PULSE_

anomaly

the difference from a long-term average, not the value itself. “+1.2 °C” means 1.2 °C warmer than that series' baseline. Anomalies are how different places and eras become comparable.

baseline

the multi-decade reference period an anomaly is measured against. Pulse names it on every chart (temperature vs 1951–80, sea ice vs 1981–2010, snowpack vs 1991–2020) because the same reading looks bigger or smaller against a different baseline.

σ (sigma)

a yardstick for “how unusual”: one standard deviation (1σ) covers most normal year-to-year wobble. A reading 2σ from average is genuinely unusual; 3σ or more is rare.

pre-industrial

the climate of roughly 1850–1900, before fossil-fuel emissions took off — the zero point for “+1.5 °C” and the Paris limits.

ppm / ppb

parts per million / parts per billion — out of every million (billion) air molecules, how many are CO₂ (methane).

ONI

Oceanic Niño Index — a three-month average sea-surface-temperature anomaly in the Niño 3.4 region of the Pacific. ONI ≥ +0.5 °C is El Niño (warmer than usual), ≤ −0.5 °C is La Niña (cooler).

SWE

snow-water equivalent — the depth of water the snowpack would yield if it melted. It is the water-supply number; deep fluffy snow can hold less water than shallow dense snow.

DHW

degree heating weeks — accumulated heat stress on a coral reef over the last 12 weeks. Around 4 °C-weeks significant bleaching becomes likely; around 8, severe bleaching and death.

FRP

fire radiative power — the energy a fire radiates, in megawatts, as measured from satellite. It ranks fire intensity, not area burned.

MAF

million acre-feet — reservoir volume; one acre-foot is an acre of water one foot deep (~1.23 million litres). Lake Mead holds about 26 MAF when full.

dead pool

the reservoir level below which water can no longer pass through the dam by gravity. Hydropower fails earlier, at the minimum power pool.

D0–D4

the Drought Monitor's severity ladder, D0 “abnormally dry” to D4 “exceptional drought”. The classes nest, so “D2+” always means severe or worse.

Every chart on the dashboard speaks the same six verbs. Hover reads — the crosshair follows your pointer, and everything it shows is also in each chart's TABLE view. Drag frames — sweep a window to zoom (or ⌘/Ctrl + wheel, or pinch); a minimap bar appears underneath for panning. Click marks — on a time-series, two clicks pin A and B and a Δ chip states the change (green when the change is good for that instrument); on a spaghetti chart (sea ice, snowpack), hovering names a year and clicking locks it — ⇧click compares up to three. Any non-default view grows a COPY LINK button so the exact framing can be shared. Esc always unwinds. Practice on the synthetic series below — the real instruments work identically.

DEMO · SYNTHETIC PRACTICE SERIES — DRAG A WINDOW · CLICK TWICE FOR Δ

DEMO · SPAGHETTI — HOVER A STRAND TO NAME IT · CLICK LOCKS · Y CYCLES

KEYBOARD — EVERY VERB HAS A KEY PATH

← →

scrub the cursor (⇧ steps ×12)

[ ]

zoom out / in around the cursor

{ }

pan one window left / right (while zoomed)

0

reset to the full extent

P

drop a pin at the cursor — A, then B, and the Δ chip lands between them

Y / ⇧Y

cycle the named year on a spaghetti chart (sea ice, snowpack)

Esc

unwind, in order: pins → zoom → leave the chart

J / K

jump down / up the page, section by section

⌘K / Ctrl-K

command palette — every instrument, explainer, and route

⌘/Ctrl + wheel

zoom at the pointer (a trackpad pinch does the same)

Charts are keyboard-first friendly: tab to any chart, then use the keys above. Everything a tooltip shows also lives in the chart's TABLE twin — hover never gates information.

What it measures

Sea-surface temperature in the Niño 3.4 box (5°N–5°S, 120°–170°W) — the standard gauge of El Niño / La Niña.

Why it matters

ENSO is the largest year-to-year climate signal on Earth; it shifts rain, drought, hurricanes and global temperature.

How to read the chart

Watch the anomaly line against the ±0.5 °C thresholds: sustained above = El Niño, below = La Niña, between = neutral. The ONI 1950→ view unrolls the whole owned index — every El Niño and La Niña for 75+ years, shaded warm/cool, with the strong episodes (1972–73, 1982–83, 1997–98, 2015–16 …) labeled at their peaks.

Thresholds & terms

El Niño (La Niña) is declared when the 3-month ONI stays ≥ +0.5 °C (≤ −0.5 °C) for 5 consecutive seasons; weekly values move first.

Record

Weekly since 1990 (dataset starts 1981). · Source: NOAA CPC ↗

What it measures

The average temperature of the ocean surface between 60°S and 60°N, updated daily — NOAA's OISST v2.1 satellite+buoy analysis, as compiled by the University of Maine's Climate Reanalyzer.

