Daily, from the ECMWF AIFS ensemble and the ECCC GDPS, anomalies against ERA5 1991–2020
Global circulation and jets
The atmosphere's response to the tropical ocean and what it does downstream: the upper-level flow, angular momentum and its torques, the overturning cells, the jets and the stratosphere — one figure at a time, picked from the menu. Every figure carries its method and sources under the caption. Most diagnostics use the full 50-member AIFS-ENS or its control; the heavier ones (angular momentum, jets, E–P flux, wave activity flux) use a 25-member subset, and each About block says which.
← → step through the options · ↑ ↓ change figure · click a still to enlarge
The full-field Pacific surface forecast behind the equatorial diagnostics: mean-sea-level pressure (contours, with H/L centres) and 10 m wind (speed shaded in knots, plus barbs), animated over the tropical and subtropical Pacific. An equatorial westerly push between the Maritime Continent and the dateline is a wind-burst signature. Switch between the ECMWF super-ensemble mean (AIFS-ENS + IFS-ENS, days 1–15) and ECCC GDPS (15 km deterministic, days 1–10) as an independent physics cross-check — same domain and styling, so differences are the models, not the plotting.
ECMWF AIFS-ENS + IFS-ENS (open data, CC BY 4.0), member-weighted super-ensemble mean, each 00/12Z cycle · ECCC GDPS 15 km deterministic (MSC Datamart), 00Z cycle, daily
GDPS forecast top-of-atmosphere OLR converted to IR brightness temperature and rendered like the live GMGSI tropical Pacific loop on the El Niño monitor page — a simulated satellite loop, 3-hourly frames to day 10. White and colour are cold cloud tops (deep convection).
ECCC GDPS top-of-atmosphere OLR (MSC Datamart) · OLR → Tb via inverse Ohring–Gruber · 3-hourly frames, days 0–10 · GMGSI enhanced-IR colortable · daily
GDPS 150 hPa wind as isotach fill and contours with wind barbs and geopotential height (6 dam interval), 3-hourly frames to day 10 — the Hadley/Walker outflow layer and the subtropical jets. White is below 25 kt.
ECCC GDPS 150 hPa u/v and geopotential height (MSC Datamart) · isotach fill, barbs, 6 dam height contours · 00Z cycle, days 1–10 · daily
The large-scale divergent circulation aloft from the AIFS-ENS ensemble mean: the 200 hPa velocity-potential anomaly shaded, with the irrotational (divergent) wind as vectors. Green is upper-level divergence (the outflow above deep convection); orange is convergence (subsidence). The vectors point out of the divergence centres, tracing the rising and sinking branches of the Walker and Hadley circulations; in El Niño the divergence shifts east over the central Pacific. Anomaly is versus the ERA5 1991–2020 climatology. The first frame is the analysis, the rest is the forecast to day 14.
ECMWF AIFS-ENS 200 hPa u/v, 50-member mean (open data, CC BY 4.0) · velocity potential via spherical-harmonic inversion · anomaly vs ERA5 1991–2020 · updated each 00/12Z cycle
The Takaya–Nakamura (2001) wave activity flux is a phase-independent vector diagnostic of quasi-stationary Rossby wave propagation: unlike watching ridges and troughs move, the flux W points along the wave packet's group velocity — where the wave energy itself is travelling, ducted along the jet-stream waveguides — and its convergence marks where the downstream flow will amplify days later. That makes it a practical forecast tool for downstream development, blocking onset, and teleconnection excitation: watch packets radiate out of the tropical Pacific convection (the El Niño–PNA pathway) or off a mountain-torque event, arc along the subtropical jet, and pile into an amplifying ridge.
