIFC statistics page (updated ~daily; upstream of record is the weekly IMSR). The hero is NIFC's own reconciled national total, not a sum reconstructed from incident records. The comparison bars show year-to-date acres for 2016–2026; the dashed rule is the 2016–2025 ten-year average (1,172,748 acres), pinned as a constant until the rolling window advances in Jan 2027. Heat is keyed to value: the current year is full alarm at roughly twice the decade average. The baseline itself is climbing, which the bars show rather than hide. The FIRMS calendar (right column) is a separate variable — daily VIIRS satellite fire detections across the contiguous US, not acres — shown alongside for a same-region, different-instrument view.
SOURCE · nifc_fire_ytd
Water · the Colorado reservoir system — % of capacity computed from storage volume, not elevation
Lake Powell
Capacity
21.9% full
Available Water
5.13 maf
≈ 1.7T gallons
Surface Elevation
3517.58 ft
Min power pool
+27.58 ft
above min power pool
ⓘ methodology & sources
Latest observation: 2026-09-07 · 2d old · fetched 2026-09-09 · NGVD 1929
21.9% = 5.13 maf storage ÷
23.42 maf live capacity.
Capacity is computed from storage volume (USBR datatype 17), not elevation —
reservoir hypsometry is nonlinear, so a percent-by-elevation would overstate
how much water is present.
Min power pool (3490 ft)
+27.58 ft
Dead pool (3370 ft)
+147.58 ft
Glen Canyon Dam. The downstream giant the entire upper system is being drained to prop up. Capacity is computed from storage volume (USBR datatype 17), not elevation.
SOURCE · USBR hydrodata 919
Lake Mead
Capacity
26.4% full
Available Water
6.9 maf
≈ 2.2T gallons
Surface Elevation
1038.93 ft
Min power pool
+88.93 ft
above min power pool
ⓘ methodology & sources
Latest observation: 2026-09-07 · 2d old · fetched 2026-09-09 · NGVD 1929
26.4% = 6.9 maf storage ÷
26.12 maf available capacity.
Capacity is computed from storage volume (USBR datatype 17), not elevation —
reservoir hypsometry is nonlinear, so a percent-by-elevation would overstate
how much water is present. (Basis: documented
available capacity, not back-solved live capacity — see source notes.)
Min power pool (950 ft)
+88.93 ft
Dead pool (895 ft)
+143.93 ft
Hoover Dam. Largest reservoir in the US by capacity; reduced Powell releases accelerate its decline. Capacity on documented available-capacity basis (26.12 maf).
SOURCE · USBR hydrodata 921
Flaming Gorge
Capacity
69.6% full
Available Water
2.56 maf
≈ 0.8T gallons
Surface Elevation
6009.32 ft
ⓘ methodology & sources
Latest observation: 2026-09-07 · 2d old · fetched 2026-09-09 · NGVD 1929
69.6% = 2.56 maf storage ÷
3.68 maf live capacity.
Capacity is computed from storage volume (USBR datatype 17), not elevation —
reservoir hypsometry is nonlinear, so a percent-by-elevation would overstate
how much water is present.
Upstream reserve on the Green River. 660kaf-1maf being released through Apr 2027 to prop up Powell, drawing it toward ~59%.
SOURCE · USBR hydrodata 917
Navajo
Capacity
47.9% full
Available Water
0.79 maf
≈ 0.3T gallons
Surface Elevation
6011.27 ft
ⓘ methodology & sources
Latest observation: 2026-09-07 · 2d old · fetched 2026-09-09 · NGVD 1929
47.9% = 0.79 maf storage ÷
1.65 maf live capacity.
Capacity is computed from storage volume (USBR datatype 17), not elevation —
reservoir hypsometry is nonlinear, so a percent-by-elevation would overstate
how much water is present.
Upstream reserve on the San Juan. No additional releases planned due to low levels and poor inflow forecasts.