Why it matters

The ocean absorbs over 90% of the heat greenhouse gases trap, so this single curve is the planet's thermometer with the weather smoothed out. When it broke away from every previous year in 2023–24 it marked the largest ocean heat excursion in the satellite era — the backdrop for coral bleaching, marine heatwaves and hurricane fuel elsewhere on this page.

How to read the chart

A spaghetti chart: every year since 1981 is a faint strand over the annual cycle; the gray band is the 1981–2010 normal ±2σ; this year is bold, with last year and the record-warm year labeled. Hover names a strand, click locks it. ANNUAL EXTREMES reduces each complete year to its warmest and coolest day with fitted trends in °C per decade.

Thresholds & terms

±2σ ≈ a 1-in-20 year by the old normal's odds — the 2023–24 strands run far above the band for months at a time. The seasonal peak lands in March–April (the Southern Hemisphere's ocean summer dominates the global mean).

Record

Daily since September 1981, with a 1–2 day lag; the freshest ~two weeks are NOAA-preliminary values, quietly revised as the final analysis lands. This source blocks datacenter addresses, so readings arrive via Pulse's residential relay (the Pi fetches, the hub stores) — the full history rides in every push. · Source: Climate Reanalyzer · NOAA OISST v2.1 ↗

What it measures

Daily sea-surface temperature in the Niño 3.4 box (5°N–5°S, 120°–170°W) — the same patch of Pacific the ENSO instrument above tracks, but at daily resolution instead of weekly averages.

Why it matters

ENSO state is declared on weekly and 3-month means, and those smoothings hide the turn. The daily curve is where an arriving El Niño or a collapsing La Niña shows first — watch it bend away from its own recent strands weeks before the ONI catches up.

How to read the chart

Same grammar as the global SST chart: every year since 1981 is a faint strand, the gray band is the 1981–2010 normal ±2σ, this year is bold with the record-warm year labeled. Read it as absolute temperature (the ENSO widget shows the anomaly): the annual cycle peaks in boreal spring, and El Niño years (1997, 2015, 2023) arc high above the band while La Niña years hug its floor. ANNUAL EXTREMES shows each year's warmest and coolest day.

Thresholds & terms

The El Niño / La Niña thresholds are defined on anomalies (±0.5 °C vs a 30-year normal), not on this absolute scale — use the band: strands above it are warm events, below it cool events. Daily values are noisy; declarations wait for the weekly and seasonal means.

Record

Daily since September 1981 with a 1–2 day lag (the freshest ~two weeks are NOAA-preliminary, revised as the final analysis lands), via the same residential relay as the global SST — the source blocks datacenter addresses; the full history rides in every push. · Source: Climate Reanalyzer · NOAA OISST v2.1 ↗

What it measures

Global mean sea level from satellite altimetry — the TOPEX/Poseidon → Jason → Sentinel-6 reference missions, merged by the University of Colorado Sea Level Research Group. Millimetres relative to the mission-era reference mean, seasonal signals removed, corrected for glacial isostatic adjustment (GIA).

Why it matters

Sea level is the planet's most unforgiving vital sign: it integrates ocean heat (water expands as it warms) and land-ice loss into one number that never 'cools back' the next winter. Roughly 110 mm of rise since 1993 sounds small — but every 10 mm of global mean reaches decades of coastal flood frequency.

How to read the chart

The thin line is each ~10-day satellite cycle; the bold line is a 60-day smooth. The banner states the rise since 1993 and the rate: the first altimetry decade ran near 2.8 mm/yr, the most recent decade near 3.9 — the rise is accelerating, not just continuing. The RATE view plots that acceleration itself: the fitted rate over each trailing decade, evaluated along the whole record — a rising curve means the rise is speeding up.

Thresholds & terms

No official threshold — the signal is the rate and its acceleration. GIA correction adds ~0.3 mm/yr versus uncorrected products, so quoted trends differ slightly between groups; this chart states its own basis.

Record

Full record 1992→ arrives with every release, so Pulse owns the whole series; new releases (a few per year, ~2–3 months of processing lag) revise history slightly and the collector upserts in place. The data-age chip honestly trails today by that lag. · Source: Univ. of Colorado Sea Level Group ↗

What it measures

NOAA Coral Reef Watch's daily bleaching heat-stress board: 214 'virtual stations' — reef sites worldwide watched by 5 km satellite sea-surface temperature — each carrying today's alert level and its accumulated heat stress (DHW, degree heating weeks).