$$\mathbf{W}=\frac{p\cos\phi}{2\,|\mathbf{U}|}\begin{pmatrix}U(\psi_x'^2-\psi'\psi_{xx}')+V(\psi_x'\psi_y'-\psi'\psi_{xy}')\\[2pt]U(\psi_x'\psi_y'-\psi'\psi_{xy}')+V(\psi_y'^2-\psi'\psi_{yy}')\end{pmatrix}$$
Computed at 250 hPa — the level of the mid-latitude jet core and the stationary-wave maximum — for each of 25 AIFS-ENS members at every daily lead to day 15, then averaged: the flux is quadratic in $\psi'$, so the flux of the ensemble mean fades with lead as members decorrelate, whereas the mean of the members' fluxes keeps the packets that are actually forecast. Each member's streamfunction anomaly $\psi'$ (spherical-harmonic inversion of the vorticity, against the ERA5 1991–2020 day-of-year climatology) is smoothed with a 5-day running mean along the lead before the flux is formed, because the theory is for quasi-stationary waves and fast synoptic packets enter the phase-independent form with error; the basic state $(U,V)$ is the ERA5 day-of-year climatology plus the 30-day mean anomaly of the AIFS 0-h analyses, and $\psi'$ is taken against that same low-passed flow — so in a year with a displaced jet the packets are steered by the waveguide that is actually there, and perturbation and basic state are consistent. Shading is the flux convergence $-\nabla\!\cdot\!\mathbf{W}$ — red marks where wave activity is piling up and the downstream flow amplifies over the following days (blue marks emission regions) — hatched where fewer than 60% of members agree on its sign; grey contours are $\psi'$ (the anomalous ridges/troughs, dashed negative), arrows the member-mean wave activity flux, green contours the ensemble-mean 250 hPa jet at that lead (25/35/45 m/s) — the actual waveguide the packets follow. Fluxes are masked equatorward of 20° and outside westerlies, where the quasi-stationary linear theory does not apply; the diagnostic is most meaningful for 3–10-day downstream development, not fast transients. Both hemispheres are shown — the Southern Hemisphere winter waveguide is the active one in austral winter, a view few live sites carry.
ECMWF AIFS-ENS 250 hPa u/v, 25 perturbed members (open data, CC BY 4.0) · Takaya & Nakamura (2001, JAS) eq. 38, member-mean flux · basic state & ψ climatology: ERA5 1991–2020 · updated each 00/12Z cycle
The upper-level flow as PV thinkers read it. Ertel potential vorticity is computed on the AIFS-ENS control member’s temperature and wind on nine levels from 700 to 100 hPa — the control alone, because PV does not survive ensemble averaging: a mean of fifty folds is a smear. The dynamic tropopause is the 2-PVU surface, found by searching upward from 700 hPa; its potential temperature is shaded (low θ, blue, is stratospheric or polar air reaching down; high θ, red, is a tropical tropopause), its pressure contoured, and the wind on it drawn. Troughs, cut-off lows and ridges are the θ pattern itself; a tight gradient is the jet; a tongue of low θ wrapping up is a Rossby wave breaking. The other two loops give PV on the 330 and 350 K isentropes with the 2-PVU line in dark red — 330 K sits near the winter jet, 350 K near the summer and subtropical one. 12-hourly to day 10, North Pacific to the Atlantic.
ECMWF AIFS-ENS control t/u/v on 700–100 hPa (open data, CC BY 4.0), 0.25° · Ertel PV on pressure levels, 2-PVU surface by linear interpolation in PV, isentropic PV by interpolation in θ · updated each 00/12Z cycle
The upper-level flow as PV thinkers read it. Ertel potential vorticity is computed on the AIFS-ENS control member’s temperature and wind on nine levels from 700 to 100 hPa — the control alone, because PV does not survive ensemble averaging: a mean of fifty folds is a smear. The dynamic tropopause is the 2-PVU surface, found by searching upward from 700 hPa; its potential temperature is shaded (low θ, blue, is stratospheric or polar air reaching down; high θ, red, is a tropical tropopause), its pressure contoured, and the wind on it drawn. Troughs, cut-off lows and ridges are the θ pattern itself; a tight gradient is the jet; a tongue of low θ wrapping up is a Rossby wave breaking. The other two loops give PV on the 330 and 350 K isentropes with the 2-PVU line in dark red — 330 K sits near the winter jet, 350 K near the summer and subtropical one. 12-hourly to day 10, North Pacific to the Atlantic.
ECMWF AIFS-ENS control t/u/v on 700–100 hPa (open data, CC BY 4.0), 0.25° · Ertel PV on pressure levels, 2-PVU surface by linear interpolation in PV, isentropic PV by interpolation in θ · updated each 00/12Z cycle
ECMWF AIFS-ENS (open data, CC BY 4.0) · anomaly vs ERA5 1991–2020 (Copernicus C3S) · archived each cycle
Top: relative AAM integrated in 1.5° latitude bands — ~3 months observed (ERA5) plus the 15-day AIFS-ENS mean forecast — so poleward-propagating westerly anomalies and building subtropical momentum show as slanted warm streaks. Bottom: each hemisphere's trajectory through (AAM anomaly, tendency) phase space, the hemispheric analog of the Weickmann–Berry global wind oscillation orbit: grey trail = last 75 days, red = the forecast. The headline sentence is generated automatically from the forecast quadrant and the latitude band driving the change — e.g. "NH: rising above-normal AAM — subtropical westerlies increasing."
