Equatorial Pacific monitor

Daily ENSO state, warm-water volume, the Indian Ocean Dipole and Atlantic genesis windows — every number with its baseline stated, and a public forecast record (2 open).

Over August–October 2026 the tropical Pacific was in a strong El Niño on the index NOAA now publishes (+1.97 °C), and a very strong El Niño on the older one it replaced (+2.46 °C).The +0.49 °C difference is background warming of the whole tropical ocean, which the relative index removes. NOAA publishes each 3-month value early in the following month.

What the satellite saw todayNiño 3.4, one day+3.36°Cdaily satellite anomaly · 90-day mean +2.99 °C

What NOAA officially publishesNOAA official index (RONI) · August–October 2026+1.97°Cstrong El Niñothree-month mean, published monthly

The traditional index (ONI) reads +2.46 — very strong El Nino. The +0.49 between them is background tropical warming, which the relative index removes and the traditional one counts as El Niño.

These are not the same measurement. The large number is a single day from satellite, against a fixed 1991–2020 baseline. NOAA's are three-month means of monthly values, centred two months earlier, and the relative one also removes the warming of the tropics as a whole. A daily value sits above a seasonal mean by construction.

observed 2026-11-02
bulletin 2026-11-04 11:39 UTC

The Pacific right nowHow warm the ocean is, how much heat it is holding, and which way it is going

What this means today

On NOAA's official index this is a strong El Nino at +1.97 (ASO 2026). NOAA switched its official ENSO monitoring to the relative index in February 2026, because the traditional one counts background tropical warming as El Niño. For ASO 2026 the traditional index reads +2.46 (very strong El Nino); the +0.49 between them is that warming.
The trajectory is warm and recharged, recharging. Warm water volume has changed +0.64×10¹⁴ m³ over 90 days while Niño 3.4 stands at +3.36 °C. This is the quadrant a warm event peaks in; the basin still holds the heat that sustains it.
Niño 3.4 is warming. -0.13 °C over a week, +0.07 over a month, +0.81 over three months. Against the 1991–2020 day-of-year climatology.
No genesis window is open. All monitored boxes are above their calibrated ventilation-index threshold.
WMO OND HOTSPOT is 144% of normal so far this season. Day 33 of 92 of its October–December season.
56% of today's Niño 3.4 is basin-wide tropical warmth, not ENSO structure. The tropics (20°S–20°N) are +1.90 °C above the 1991–2020 average. Subtracting that leaves +1.46 °C, or +2.04 °C rescaled to NOAA's relative index — a daily proxy for it, not the published value.
Today's satellite Niño 3.4 is +3.36 °C. A daily anomaly against the 1991–2020 climatology. It sits above any seasonal index by construction: NOAA's categories are defined on a three-month mean, and the 90-day mean here is +2.99 °C.

Surface

Niño 3: +4.49°C
Niño 4: +1.51°C
Niño 1+2: +5.80°C
Niño 3 − Niño 4: +2.98°C
90-day running mean: +2.99°C

OISST anomalies against the PSL 1991–2020 day-of-year climatology

nino34_run is a 90-day running mean — a proxy for the ONI, not the ONI. The official index is a 3-month running mean of monthly ERSSTv5 anomalies.

The Pacific, week by week

Sea surface temperature against the same week of 1991–2020 — press play to watch the warm tongue spread

2026-10-11

cooler than normal −3 °C
−1 °C
normal
+1 °C
warmer than normal +3 °C

Weekly mean sea surface temperature anomaly from NOAA OISST v2.1, against the PSL 1991–2020 day-of-year climatology — the same baseline as the Niño indices on this page, and not the ONI's ERSSTv5 base period. The dashed boxes are Niño 4, 3.4, 3 and 1+2 from west to east, so the map and those numbers refer to visibly the same water.

Subsurface

Warm water volume: +27.62×10¹⁴ m³
WWV anomaly: +1.55×10¹⁴ m³
Mean 20 °C depth: +144.37m
Zonal tilt (west − east): +1.11×10¹⁴ m³

anomalies against the mean of the accumulated record

A short, moving baseline: it shifts as the record grows, so the same day can carry a different anomaly after a longer backfill. For quotable numbers use PMEL's, in wwv_pmel_monthly.