SOURCE · USBR hydrodata 920
Blue Mesa
Capacity
21.9% full
Available Water
0.16 maf
≈ 0.1T gallons
Surface Elevation
7418.62 ft
Min power pool
+58.62 ft
above min power pool
ⓘ methodology & sources
Latest observation: 2026-09-07 · 2d old · fetched 2026-09-09 · NGVD 1929
21.9% = 0.16 maf storage ÷
0.75 maf live capacity.
Capacity is computed from storage volume (USBR datatype 17), not elevation —
reservoir hypsometry is nonlinear, so a percent-by-elevation would overstate
how much water is present.
Min power pool (7360 ft)
+58.62 ft
Dead pool (7220 ft)
+198.62 ft
Gunnison River, part of the Aspinall Unit — one of the four CRSP Initial Units built alongside Powell, Flaming Gorge, and Navajo. Its releases have been used to prop up Powell during the current drought.
SOURCE · USBR hydrodata 913
Lake Havasu
Capacity
79.5% full
Available Water
0.51 maf
≈ 0.2T gallons
Surface Elevation
447.05 ft
Min power pool
+7.05 ft
above min power pool
ⓘ methodology & sources
Latest observation: 2026-09-07 · 2d old · fetched 2026-09-09 · NGVD 1929
79.5% = 0.51 maf storage ÷
0.65 maf total capacity.
Capacity is computed from storage volume (USBR datatype 17), not elevation —
reservoir hypsometry is nonlinear, so a percent-by-elevation would overstate
how much water is present. (Basis: documented
available capacity, not back-solved live capacity — see source notes.)
Min power pool (440 ft)
+7.05 ft
Dead pool (400 ft)
+47.05 ft
Parker Dam, downstream of Hoover and Davis — the diversion point for the Central Arizona Project and Metropolitan Water District canals. Capacity on documented total-capacity basis (646.2 kaf; no separately published live/active split).
SOURCE · USBR hydrodata 923
Autonomous amplifiers
Earth system mechanisms that, once activated by human-caused warming, amplify themselves independent of any further human action.
AMOC slowdown
ONGOING
AMOC slowdown
The subpolar North Atlantic has cooled anomalously against a warming background — a cold blob fingerprint consistent with a weakened AMOC suppressing poleward heat transport.
RAPID array measurements show a −1.1 Sv/decade transport decline since 2004. Greenland meltwater is accelerating freshwater discharge into the formation zone, reducing the density gradient that drives overturning.
⟷ Greenland coupling — Greenland’s elevation feedback is accelerating freshwater discharge into the North Atlantic formation zone, reducing the density gradient that drives overturning. See Greenland Ice Sheet panel.
Transport trend
−1.1 Sv/decade
Cold blob
+1.64°C
SST anomaly vs 1991–2020
Potential CO₂ release
47–83
ppm if collapse
Human control
NONE
ⓘ methodology & sources
Cold blob: NOAA OISST v2.1 via ERDDAP, monthly mean 45–65°N 5–65°W vs 1991–2020 climatology. Transport trend: RAPID-MOCHA-WBTS program, rapid.ac.uk; McCarthy et al. 2015 Geophys. Res. Lett. Consequence if collapse: Zhu et al. 2023 Nat. Clim. Change (47–83 ppm CO₂ from Southern Ocean outgassing, independent of human emissions).
Greenland Ice Sheet
CROSSED
Greenland Ice Sheet
7 m
sea level equivalent — full Greenland commitment
The Greenland Ice Sheet sits in a trap of its own lowering. As surface ice melts, the sheet surface drops into warmer air — the atmosphere cools with altitude at −6.5°C per 1,000 m, so every meter lost brings the remaining ice into a warmer layer. More melt, lower surface, warmer air, more melt.
Above approximately 1.5°C of global warming, this loop cannot be closed. That threshold is behind us. The sheet’s full contribution — 7 metres of sea level equivalent — is now a question of centuries, not if.
⟷ AMOC forcing — Greenland meltwater discharge is a primary autonomous freshwater input disrupting North Atlantic circulation. See AMOC panel.