ERA5 (local store) · AIFS-ENS ensemble mean (open data, CC BY 4.0) · clim: ERA5 1991–2020 · updated daily
The torques above change the global AAM; this shows where that angular momentum sits and how the forecast moves it, as a latitude–pressure cross-section of the relative-AAM density whose integral over latitude and pressure through the 50–1000 hPa layer is essentially the global AAM. Top panel, absolute: where the AAM resides, dominated by the subtropical jets (the winter hemisphere’s is strongest). Bottom, change versus the 0-h analysis: where the forecast is adding (red) or removing (blue) angular momentum, with the per-latitude breakdown of the global ΔAAM along the top strip. Black contours are the zonal-mean zonal wind (bold = zero). The bottom panel integrates to the ΔAAM in the title, tying it back to the torque budget above. AIFS-ENS ensemble mean over the control and 25 perturbed members; the slider steps the forecast Day 0–15.
ECMWF AIFS-ENS zonal wind (13 levels, 50–1000 hPa) + surface pressure (open data, CC BY 4.0) · ensemble mean · updated daily
Atmospheric angular momentum changes only through torques the Earth exerts on the atmosphere. Three torques act: friction, the surface wind stress; mountain torque, pressure pushing on resolved topography; and the gravity-wave drag the model exerts on sub-gridscale terrain. The first two are resolved live — the open-data forecast stream carries no gravity-wave stress, so that term is left in the residual. The budget applies to total AAM — the wind (relative) part plus the mass term the rotating Earth carries in its surface-pressure field. Only the wind term is computed here; it tracks the total closely on subseasonal timescales but is not identical to it.
$$\frac{dM}{dt}=T_{\text{fric}}+T_{\text{mtn}}\;(+\,T_{\text{gw}}),\qquad T_{\text{mtn}}=a^2\!\iint h_s\,\frac{\partial p_s}{\partial\lambda}\,\cos\phi\;d\lambda\,d\phi$$
Everything here is an anomaly against the ERA5 1991–2020 climatology at the same time of year, and that is deliberate. The absolute budget cannot be made to close: ERA5’s own three terms sum to −4.5 ± 0.7 Hadley in the annual mean against a required zero, the resolved mountain torque is not resolution-converged (+3.1 Hadley at 0.25°, −7.7 at 0.5°, −23.7 at 1°), and the semidiurnal pressure tide cannot be sampled away at twice a day. Every one of those errors sits in the mean and cancels in an anomaly.
Maps: friction (top) and mountain (bottom) torque-density anomaly, day 0–15 — red adds westerly momentum, blue removes it. Grey contours are the standardized surface-pressure anomaly the mountain torque acts on, so colour and contour read together: high to the west of a range and low to its east is a braking event. Series: the same anomalies integrated over the globe and each hemisphere; the gap between net surface torque and $dM/dt$ is the unresolved gravity-wave-drag torque (the sub-gridscale companion of the mountain term, which ERA5 carries but the open-data forecast fields do not) plus, hemisphere by hemisphere, cross-equatorial transport — an internal flux that cancels in the global integral. By range: which barriers are doing the work.
How much is this worth? Tested against 30 years of ERA5: over a two-week-to-two-month window the net surface torque tracks a substantial fraction of the actual change in atmospheric angular momentum, and the mountain term carries most of that information. Two limits are worth knowing. The regression slope comes out well below 1, which is what errors-in-variables predicts when a 30 Hadley-scale quantity is sampled only every five days — the torques are correctly sized and merely sampled too coarsely to integrate. And the skill vanishes beyond a season — AAM is mean-reverting, so there is no annual-scale change left to explain. Read these as a driver of week-to-month swings, never as something to accumulate.
Friction uses a drag coefficient solved per grid cell and calendar month against ERA5’s own boundary-layer stress, rather than one global constant. A drag coefficient is a property of the surface, not of a latitude: the fitted field runs 1.3×10−3 over ocean to 1.8×10−2 over rough land and reproduces ERA5’s stress with a median $r^2$ of 0.95, where the old constant carried 72% of the true variability and had the wrong sign in the global mean. The climatology these anomalies are taken against is rebuilt with the identical field, so forecast and reference are the same quantity.