Indian Ocean Dipole

DMI: +0.74K
30-day mean: +0.74K
West box: +1.08K
East box: +0.34K

dmi_run is a 30-day running mean — a daily proxy for a monthly published index. For anything quoted use BoM's or NOAA's.

StormsWhere the atmosphere will let a cyclone form, what has formed, and whether we called it right

Genesis windows all closed

El Niño raises the bar for tropical cyclones without forbidding them. What matters day to day is whether a window opens — a stretch where the hostile conditions relax locally. A box is open when its ventilation index stays below that box's own calibrated threshold for long enough to count. Inside a window genesis is about twice as likely as outside one: it means genesis is not forbidden, not that it is likely.

Box	VI	7-day	Shear	SST	State
CaribbeanATL_CARIB	0.098	-0.076	16.4	29.5 °C	closed falling
Gulf of MexicoATL_GOM	0.232	+0.122	25.3	27.9 °C	closed rising
Main Development RegionATL_MDR	0.156	-0.007	22.8	28.9 °C	closed falling
Eastern North PacificENP	0.141	+0.062	16.8	28.9 °C	closed rising
Western North PacificWNP	0.119	+0.015	13.8	29.1 °C	closed risinglow skill

VI, the ventilation index, folds three things into one number: vertical wind shear, how dry the middle of the atmosphere is, and how much a storm could theoretically intensify. Low is favourable. The three can fail independently, which is why shear alone answers only a third of the question. 7-day is the change over the past week — a box above its threshold but falling is the one to watch.

A box marked low skill has a calibrated threshold, but its open window barely beats the season around it: genesis is only 1.25x more likely inside the window than outside it, against better than twice for the main development region. It is shown for completeness and should not be read as equivalent to the boxes above it.

Atlantic genesis windows (ERA5 2026-10-27, 7-day means, about 6 days behind today)

shear anomaly baseline: 1991-2020 (20+ years/day)

Every cyclone on earth right now

All basins, with the whole track — drag the slider to watch them develop

2026-11-03 00:00

Tracks current to 2026-11-03.

category 4–5 (113 kt and above)
category 2–3 (83–112 kt)
category 1 (64–82 kt)
tropical storm (34–63 kt)
depression (below 34 kt)

From the IBTrACS active file, which covers every basin rather than the three the National Hurricane Center warns for — this is where the western Pacific typhoons and the north Indian Ocean come from. It also keeps a storm after NHC stops issuing on it: Polo was gone from the live advisory list while IBTrACS still held all 93 of its fixes, including the 155 kt peak. Each dot is a three-hourly fix coloured by the wind measured then, so dragging the slider shows when a storm actually intensified, which a finished track cannot tell you. Positions lag real time by about a day.

Has the genesis detector been right?

Its own windows, scored against the storms that actually formed

Out of sample, from the end of the calibration: 2026-08-31 → 2026-10-06 (37 d)

Box	Storms caught		Windows with no storm		Missed
Box	observed	fitted 1991–2020	observed	fitted 1991–2020	Missed
ATL CARIB	—	0.57	—	0.67	—
ATL GOM	—	0.56	1 of 1	0.53	—
ATL MDR	0 of 1	0.64	—	0.38	Gonzalo
ENP	3 of 3	0.47	0 of 1	0.09	—
WNP	—	0.37	1 of 1	0.35	—

The thresholds were fitted on 1991–2020, so the fitted columns are an in-sample upper bound and these are the only out-of-sample numbers there are. Counts are shown rather than rates until there are enough events to carry one: three of three is not a detection rate of 1.00, and the fitted column rests on thirty seasons. Scored exactly as the calibration scores — a storm counts as caught if it forms inside an open window or within two days of one closing. The record ends where ERA5 does, about six days back, so a storm that formed this week is not scored yet.

Where it landsEl Niño is felt as rain that does or does not fall — on the Horn, the monsoons, the Amazon

Watching the Horn

Meteosat over the Jubba and Shabelle, every 15 minutes

The Horn is in darkness in this frame, and Convection needs sunlight for two of its three channels — over this ground the yellow cores are gone, not the storms. Anything still lit further west keeps them. Switch to Infrared 10.8 µm, which works at any hour and anywhere.