Surface temp · 14d mean
+0.9°C
above 1991–2020 average
Domain above 0°C · 14d
69%
of ice sheet domain
This month rank
#5/31
warmest on record
2003–2010 avg
0.66
mm/yr sea level
2015–2025 avg
0.63
mm/yr sea level
1.5°C threshold
CROSSED
Human control
NONE
ⓘ methodology & sources
Surface temperature: ERA5 derived daily statistics (2m temperature), daily mean area-weighted over 60–84°N 73–10°W, 14-day rolling mean anomaly vs 1991–2020 climatology, ~5d lag via ERA5T preliminary. Mass balance data: NASA GRACE/GRACE-FO JPL Mascon RL06.3Mv04; Tellus Level-4 regional Greenland time series; podaac.jpl.nasa.gov; 2003–present monthly. Commitment threshold: Robinson et al. 2012 Nature; Ridley et al. 2010 Climate Dynamics. Sea level equivalent: IPCC AR6 WGI Ch9. Coupling mechanism: freshwater flux suppresses North Atlantic deep water formation (Köhl et al. 2023; Caesar et al. 2021).
Permafrost thaw
ACTIVE
Permafrost thaw
Permeability rises by orders of magnitude across the −5°C to +1°C transition zone.
Arctic warming at 4× the global average rate — closing the gap to the transition zone.
Feedback operates independently of subsequent emissions reductions once the transition begins.
Arctic mean temp
-8.7°C
ERA5 annual mean
Warming rate
+0.68°C
°C / decade since 1979
Methane rise
+9.6
ppb / yr · 5-yr mean
Carbon stock
~1,500
Gt C
Active layer depth
~57 cm
CALM 2023 mean
Human control
INDIRECT
ⓘ methodology & sources
Sources: IPCC AR6 WGI Ch. 5 (carbon stock ~1,460–1,600 Gt C); Turetsky et al. 2019 Nature Geosci. (abrupt thaw); Burt & Williams 1976 (permeability transition). Active layer: CALM circumpolar network (~170 sites); NOAA Arctic Report Card 2023. Arctic temperature: ERA5 via Climate Reanalyzer, 66.5–90°N annual mean. CH₄ growth rate: NOAA GML global marine surface network monthly mean (ch4_mm_gl.txt); 5-year mean rate; multiple sources contribute to CH₄ growth (permafrost, wetlands, fossil fuels) — not permafrost-exclusive.
Boreal forest dieback
ACTIVE
Boreal forest dieback
Boreal forests (Canada, Russia, Alaska) cover 30% of the world’s forested area and store 30–40% of all terrestrial carbon — the largest land carbon pool on Earth.
Warming at 2–4× the global average drives a self-reinforcing cycle: warmer winters let bark beetles survive and kill billions of trees; dead forests burn; fires expose permafrost; permafrost releases CO₂ and CH₄ that drives further warming.
The 2023 Canadian fire season burned 18.4 Mha — the largest on record by a factor of two. North American boreal released 1.89 Gt CO₂, equal to 3.3× Canada’s entire annual fossil-fuel output in a single fire season.
1,083,243
FIRMS detections · Canada + Alaska · YTD 2026-09-06
2026 pace vs. 2023 record · same calendar day
41% of 2023 pace
¹ SNPP only ² SNPP + NOAA-20 ³ + NOAA-21 (2026-07-16+)
2023 fire season
1.89 Gt CO₂
5.3× the long-run average
2023 fires vs. Canada fossil fuels
3.3×
one fire season = Canada’s annual CO₂ output
Long-run average
0.36 Gt CO₂/yr
N. America boreal · 1997–2024
2026 season · live
1,083,243
satellite fire detections · Canada + Alaska · YTD
Human control
NONE
structural · decades timescale
② methodology & sources
Historical emissions: GFED5.1 CO₂ summary table (globalfiredata.org; van der Werf et al. 2023, Nat. Clim. Change). BONA (Boreal North America) region; units 1×10¹³ g CO₂ converted to Gt. Bars show all fire types combined. Current-year detections: NASA FIRMS, combining three co-orbiting VIIRS instruments — SNPP, NOAA-20, NOAA-21 (SP archive + NRT each; NOAA-21 has no SP archive product yet) — bbox 168°W–52°W 50°N–84°N, nominal + high confidence only (firms.modaps.eosdis.nasa.gov; FIRMS_MAP_KEY required). Detection count sums all three satellites' passes, so it is not a deduplicated unique-fire count; figures collected before 2026-07-16 used SNPP alone and are not directly comparable at the raw-count level. Canada fossil-fuel reference: Environment and Climate Change Canada 2024 National Inventory Report (~0.57 Gt CO₂ fossil + industrial). Feedback pathway: fire carbon release → warming → drying + beetle kill → more fire; coupled to Permafrost thaw card #01.