Why it matters

Marine heatwaves bleach corals, and tropical reefs are the most exposed tipping element on the board upstairs (~1.5 °C central estimate) — a quarter of ocean life depends on them. This is the live telemetry of that pressure: where the ocean is hot enough, right now, to be killing reefs.

How to read the chart

Map dots are stations coloured by alert level — yellow WATCH (warm), orange WARNING (HotSpot ≥ 1 °C), red ALERT LEVEL 1 (DHW ≥ 4 — significant bleaching expected), dark-red ALERT LEVEL 2 (DHW ≥ 8 — severe bleaching, mortality likely). The STATIONS view ranks the reefs under stress by DHW; each links to its official CRW gauge page.

Thresholds & terms

DHW adds up how far SST has run above the reef's bleaching threshold over the trailing 12 weeks, in °C-weeks. 4 is the bleaching line, 8 the mortality line. The product date runs ~2 days behind (satellite processing).

Record

The live board is replaced each sweep; the daily station counts per alert level join the owned history from 2026-07-14 — a stress-trend view arrives once seasons of history accrue. · Source: NOAA Coral Reef Watch ↗

What it measures

Accumulated Cyclone Energy (ACE) — one number for how much tropical cyclone activity a season has actually produced, summed from every storm's wind speed every six hours while it is at tropical-storm strength or above.

Why it matters

Counting storms is a poor measure of a season: twelve brief tropical storms and four long-lived major hurricanes can carry the same count and wildly different energy. ACE weights by intensity AND duration, so it captures what a season really did — and unlike an active-storm board, it still says something in February.

How to read the chart

The curve is this season's running total by day of year, against the 1991-2020 normal band and every season back to 1980 as faint context. The basin table gives all six basins side by side. Northern basins run with the calendar year; the southern Indian and south Pacific seasons straddle the new year, so their totals are labelled calendar-year to date rather than presented as season totals.

Thresholds & terms

Computed from this hub's own backfilled record over the 1991-2020 normals window: the Atlantic averages 122.5 per season (NOAA publishes ~123 for the same window, which is the cross-check) with a median of 129.3 and a range of 32 to 245. Globally the six basins together average 789 a year, from 525 to 1,169. The west Pacific is the busiest basin by far at ~302.

Record

Daily accumulation per basin, backfilled to 1980 (the satellite era, where IBTrACS coverage is uniform) and kept current daily — 119,077 owned readings. Validated against published seasons before it shipped: 1995 -> 227.4, 2005 -> 245.3, 2017 -> 224.9, 2020 -> 180.4, 2024 -> 161.6, all within ~2%. Values for the running season are provisional and revise as agencies finalise tracks. · Source: NOAA NCEI IBTrACS v04r01 ↗

What it measures

Water-surface elevation of Lake Powell behind Glen Canyon Dam — the Upper Basin's storage bank for its Colorado River Compact delivery obligations.

Why it matters

Below minimum power pool Glen Canyon Dam stops generating for ~5 million customers, and the river approaches the physical limits of moving water downstream at all.

How to read the chart

The line is pool elevation vs the dam's operating thresholds (dashed). Header storage % compares contents to 24.32 MAF capacity. COMPARE YEARS overlays whole years by month — watch the spring snowmelt pulse arrive (or fail to). STORAGE 1963→ swaps feet for contents: the whole owned record in million acre-feet against full capacity — the 17-year fill, the 1983 near-spill, the post-2000 decline.

Thresholds & terms

Full pool 3,700 ft · minimum power pool 3,490 ft (turbines starve below it) · dead pool 3,370 ft — river outlet works only.

Record

Daily since December 1963 (first fill behind the new dam). · Source: USBR hydrodata (site 919) ↗

What it measures

Water-surface elevation of Lake Mead — the largest US reservoir, formed by Hoover Dam on the Colorado River and supplying ~25 million people.

Why it matters

The single clearest gauge of the Colorado River crisis: two decades of megadrought plus overallocation drained it to ~27% full by 2022. Federal shortage declarations key off this number.

How to read the chart

The line is elevation above sea level; dashed rules are the operating thresholds. Shortage tiers cut state allocations as the lake falls through 1,075 / 1,050 / 1,025 ft. COMPARE YEARS overlays whole years by month (the 'bathtub ring' in chart form).

Thresholds & terms

Full pool 1,229 ft · Tier 1 < 1,075 · Tier 2 < 1,050 · Tier 3 < 1,025 (tiers are set by the *projected* Jan-1 level under the 2007 Guidelines/DCP) · dead pool 895 ft, below which water cannot pass Hoover Dam.

Record

End-of-month elevations since February 1935 (first fill); daily values for the current month. · Source: USBR Lower Colorado daily report ↗

What it measures

Snow water equivalent (SWE) — the depth of water the snowpack would yield if melted — averaged across the 136 SNOTEL telemetry stations of the Upper Colorado River Basin (hydrologic region 14).