ECMWF AIFS-ENS surface pressure & 10 m wind (open data, CC BY 4.0) · drag coefficient and torque climatology from ERA5 1991–2020 (Copernicus C3S) · instantaneous once-daily fields · updated daily
The zonal-mean meridional mass streamfunction $\Psi(\phi,p)$ traces the overturning cells (Hadley, Ferrel, polar) from the AIFS-ENS 0-h analysis:
$$\Psi(\phi,p)=\frac{2\pi a\cos\phi}{g}\int_0^p [v]\,dp',\qquad [v]=\text{zonal-mean }v$$
Shown as the anomaly $\Psi'$ from the ERA5 1991–2020 harmonic climatology (mean + annual + semiannual, by day-of-year), how the overturning departs from its normal seasonal state. The black contours are the absolute $\Psi$ (the actual cells) and the arrows show the vertical motion (up is ascent, scaled by strength); the strong equatorial signal tracks the El Niño Walker/Hadley response. Each frame is a 7-day running mean (the 0-h analyses are stashed into a rolling history and trailing-week-averaged), which smooths the day-to-day wobble so the evolution of the Hadley-cell strength is legible. Fixed colour scale.
ECMWF AIFS-ENS 0-h analysis meridional wind (open data, CC BY 4.0) · anomaly vs ERA5 1991–2020 Ψ climatology (Copernicus C3S) · 7-day running mean · updated each cycle
The equatorial companion to the Hadley-cell plot above: the zonal overturning streamfunction $\Psi_W(\lambda,p)$ built from the 5°S–5°N divergent zonal wind (velocity-potential decomposition, so the rotational flow that dwarfs it near the equator is removed):
$$\Psi_W(\lambda,p)=\frac{\Delta y}{g}\int_0^p u_D\,dp',\qquad u_D=\text{5°S–5°N divergent }u$$
Colour is the anomaly from the ERA5 1991–2020 harmonic climatology; black contours are the absolute cells (ascent where $\Psi_W$ increases eastward — arrows show the vertical branches). The headline Pacific Walker cell index is the mean $\Psi_W$ over 140°E–160°W, 300–700 hPa: on ERA5 Decembers it runs about +0.3…+0.4×10¹⁰ kg/s in La Niña and reverses sign (−0.1) in the 1982/1997/2015 super El Niños — a direct, mass-flux measure of how far the Walker circulation has weakened, shifted or reversed. Each frame is a 7-day trailing mean of the 0-h analyses.
By ocean basin. The figure below asks what the Indian and Atlantic Oceans are doing to this circulation once ENSO is taken out, two ways. First a regression on ERA5 and ERSST 1991–2020: the two-level Walker proxy (divergent zonal wind at 200 minus 850 hPa along the equator) is regressed on Niño-3.4 and its three-month lag first, and the Indian Ocean Dipole, Indian Ocean basin, Atlantic Niño and tropical North Atlantic indices enter only as the residuals left after that, so their parts are ENSO-independent by construction. Second an idealised simulation: the Gill (1980) linear equatorial response to the observed SST anomaly of each basin, with the ENSO regression pattern subtracted from the Indian and Atlantic fields before forcing and a single amplitude calibrated on ENSO. All anomalies are detrended, so a warming ocean is not read as a forcing. Where the two legs agree in sign and size the attribution is robust; where they part, the statistical link is not a simple SST-forced one.
ECMWF AIFS-ENS 0-h analysis u/v, 13 levels (open data, CC BY 4.0) · velocity potential via spherical harmonics · anomaly vs ERA5 1991–2020 ΨW climatology · 7-day running mean · updated each cycle · basins: OISST v2.1 and ERSSTv5 (NOAA), regression and Gill model as described, updated twice daily
A robust extratropical El Niño signature is a stronger, equatorward-shifted subtropical jet in each winter hemisphere. Top: the zonal-mean zonal wind $[u](\phi,p)$ at the analysis — colour is the anomaly against the ERA5 1991–2020 day-of-year normal, contours the absolute jets, ▼ the 200 hPa subtropical cores. Below: the forecast evolution of each core's speed and latitude (max $[u]$ at 200 hPa within 15–45°), ensemble members and mean against the climatological normal ±1σ band, with the analysis departure quoted in σ — the "how anomalous are the jets right now" number. The animation steps the ensemble-mean cross-section through the 15-day forecast (▼ forecast cores, ▽ their climatological positions).
ECMWF AIFS-ENS 13-level zonal wind, control + 25 perturbed members (open data, CC BY 4.0) · clim: ERA5 1991–2020 harmonic day-of-year normal ±1σ (WeatherBench2 / Copernicus C3S) · updated daily
The 200 hPa jet over 10–70°N, 100°E–120°W, member by member: an extension index (the leading Nov–Mar EOF of the ERA5 1991–2020 anomalies, positive when the jet reaches east across the Pacific), a shift index (the second EOF, positive poleward), the plain exit-region wind anomaly over 30–40°N 170°E–150°W, and the jet terminus — the easternmost longitude the 30 m/s core reaches. The black tail is ERA5 through the same projection (about six days behind), the grey dots are the AIFS analyses, so the forecast starts from an observed state rather than a model one.