Amazon carbon flip
WEAKENING
Amazon carbon flip
Eastern Amazon now net carbon emitter.
Previously absorbing ~2 Gt CO₂/yr.
A structural reversal in the Amazon's role in the global carbon budget.
Net emission
0.8
Gt CO₂ / yr
Previous sink
~2.0
Gt CO₂ / yr
Response time
YEARS–DECADES
Human control
LOW
ⓘ methodology & sources
Sources: Gatti et al. 2021 Nature 595 (Eastern Amazon flux 2010–2018); Global Carbon Project 2023. Sign convention: positive = net absorption (sink); negative = net emission. Data schematic — see source for annual uncertainty ranges.
Coral reef collapse
ACTIVE
Coral reef collapse
42.6%
% of reefs globally under bleaching alert · rolling 365d
Coral reefs cover <1% of the ocean floor but support ~25% of all marine species and food security for 500 million people.
Bleaching occurs when thermal stress exceeds 4°C-weeks above the maximum monthly mean. Bleached reefs release stored carbon and shift from net sink to net emitter.
The 4th global bleaching event (2023–24) was the most extensive on record. Event intervals have compressed from decade-scale to near-annual.
Pacific
35.6%
% at AL1 · rolling 365d
Atlantic
94.0%
% at AL1 · rolling 365d
Indian Ocean
34.9%
% at AL1 · rolling 365d
Human control
INDIRECT
ⓘ methodology & sources
NOAA Coral Reef Watch CoralTemp v3.1 5km satellite product. Bleaching Alert Area (BAA): % of reef pixels globally at Alert Level 1 (DHW ≥4°C-weeks, bleaching expected with some mortality) within the rolling past 365 days. 5-day maximum composite. No auth; daily update. Active global bleaching event threshold: ≥20% globally and ≥12% in each tropical ocean basin concurrently for ≥22 weeks (NOAA/ICRI 2024 criteria). Carbon feedback: Hoegh-Guldberg et al. 2017 Science 356; IPCC AR6 WGII Ch. 3.2.
Thwaites glacier
ACTIVE
Thwaites glacier
3.3 m
Committed sea level rise:
If full Thwaites buttressing fails
Thwaites sits on a retrograde bed — rock that slopes deeper inland. As warm ocean water pushes the grounding line inward, it retreats onto deeper bed, exposing more ice face to melt, accelerating the retreat further. The geometry is the amplifier.
Human warming pulled the trigger. The bed sustains the process regardless of what happens to emissions next. The ITGC finds no scenario in which retreat reverses on human timescales.
Thwaites buttresses neighboring West Antarctic glaciers. If the buttressing fails, neighboring glaciers accelerate. The committed rise is not from Thwaites alone.