Why it matters

This snow IS the Colorado River: roughly two-thirds of the river's flow starts as Upper Basin snowpack, so a bad snow year here becomes next year's drop in Lake Powell and Lake Mead. April 1 SWE is the West's classic water-supply forecast date.

How to read the chart

A water-year chart (Oct 1 → Sep 30): this year's basin-mean SWE against the 1991–2020 median curve (dashed), last year, the worst season on record, and every backfilled year as faint context. The banner states % of median in season — and says MELTED OUT honestly in summer instead of quoting a % of a near-zero median. PEAKS 1991→ lines up every season's peak SWE as bars against the median of season peaks; incomplete seasons render dim so they can never read as low years.

Thresholds & terms

No official thresholds — the reference is the median curve itself. Peak normally lands in early April; peaking early AND low (like WY2026's 66%-of-median peak on Mar 8) is the drought double-hit: less snow, gone sooner. The basin index is Σ SWE ÷ Σ median over stations reporting both.

Record

Daily basin aggregate backfilled to water year 1991 (the normals period start); station records reach the 1960s. Kept fresh daily from the AWDB API. · Source: USDA NRCS SNOTEL (AWDB) ↗

What it measures

Daily mean streamflow (discharge, in cubic feet per second) at the three gauges that define the Colorado River's plumbing: COLORADO RIVER AT LEES FERRY, AZ — the compact point just below Glen Canyon Dam, where Lake Powell's releases become the Lower Basin's supply — plus the river's two big Powell inflows, COLORADO RIVER NEAR CISCO, UT and GREEN RIVER AT GREEN RIVER, UT.

Why it matters

This is the middle of the water story the rest of this section tells: snowpack melts into these gauges, these gauges fill Lake Powell, and Lees Ferry is what leaves it. Lees Ferry is also the legal hinge of the whole basin — the 1922 Colorado River Compact splits the river there, and the Upper Basin's delivery obligation is measured as a running 10-year total past this gauge. Cisco + Green River together carry most of what ever reaches Powell.

How to read the chart

The hydrograph plots this water year's daily flow against the median and the 10th–90th percentile band of the regulated era, so 'normal' means normal for a dammed river. VOLUME sums each water year to a total in acre-feet — the number water managers actually argue about — with the compact's 7.5 MAF/yr reference on Lees Ferry. Note the two eras: before Glen Canyon Dam closed in 1963, Lees Ferry ran wild (a spring snowmelt spike an order of magnitude above winter baseflow); after, it is a managed release schedule, which is why the climatology is computed from 1964 on.

Thresholds & terms

No 'safe' line — the references are the regulated-era median, and at Lees Ferry the 7.5 MAF/yr the Compact contemplates (the Upper Basin's obligation is 75 MAF per 10 years). Flow is reported in cubic feet per second; 1 cfs sustained for a year ≈ 724 acre-feet. Reservoir releases mean Lees Ferry is a policy signal as much as a weather one — Cisco and Green River are the closer read on what nature actually delivered.

Record

Daily record owned from water year 1895 for Green River (1894-10-01 — 132 years), 1914 for Cisco, and 1922 for Lees Ferry; backfilled once from the USGS OGC API, then kept fresh twice a day. The most recent day or two is Provisional until USGS review. · Source: USGS Water Data ↗

What it measures

Drought from two directions, because no single product does both. WORLDWIDE: the Copernicus Global Drought Observatory's combined drought indicator, published as events through GDACS — where drought is currently being analysed on Earth, how much land each one covers, and an official report page for each. IN THE US: the weekly US Drought Monitor, the most detailed drought map made anywhere — author-drawn each week by rotating experts who combine rain and snow deficits, streamflow, soil moisture, reservoir levels and on-the-ground reports into five severity classes (D0 abnormally dry, D1 moderate, D2 severe, D3 extreme, D4 exceptional).

Why it matters

Drought is the slow disaster behind the reservoirs, the wildfires and the crop losses on the rest of Pulse — and it is not a US story. The Monitor is what American federal drought aid, water restrictions and agricultural disaster declarations key off; the GDO events are how the EU's emergency services see the same hazard everywhere else, from the Horn of Africa to the Amazon. Seeing them side by side is the point: the western US megadrought is one entry on a much longer list.

How to read the chart

MAP is the world: each event's affected-area footprint shaded on the basemap, with a ◆ marker at its centre — click or tap any marker (or press N and P to step through them) for that event's countries, area, GDO assessment and a link to the official report. WORLD lists the same events ranked by area, and ◎ on any row jumps the map to it. US is the Drought Monitor choropleth, each area painted its worst current class on the D0→D4 ramp, with a ⬤ on every state — click one for that state's own D0–D4 breakdown and its change since last week; the STATES cut ranks all 52 jurisdictions. TREND shows the percentage of the Lower 48 in drought over the full record ('D1+' is moderate-or-worse, the usual 'in drought' headline; 'D3+' is extreme-or-worse); its STACK cut paints all five classes at once, light→dark, so the band between two curves is exactly the share of the country in that one class — 2012 and 2021–22 rise like mountain ranges.