The link to the Himalayan mountain torque above is tested, not assumed: over 30 years of ERA5 the exit-region wind correlates with the torque at $r\approx0.25$ five to seven days later, and the composite extension after torque days of +1.5σ or more rises to about +0.2σ a week on — modest, but outside what random dates produce. The dashed red curve on the plumes is that composite anchored on this cycle’s torque peak: the historical expectation the members can be read against. The second figure follows the chain downstream to 500 hPa, where the composite response is a Gulf-of-Alaska ridge nine to twelve days after the torque (weaker over Alaska itself), set beside the model’s own height anomalies and two ridge indices.
ECMWF AIFS-ENS 200 hPa u and 500 hPa z, 51 members (open data, CC BY 4.0) · reference: ERA5 1991–2020 via WeatherBench2 (Copernicus C3S), EOFs after Jaffe et al. (2011) and Winters et al. (2019) · observed tail: ERA5 (ARCO) and AIFS analyses · updated daily
The stratospheric vortex is the slow variable that sets the odds for the surface weeks later, so it is worth watching before it moves. Top: the zonal-mean zonal wind at 10 hPa, 60°N — the standard sudden-stratospheric-warming diagnostic. A reversal to easterly here in winter is a major SSW (the original WMO definition also asked for a reversed 60–90° temperature gradient), which is why the zero line is drawn heavy and the member fraction crossing it is called out. Middle: the same wind at 100 hPa. Anomalies that reach this far down are the ones with a path to the troposphere — a 10 hPa event that never appears at 100 hPa usually never reaches the surface. Bottom: the 100 hPa polar-cap (65–90°N) height anomaly, positive for a weak or displaced vortex; this is the field that actually leads the AO/NAO response. Blue traces are AIFS-ENS members, orange their mean, dashed the control. The 100 hPa height is control-only because ECMWF open data publishes no perturbed geopotential at that level.
ECMWF AIFS-ENS zonal wind and geopotential at 10/100 hPa (open data, CC BY 4.0) · reference: MERRA-2 1980–2026 day-of-year 10th–90th percentiles (NASA GMAO) · analysis tail: ERA5 · updated daily
The vortex as it actually looks, rather than as an index. Wind speed is
shaded with streamlines over it at 10 hPa. This is where the vortex lives and where an SSW is declared — the level the WMO diagnostic is evaluated on. The red contour
is u = 0: where the westerlies end in the zonal sense (not the PV-gradient
vortex edge), and the outer bound of the corridor $0<\bar{u} ECMWF AIFS-ENS control (open data, CC BY 4.0) · 10 hPa speed shaded, streamlines overlaid, red contour u = 0 · fixed colour scale · analysis plus 12-hourly steps to day 15 · updated each 00/12Z cycle
Where the torque budget shows how the Earth exchanges angular momentum with the atmosphere at the surface, the Eliassen–Palm flux shows how the waves redistribute it internally — the same quasi-geostrophic diagnostic used to diagnose sudden stratospheric warmings, computed live from AIFS-ENS and stepped through the full 15-day forecast.
ECMWF AIFS-ENS ensemble — control + 25 perturbed members, per-member fluxes averaged (open data, CC BY 4.0), 14 levels, day 0–15 · QG E–P flux, wavenumbers 1–3, global-mean static stability · poleward of 82° masked · rendered in Julia (CairoMakie) · updated each 00/12Z cycle
The zonal wavenumber-1 component of geopotential height, 100 hPa beside 500 hPa for one hemisphere at a time — the Hemisphere selector switches between them, and the loop opens on whichever one currently carries the larger wave. The two levels share a frame because that is the comparison that carries the physics: 100 hPa is what the vortex feels, 500 hPa is the tropospheric source, and a ridge that leans westward with height is actively driving the vortex while one sitting over the same longitude at both levels is not. Each level keeps its own colourbar — wave amplitude grows with height, so a shared scale would flatten the lower one. Because only k = 1 is retained the field is an anomaly by construction: red is a ridge, blue a trough, the zonal mean already removed. Thin contours are the full height field; values inside the deadband are left white. The map is the ensemble mean — the part of the wave the forecast agrees on, which is the only part whose phase is worth reading.
ECMWF AIFS-ENS ensemble mean (open data, CC BY 4.0), 100 and 500 hPa, analysis to day 15 · zonal wavenumber-1 Fourier component of geopotential height · superposition index vs ERA5 1991–2020 (WeatherBench2) · polar stereographic · updated each 00/12Z cycle