BedMachine v3 · central flowline · Morlighem et al. 2020
MEaSUREs NSIDC-0498 · Rignot et al. 2014 · Milillo et al. 2019
Thwaites alone
~65 cm
sea level rise at full collapse
GL retreat 1992–2011
14 km
Rignot et al. 2014 · western sector
Current retreat rate
~1 km/yr
accelerating · InSAR observations
Reversal timescale
NONE
geometry now drives the process
Human control
NONE
trigger pulled · ITGC consensus
ⓘ methodology & sources
Bed topography: BedMachine Antarctica v3 (Morlighem et al. 2020, Nat. Geosci. 13, 132–137; doi:10.1038/s41561-019-0510-8); representative values along central Thwaites flowline. Grounding line retreat: MEaSUREs Antarctic Grounding Line from Differential Satellite Radar Interferometry v2 (NSIDC-0498); Rignot et al. 2014 GRL (14 km western sector 1992–2011); Milillo et al. 2019 Sci Adv (continued retreat 2011–2017, up to 4 km/yr in eastern sector). Current rate ~1 km/yr average; series extended to 2023 at observed rate. Committed SLR: Thwaites alone ~65 cm (Joughin et al. 2014, Science 344, 735–738); full WAIS with buttressing failure 3.3 m (Bamber et al. 2019, Nature 575, 58–62). Marine Ice Sheet Instability (MISI): Weertman 1974; Schoof 2007, Science 315, 838–841. ITGC: International Thwaites Glacier Collaboration (itgc.org), ongoing since 2019.
SYNTHESIS
Pulling it all together
Persistent planetary energy imbalance propagates through interconnected Earth systems with different response times, thresholds, and recovery capacities. This page synthesises the observatory into a single systems-level view.
OBSERVATORY STATE · ANALYSIS
The energy imbalance is the upstream fact from which the rest of this page descends. The planet is now retaining roughly three-quarters more energy per square metre than it did across the CERES-era mean — the imbalance has not merely persisted, it has widened against its own recent baseline. Everything downstream is a ledger entry for where that retained energy goes. Most of it enters the ocean, which is why ocean heat and sea-level rise register as the imbalance's most direct receipts: the sea-level rate has stepped up from its 1993–2005 pace, not through any new mechanism but because a warmer, expanding ocean is the imbalance made visible.
All seven tracked domains read elevated simultaneously, and El Niño is one of them — the ENSO state is active at +2.70°C, so this is not a case of stress appearing without a warm-phase driver. The honest claim is narrower and harder to dismiss: the breadth of elevation exceeds what this ENSO state alone has historically produced. An El Niño can warm the surface, tilt drought, and prime fire seasons, but the matrix shows heat content, sea ice, food-system buffers, and drought co-elevated alongside it. These domains share the energy imbalance as a common driver; what they do not share is subordination to a single climate mode. Seven of seven does not fall out of ENSO's envelope.
What this cycle has already committed is the ocean's accumulated heat and the sea-level response it drives — quantities that respond to the imbalance over decades, not seasons, and that will keep rising after the current El Niño relaxes. Read the other panels with that lag in mind: the surface-mode signals will recede when ENSO turns, but the ocean and cryosphere readings are recording the slower, retained fraction. When ENSO subsides, watch which domains stand down and which do not — the ones that remain elevated are measuring the imbalance directly, not the mode.
ⓘ how this analysis is generated
This synthesis is generated by claude-opus-4-8 reading the observatory's current instrument readings across all panels.
Inputs: CERES EBAF (EEI), NCEI/Argo (OHC), NSIDC (sea ice), USDA FAS PSD (grain stocks), NIFC (fire), NOAA CPC (ENSO), NASA GISTEMP (global temperature), CU Sea Level (GMSL), RAPID array (AMOC), NOAA GML (CO₂).
Generated 2026-08-21. The analysis follows the observatory's editorial posture: unhedged, no net-zero conditionals, damage-limitation framing. No prescription.
ENERGY PROPAGATION
The cascade
EEI as the single upstream driver. Energy propagates through each system with different response times and remaining human control. Click any node to view its source panel.
REVERSIBILITY
Response times — where each system sits on its curve
Not a prediction. A "where are we in the process" read — each system's position relative to its own response and recovery curve.