Thresholds & terms

D0 abnormally dry (not yet 'drought' — recovering or heading in) · D1 moderate · D2 severe · D3 extreme · D4 exceptional (rare, the worst class). US percentages are CUMULATIVE — 'D2+' means severe OR worse, so the classes nest. THE TWO HALVES ARE NEVER MIXED, and they are not measuring the same thing: the Drought Monitor is author-drawn and covers every kind of drought impact, while GDO's indicator is model-derived and scores AGRICULTURAL drought — crop-relevant moisture deficit — so a country can appear on one and not the other without either being wrong. No US percentage is ever added into a world total. The world board also shows only the CURRENT analysis period: GDO re-runs about every 10 days and GDACS keeps finished events in the same list, so events whose last period has passed are counted in the footer and never mapped or summed. The US map is the contiguous states only — Alaska, Hawaii and Puerto Rico are in the statistics and the STATES list but off that frame.

Record

The Drought Monitor has been released every Thursday since 2000 and the trend series carries the full 2000→now history (the collector keeps it even as the upstream window moves). The world board began accruing on 2026-07-25. Both map files are reduced SERVER-SIDE before they reach your browser — the Monitor's 19 MB of polygons to about 45 KB, and roughly a megabyte of world drought geometry to about 6 KB — so the shapes are smoothed for a continental view, not survey-accurate at street scale. · Source: US Drought Monitor (NDMC / USDA / NOAA) · GDACS / Copernicus GDO ↗

What it measures

Three views of fire on Earth, from three authorities. NASA FIRMS gives the satellite thermal-anomaly detections — every spot the VIIRS instruments on NOAA-20 and NOAA-21 saw burning in the last 24 hours, worldwide, at 375 m resolution. The two satellites fly the same orbit about 50 minutes apart, so between them a fire is looked at roughly twice as often as one satellite manages, and the extra detections sharpen the outline of every fire on the map. GDACS, the EC/UN global disaster alert system, gives the world's tracked fire events with an estimated burnt area in hectares, the country, and a report page for each. The US National Interagency Fire Center gives the named American incidents with acreage, containment and assigned personnel. No one source covers the planet: NIFC names fires only inside the US, GDACS publishes only the largest hundred, and the satellite sees everything but names nothing — so all three are on the board, and every row says which one it came from.

Why it matters

Fire is where several Pulse instruments meet: heat and drought dry the fuels, fire weather ignites them, and the smoke and carbon feed back into the atmosphere the CO₂ chart is tracking. Hotter, drier summers are lengthening fire seasons and growing burned area across much of the world — the boreal fires routinely visible on this map at high latitudes are themselves a climate signal.

How to read the chart

The MAP bins detections into half-degree cells coloured by total fire radiative power (FRP, in megawatts) — how much energy the fires in that cell are radiating; brighter cells are bigger, hotter fire complexes. Over that grid sit three kinds of marker, each a different SHAPE so you never have to read colour alone: a filled dot ⬤ is a satellite fire cluster, a diamond ◆ is an official GDACS fire event, a ring ○ is a named US incident. Click, tap or focus any marker and a popup gives everything that source knows about that fire — burnt area, containment, personnel, intensity, footprint, and the report link where one exists. Press N and P to step through every marker from the keyboard, Escape closes; OFFICIAL and SATELLITE switch the two marker families off and on. WORLD lists the same fires as text — OFFICIAL EVENTS ranks the globe by burnt area with a link to each GDACS report, SATELLITE CLUSTERS ranks by radiative power and names the country each is burning in — and ◎ on any row jumps the map to that fire with its popup open. US ranks the American board by acreage with containment percentages, where 100% contained shows green.

Thresholds & terms

Honest framing, and this instrument needs more of it than most. Detections are HOTSPOTS, not fire perimeters — anything hot enough triggers one, including gas flares, volcanoes and industrial heat (the persistent dots over oil fields are not wildfires). A large share of the African, Australian and South American signal is agricultural and savanna burning: seasonal land management, not disaster — so the biggest cluster on the board is often not the worst fire on Earth. Satellite clusters are built by grouping neighbouring detections, which means one long burning front reads as a single very wide cluster; each one reports how many kilometres across it is, so a 300 km 'cluster' declares itself as a landscape-scale burn front rather than posing as one fire. A cluster is never labelled with a nearby fire's name — proximity is not identity, and a popup lists nearby official reports with their distances instead, for you to judge. Low-confidence detections (a sun-glint artifact class, roughly 9% of rows) are dropped before anything is counted. Clusters are built from both satellites together, and a cell needs at least two detections before it can extend a cluster outward — without that rule the denser two-satellite view welds neighbouring fires into continent-sized blobs through threads of stray detections. The 24 h file is a rolling window; the satellite passes each spot about twice a day, and cloud or smoke can hide fires beneath it. FRP is energy radiated, not area burned; GDACS hectares are a modelled estimate, not a survey.