2026-11-05 01:00:00 UTC · Meteosat at 0° longitude, from EUMETSAT. GOES cannot see this ground — it sits at 75.2°W, about 120° away, past the limb — and NASA GIBS does not carry Meteosat, so this comes from EUMETSAT directly. Step back through the hours to watch a system build: the instrument repeats every 15 minutes. In daylight, Convection marks vigorous storms with bright yellow cores — yellow means tiny ice crystals at the cloud top, which only a strong updraught makes, so it is reading the violence of a cell rather than simply its height. Two of its three channels are reflected sunlight, so after dark the yellow disappears and the same storms read red and magenta; use infrared at night, which is cloud-top temperature and works at any hour — the coldest and brightest tops are the deepest. Thin cirrus and mountain wave clouds carry small ice crystals too and can mimic the yellow, which over the Ethiopian highlands is worth remembering. Riverine flooding here is made upstream in the Ethiopian highlands, days before it arrives at Belet Weyne — so watch the headwaters view, not only the corridor.

What the window is looking at

GOES-East, ten-minute imagery over the genesis boxes

Active right now: Vance EP 100 kt. Their past tracks, forecast tracks and forecast cones are drawn on the map, from the National Hurricane Center. A storm inside a dashed box while that box reads OPEN is this page's own genesis calibration being tested in public rather than in a backtest.

The dashed box is the genesis box itself. Switch to band 13 to see what the window is actually about: it measures cloud-top temperature, so the coldest, brightest tops are the deepest convection, and an open window means the environment is not preventing it. GeoColor is the recognisable view and blends visible by day with infrared at night, so the night side is not a black hole. Frames are the newest GIBS holds, usually within twenty minutes. The west Pacific box is drawn by Himawari instead: GOES-East sits at 75.2°W and the west Pacific is past its limb. Himawari has no GeoColor product, so its wide view is the red visible band and shows only the daylit side — band 13 is the one that works on both.

When the rains start

Onset date each year since 1979, and whether El Niño moves it

Sahel, 10 N-18 N, 10 W-10 E

Onset 2026: 2026-07-15 8 days early
Normal (1991–2020): 23 Jul ± 14 d

ENSO phase	years	median onset
El Niño	8	−2.7 d vs normal
neutral	33	−1.7 d vs normal
La Niña	7	−1.7 d vs normal

No ENSO signal in the onset date here. no detectable ENSO signal in onset date over this record: El Nino, La Nina and neutral summers all sit within a day or two of the same median. The Sahel's ENSO link is mainly in seasonal total, and reporting an onset signal here would be reading noise.

peninsular India, 8 N-20 N, 72 E-87 E

Onset 2026: 2026-06-21 18 days late
Normal (1991–2020): 03 Jun ± 13 d

ENSO phase	years	median onset
El Niño	8	+7.2 d vs normal
neutral	33	+0.7 d vs normal
La Niña	7	−4.3 d vs normal

El Nino summers are associated with a weaker and later Indian summer monsoon, one of the oldest and best attested ENSO teleconnections, and it is clearly present here: El Nino summers run about 11 days later than La Nina ones.

Onset date each year since 1979, and whether El Niño moves it — An agronomic onset, not an official declaration: the first day with 20 mm over three days that is not followed by a seven-day dry spell within the next month — the second clause is what separates the start of a season from a false start that kills a planting. India's Meteorological Department declares the monsoon over Kerala by its own rule and is the authority on that date; this is rainfall only, from CPC gauge analysis, so that anyone can recompute it. Its 1979–2026 normal for peninsular India falls on 3 June against IMD's climatological 1 June. Phase is the ONI of the same June–August, which separates a developing El Niño from a decaying one — they move the date in opposite directions — and is a hindsight label, not a forecast.