SYNCHRONIZATION
Co-elevation — systems stressed simultaneously
The argument is not that any one system is at a record. It is that multiple systems are elevated at once. That synchronisation is what makes 2026 structurally different from prior climate variability — including prior El Niño years.
| YEAR | EEI | OHC | SEA ICE | FOOD | FIRE | ENSO | DROUGHT | NOTE |
|---|---|---|---|---|---|---|---|---|
| 2026 NOW | ▲ | ▲ | ▲ | ▲ | ▲ | ▲ | ▲ | 7 domains tracked — 7 of 7 elevated |
| 2023 | ▲ | ▲ | ▲ | ○ | ○ | ▲ | ▲ | Record OHC. Severe Antarctic sea ice deficit. El Nino developing. Food and US fire not at threshold. |
| 2016 | ○ | ▲ | ▲ | ○ | ○ | ▲ | ▲ | Strongest El Nino in satellite record. Record winter sea ice low. Ocean heat elevated. Food and fire systems not stressed. |
| 2012 | ▲ | ○ | ▲ | ▲ | ▲ | ○ | ▲ | Arctic record minimum. US drought. Record fire year. No El Nino forcing — structural stress. |
| 1998 | — | — | ○ | ○ | — | ▲ | ▲ | El Nino dominant — strongest on record at the time. Energy and ocean systems not yet in satellite era. |
▲ elevated / stressed relative to historical baseline
○ within historical envelope
— pre-satellite era or data unavailable
ⓘ methodology — threshold definitions
EEI: Annual mean > CERES record mean (1.11 W/m², 2001–present). Pre-2001: —.
OHC: 5-year mean gain > 2005–present mean rate (10.5 ZJ/yr). Pre-2005: —.
Sea ice: Annual mean NSIDC extent >1 SD below 1981–2010 climatological mean (11.47 M km²). Current year: any days below per-DOY all-time minimum.
Food: Global corn stocks-to-use < 90-day buffer floor (USDA FAS PSD).
Fire: NIFC acres ≥ 125% of prior 10-year average. Historical years: full-year totals vs contemporary 10yr average.
ENSO: Niño 3.4 ≥ +0.5°C for 3+ consecutive months, 1991–2020 baseline (NOAA CPC).
Drought: Global % of land area at/below SPEI3 −1.0 (moderately dry or worse, Copernicus CDS ERA5-Drought) > the product's own 1991–2020 era mean. Current year: trailing 12-month mean of the % series. A global measure — not the Southwest U.S. Palmer Drought Severity Index tracked separately on the American West panel.
OHC: 5-year mean gain > 2005–present mean rate (10.5 ZJ/yr). Pre-2005: —.
Sea ice: Annual mean NSIDC extent >1 SD below 1981–2010 climatological mean (11.47 M km²). Current year: any days below per-DOY all-time minimum.
Food: Global corn stocks-to-use < 90-day buffer floor (USDA FAS PSD).
Fire: NIFC acres ≥ 125% of prior 10-year average. Historical years: full-year totals vs contemporary 10yr average.
ENSO: Niño 3.4 ≥ +0.5°C for 3+ consecutive months, 1991–2020 baseline (NOAA CPC).
Drought: Global % of land area at/below SPEI3 −1.0 (moderately dry or worse, Copernicus CDS ERA5-Drought) > the product's own 1991–2020 era mean. Current year: trailing 12-month mean of the % series. A global measure — not the Southwest U.S. Palmer Drought Severity Index tracked separately on the American West panel.
CONTROL ASYMMETRY
What remains controllable
The honest close. Systems where human influence remains HIGH or MODERATE get one visual treatment. Systems where it is LOW or VERY LOW get another. The line is drawn at the boundary between HIGH/MODERATE and LOW/VERY LOW human control — not editorial judgment, but a direct output of the systems science.
| SYSTEM | HUMAN INFLUENCE | HUMAN CONTROL |
|---|---|---|
| LEVERS THAT REMAIN | ||
| CO₂ emissions | DIRECT | HIGH |
| Aerosol masking trap | DIRECT (involuntary) | MODERATE |
| Atmospheric moisture | MODERATE | MODERATE |
| RESPONSES ALREADY IN MOTION | ||
| Ocean heat accumulation | INDIRECT | LOW |
| Permafrost thaw | INDIRECT | LOW |
| Sea level rise | INDIRECT | LOW |
| Ice sheet response | INDIRECT | VERY LOW |
| Ocean circulation (AMOC) | INDIRECT | VERY LOW |