Record

The reduced map is replaced each sweep; the daily global totals (detections, ΣFRP, cluster count) and US totals (incidents, acres) join the owned history from 2026-07-14, with the world event count and total burnt hectares added 2026-07-25. Each satellite keeps its OWN daily totals rather than being added together: the NOAA-20 series has run unbroken since 2026-07-14, so NOAA-21 (added 2026-07-25) records separately and the combined figure you see in the banner is the two summed — a record that suddenly doubled because a second satellite arrived would be worse than useless — a fire-season trend view arrives once seasons of history accrue. · Source: NASA FIRMS (VIIRS NOAA-20 + NOAA-21) · GDACS · NIFC WFIGS ↗

What it measures

Sea ice extent — the ocean area with at least 15% ice concentration — for the Arctic and Antarctic, measured by passive-microwave satellites (NSIDC Sea Ice Index v4).

Why it matters

Ice is the planet's mirror: bright ice reflects sunlight that dark open water absorbs, so losing it accelerates the warming that melted it. Arctic September extent is down roughly 40% since 1979.

How to read the chart

A spaghetti chart: every year since 1978 is a faint line over the annual cycle; the gray band is the 1981–2010 normal ±2σ; this year is the bold blue line, with last year and the record-minimum year labeled. Outside the band = outside ~95% of the old climate's range. MIN / MAX 1979→ reduces each complete year to its annual minimum and maximum extent with fitted trends, stated the NSIDC way as % per decade vs the 1981–2010 mean — the Arctic September minimum is the fastest-shrinking number on this site.

Thresholds & terms

±2σ ≈ a 1-in-20 year by the old normal's odds. The record-minimum year (Arctic: September 2012) marks how far a bad year can go.

Record

Daily since late 1978, both poles. The Antarctic band is computed from the same daily record (no official v4 climatology file). · Source: NSIDC Sea Ice Index ↗

What it measures

Globally averaged surface CO₂ concentration, in parts per million, from NOAA's marine boundary layer sampling network.

Why it matters

CO₂ is the climate's main control knob and its longest lever — a large share of each year's emissions stays in the air for centuries. Pre-industrial was ~278 ppm; we are ~54% above it.

How to read the chart

TREND is the deseasonalized curve (the headline number). SEASONAL keeps the annual sawtooth — the Northern Hemisphere growing season drawing carbon down each summer, the planet 'breathing'. The vertical gap between successive years is the growth rate — and the GROWTH view turns exactly that into bars, each full year's increase in ppm/yr (the owned daily record starts 2016; El Niño years usually grow faster).

Thresholds & terms

There is no safe line, only milestones: 350 ppm (the climate-stability target that named a movement) · 400 passed globally in 2015 · growth ≥ 2 ppm/yr means accelerating, not just rising.

Record

GML global daily product from 2016 here; the Mauna Loa record extends the curve to 1958. · Source: NOAA GML ↗

What it measures

Globally averaged surface methane (CH₄) concentration, in parts per billion, from NOAA's global cooperative air-sampling network — the deseasonalized trend plus the monthly mean that carries the seasonal cycle.

Why it matters

Methane is the second-largest driver of human-caused warming after CO₂. Molecule for molecule it traps far more heat — roughly 80× CO₂ over 20 years — but it breaks down in about a decade, which makes cutting it the fastest lever on near-term warming. After a near-plateau from 1999–2006 it started rising again and has accelerated since 2020, for reasons still being untangled (wetlands, fossil fuels, a shifting atmospheric sink).

How to read the chart

TREND is the deseasonalized curve (the headline). MONTHLY keeps the seasonal wiggle. Pre-industrial methane was ~722 ppb, so today's ~1,930 ppb is about 2.7× — a bigger relative jump than CO₂'s. Growth is the change versus twelve months earlier. The GROWTH view shows each year's increase since 1984 as bars: the 1999–2006 plateau, then the renewed — and lately near-record — rise.

Thresholds & terms

No safe line — the signal is the growth rate. Recent years near +10–15 ppb/yr are among the fastest in the record; the 1999–2006 plateau was near zero. It's a global mean, so it lags any single site by weeks.