Flood and drought extent

Observed inundation and agricultural stress over the two impact panels

Flood classesrecurring floodflood (unusual for this place)

Vegetation healthstressed → healthy

2026-11-04. Flood is NASA GIBS's two-day VIIRS composite at 250 m — observed water, so a cloudy day shows less flooding rather than none. Drought is FAO's own rendering of the vegetation health index for the current dekad, which shows where the stress is; the numbers in the table above are the masked area means for the same dekad and are a different quantity. Only the two flood classes are drawn. The source layer also paints every ocean pixel as surface water and every cloudy one grey, which buried the few hundred flood pixels under a continent-sized sheet — and ocean is not a flood. The cost of removing them is that blank does not distinguish dry ground from ground the satellite could not see, and an optical sensor sees cloud exactly when flooding arrives. Drought is FAO's own rendering of the vegetation health index for the current dekad, which shows where the stress is; the numbers in the table above are masked area means for the same dekad and are a different quantity. Toggle either layer at the top right. Rainfall is the last 90 days as a percentage of the 1991-2020 average for the same days of the year, through 2026-11-02.

Amazon river stage

Satellite altimetry at six virtual stations, against each station's own seasonal cycle

El Niño dries the Amazon, and a river is the basin's memory of that. Rain falling anywhere upstream arrives here days to weeks later, already summed over hundreds of thousands of square kilometres, so the level of the river says what the whole catchment received — not what fell on the gauge. That is why a falling river is evidence of drought that has already happened, somewhere you were not looking.

A diagram of how the rivers connect, not a map — not to scale, and the bends are invented. The main stem runs west to east the way the water does, and thickens downstream because every tributary adds to it: Óbidos carries what the entire basin above it delivered, which is why it is the gauge to watch. The six sites are not towns we chose but places where a satellite's ground track crosses the river, so the altimeter can measure the water surface; of the candidates at each one we take the station still reporting.

Each dot is one gauge, coloured by how far today sits from that station's own seasonal cycle, measured in standard deviations of its own scatter rather than in metres.

above its seasonal cycle
close to it
below its seasonal cycle

Station	How far from normal	Is that a lot?	Has it been lower?	Last measured
Amazon basin
AmazonOBIDOS	-1.14 m	-1.4	10%unusual	2026-09-24 41d
PurusLABREA	-0.83 m	-0.5	33%	2026-10-23 12d
SolimoesTABATINGA	-0.72 m	-0.4	32%	2026-10-25 10d
TapajosSANTAREM	-0.43 m	-0.9	10%unusual	2026-09-18 47d
MadeiraPVELHO	-0.20 m	-0.1	41%	2026-10-16 19d
NegroMANAUS	+0.89 m	+0.4	76%	2026-10-21 14d
Jubba and Shabelle, Somalia
ShabelleBULOBURTE	-2.21 m	-1.2	6%unusual	2026-09-13 52d
ShabelleBELETWEYNE	-1.88 m	-1.2	7%unusual	2026-10-04 31d
ShabelleJOWHAR	-1.84 m	-1.4	4%unusual	2026-09-13 52d
JubbaBAARDHEERE	-1.29 m	-1.3	7%unusual	2026-10-08 27d
JubbaDOOLOW	-0.92 m	-1.1	5%unusual	2026-09-21 44d

Amazon basin

Colour is the departure in standard deviations, not in metres: blue above the seasonal cycle, brown below, grey close to it. The line joins the four main-stem gauges west to east and is a schematic connector between them, not the river channel — the Amazon meanders far outside it. Madeira, Purus and Tapajós are tributaries and are shown as points only.

Jubba and Shabelle, Somalia

Colour is the departure in standard deviations, not in metres: blue above the seasonal cycle, brown below, grey close to it.

“sd” is short for standard deviation — one station's ordinary amount of wobble around its own seasonal cycle. It is measured per station because they are not comparable in metres: the Rio Negro rises and falls 12.8 m across a year and the Tapajós 2.9 m, so the same number of metres means quite different things at each. Roughly two readings in three fall within ±1 sd, so anything inside that band is unremarkable by construction; ±2 sd is reached about one time in twenty.

Colour alone cannot tell you whether a reading matters, so the table gives how often each station has been lower in its own record. A departure inside about one standard deviation is ordinary — these rivers sit there most of the time. Óbidos is currently the lowest of the six, and a quarter of its own record is below it; that is a low river, not a drought. An event is a value well outside the ordinary range AND sustained across successive passes, which are two to four weeks apart.

Read the right-hand column, not the metres. A station's departure only means something against its own scatter: +1.06 m on the Rio Negro, whose seasonal range is 12.8 m, is half a standard deviation and an ordinary week. Levels are elevations above the EGM2008 geoid, so each station is comparable with its own history and with no other station.