Record

Monthly since 1983; the collector keeps the full record even as the upstream file's window moves. · Source: NOAA GML ↗

What it measures

Global mean surface temperature anomaly — land and ocean combined — versus the 1951–1980 average, from NASA GISTEMP v4.

Why it matters

The headline climate number. The Paris targets (1.5 °C / 2 °C) are defined on long-term means of exactly this kind of series.

How to read the chart

The thin line is monthly (noisy by nature); the bold line is the 12-month running mean. The stripe strip below compresses the whole record into color — find your birth year. Baselines matter: this chart's 1951–80 zero sits ~0.3 °C above pre-industrial, so +1.2 °C here ≈ 1.5 °C vs pre-industrial. The GRID view spreads every month since 1880 into a year × month heatmap on the same blue↔red scale — hover any cell for its rank ('2nd-warmest MAR on record').

Thresholds & terms

El Niño years spike the monthly line; the running mean is the honest trend. Sustained ≥ +1.2 °C on this baseline is the Paris 1.5° neighborhood.

Record

Monthly since January 1880. · Source: NASA GISS ↗

What it measures

Sixteen parts of the climate system that scientists assess could pass a self-reinforcing threshold — a 'tipping point' — beyond which change keeps going on its own and is often effectively irreversible on human timescales. They include the Greenland and West Antarctic ice sheets, the Atlantic overturning circulation (AMOC), the Amazon rainforest and tropical coral reefs.

Why it matters

These are the step-changes hiding behind the smooth curves elsewhere on Pulse. Crossing one doesn't read as a gentle trend — it commits the Earth system to large, often irreversible shifts: metres of sea-level rise, abrupt regional climate change, big carbon releases. How close today's warming sits to each assessed threshold is the single most important piece of context for the temperature record.

How to read the chart

Each row is one element on a shared axis of global warming above pre-industrial. The BAR is the assessed threshold range (lowest to highest estimate); the NOTCH is the central best estimate. The bright vertical line is current warming, and it crosses every row. A row turns red (▲ IN RANGE) once current warming reaches that element's LOW estimate, amber (◆ NEARING) when it is within 0.5 °C of it, and stays dim (● BELOW) otherwise. Grouping — cryosphere, ocean circulation, biosphere — carries the domain; colour carries only the state. Select any row for its mechanism, timescale and consequences.

Thresholds & terms

Honest framing is the whole design. These are ASSESSED TEMPERATURE THRESHOLDS, not forecasts and not probabilities — there is no live 'tipping telemetry'. 'In range' means warming has reached the earliest estimate at which tipping becomes possible, NOT that the element has tipped. Each estimate carries real uncertainty (hence a range), and the transitions unfold over decades to millennia (shown per element). Current warming here is a single 12-month mean (NASA GISTEMP, +0.3 °C to shift its 1951–80 baseline to pre-industrial); the Paris 1.5 / 2 °C limits are assessed on ~20-year means, so one year touching 1.5 °C is not the same as breaching Paris.

Record

Thresholds from the peer-reviewed synthesis of 16 tipping elements, cross-checked against the 2023 Global Tipping Points Report (https://global-tipping-points.org/). Report-driven — updates when the science updates, not on a fetch schedule; current warming refreshes monthly with GISTEMP. · Source: Armstrong McKay et al. 2022, Science ↗

What it measures

Five live hazard feeds merged into one severity-ranked stream: active tropical cyclones (NHC), US Severe/Extreme weather alerts (NWS), open natural events (NASA EONET), global disaster alerts (GDACS), and earthquakes ≥ M2.5 in the last 24 h (USGS).

Why it matters

Climate change loads the dice for exactly these events — the wall is the 'happening now' half of the dashboard, next to the slow curves that explain why the dice are loaded.

How to read the chart

Rows sort by severity, then recency; the glyph + label carry severity (▲ CRITICAL / ◆ WARNING / ● WATCH / · INFO), never color alone. The ↗ on a row — and the announcement button in the detail card — opens the authoritative source in a new tab: the NHC advisory, the InciWeb/IRWIN fire report, the GDACS event report, and the USGS event page each go straight to that specific event; NWS retired its per-alert pages (alerts.weather.gov), so a storm-based alert opens the NWS point-forecast for its location — which shows the live 'Hazardous Weather Conditions' page; a zone-based alert has no public human page (its only record is a machine-readable data feed), so those rows carry no link rather than handing you a raw file to download — better no link than one that dumps JSON. Either way you go from 'something is happening' to the source in one click. The list and the map share one selection: hover, tap, focus or arrow to a row and its map dot rings, and the card beneath the map fills in with coordinates and the link. The map is a real map — zoom with + / − (or double-tap, or pinch), pan by dragging or with the arrow keys, FIT frames the current events, ⌂ resets; dots stay the same on-screen size at every zoom so size keeps meaning magnitude. Because the dots hold their size, events in the same spot — an aftershock swarm, a cluster of alerts — would overlap into one blob, so they merge into a single numbered badge coloured by the worst event in it; zoom in and the badge splits back into its members, and a swarm packed too tightly for any zoom to separate fans out when you click it so you can still pick any single event from it. A computed day/night shade shows where it is currently dark (pure geometry, no data feed). Filter the stream by type (storms · alerts · quakes · fires) and by a time window (last 1 / 6 / 24 hours) to sharpen 'happening now' — the window filters on each row's own timestamp, the same age the row shows. SHARE VIEW copies a link that reopens exactly what you're looking at — the filter, the time window, the selected event and the map framing; an event that has since aged out simply reopens with nothing selected rather than pointing at the wrong one. NWS alerts are area-based, so they stay in the list. 'No active tropical cyclones' is a normal, good state — the wall says so rather than showing an empty box.

Thresholds & terms

CRITICAL = GDACS Red, NWS Extreme, or a hurricane · WARNING = GDACS Orange, NWS Severe, tropical storm, or M ≥ 6 · WATCH = M5–6, volcanoes, storm systems · INFO = routine monitoring (most wildfires, GDACS Green, smaller quakes).

Record

Live only — each sweep replaces the last (15–30 min per source); Pulse keeps history for the trend instruments, not the event stream. · Source: NHC · NWS · EONET · GDACS · USGS ↗

What it measures

Headlines from four hand-picked climate desks — The Guardian's climate-crisis wire, Carbon Brief's analysis, Yale Environment 360's features, and NASA Earth Observatory's daily earth-observation stories — merged into one stream. (The NASA slot originally carried the agency's general breaking-news wire; it was swapped for the Earth Observatory desk when Artemis ceremonies started pinning as BREAKING on a climate page.)

Why it matters

Numbers say what the planet is doing; reporting says what people are doing about it. Four outlets with different editorial angles beat any single homepage.

How to read the chart

Newest first — time is the only ranking; there is no engagement algorithm here. Duplicate stories are folded by title. A story pins as BREAKING only while it is genuinely fresh (≤ 2 hours old).

Thresholds & terms

Feeds are fetched every 45–60 minutes with an identifying User-Agent; two candidate feeds (Inside Climate News, NOAA) refused automated readers and were excluded rather than scraped.

Record

Whatever each outlet's RSS window carries (typically the latest 10–50 stories). · Source: Guardian · Carbon Brief · Yale E360 · NASA EO ↗

What it measures

A standings band, derived — not fetched. Each tile ranks the latest reading against the whole history Pulse has stored for that series: the warmest month in the 146-year temperature record, the Arctic's rank for today's date across ~48 satellite years, how far each Colorado reservoir sits above its drought low, where the Upper Colorado snowpack's season peak lands among 36 owned water years, and how close CO₂, methane, and global sea level sit to their next round milestones (sea level counts millimetres above its 1993 altimetry baseline — a rise that never gives back).

Why it matters

This is the payoff of owning the history (Pulse keeps every reading even as upstreams trim their windows). A single number — '427.8 ppm' — means more when you can see it's higher than at any point in 800,000 years, or that the globe has run 37 straight months at least 1 °C hot.

How to read the chart

Every claim is a fact about the OWNED data, computed on each load — nothing is invented. A tile says '43rd-lowest for this date' when that's the truth and 'record low' only when it genuinely is, so the strip is honest in the quiet months and lights up when a real record lands (the Arctic tile turns dramatic each September). Two honesty guards are baked in: a reservoir's 'record low' skips its initial filling era — Lake Mead's raw minimum is 700 ft while Hoover Dam was still filling it in the 1930s, which is not a drought — so 'modern low' is measured only after the lake first reached full pool; the snowpack tile claims a record season peak only once the season is FINAL (a mid-winter year can't fake a low peak, and an incomplete past season can't set one); and temperature records name the baseline (anomalies are vs 1951–80, the streak counts months ≥ +1.0 °C).

Thresholds & terms

MILESTONE = the next round mark ahead (CO₂) · RECORD / RECORD LOW = a genuine all-time extreme in the owned series · NEAR-LOW = within reach of the modern low · BELOW/ABOVE AVG = the day-of-year standing when it isn't near a record. Emphasis (a brighter tile) marks the records and near-records so the notable ones read first.

Record

Recomputed live from the SQLite history on every page load; the ranges are as deep as Pulse's stored series — temperature to 1880, sea ice to 1978, Powell to 1963, Mead to 1935, sea level to 1992. · Source: Derived from Pulse's owned history (NOAA · NASA · NSIDC · USBR) ↗