wx mapwall

Sources & methods

Data

ERA5 accessed September–October 2026 under CC BY 4.0. Contains modified Copernicus Climate Change Service information (2026). Neither the European Commission nor ECMWF is responsible for any use that may be made of this information.

GEFS 10 hPa ensemble watch

Ensemble watches are started and ended by the website operator. While active, one daily 00Z initialization is acquired at 10 hPa. A paused watch retains its last results and their initialization dates, with no new ensemble downloads. There is no automatic reversal threshold or automatic restart.

Each forecast is one 00Z initialization, at six-hourly valid UTCs through 384 hours. Wind is eastward-positive in m/s at the exact 60°N/S latitude rows. Each member’s complete 720-point longitude circle is averaged first; the ensemble mean and type-7 10th–90th percentiles are then calculated from the 31 member means (c00 and p01–p30). The chart shows the 31 member lines with the c00 control in bold. Curves default to a centered 24-hour time average of each member and the separate GFS forecast, calculated by integrating linearly between the six-hour samples. The window shortens symmetrically to 12 hours near an endpoint or gap, then to an unchanged raw point at the boundary; it never crosses a gap or initialization. Raw 6h mode shows the original curves. This averaging applies only to forecast curves; the separately labelled daily 00Z climatological minimum–maximum range is unchanged and is not a probability band. The timeline remains six-hourly. The c00 and GFS readouts match the displayed curves and state their actual averaging window; the member table retains raw six-hour values. The selected-time mean and type-7 10th–90th percentiles remain explicitly labelled raw statistics, the percentiles as member spread, not a calibrated confidence interval. Every displayed time requires all 31 validated members; incomplete times remain gaps.

The bold c00 line is the GEFS control. The orange operational GFS line is a separate, same-initialization forecast, excluded from all GEFS statistics. Its validated scalar winds and source hashes are frozen with each retained initialization. Missing GFS times stay blank. Initialization changes preserve the selected valid UTC, or the nearest available time.

At 10 hPa, muted shading shows the NCEP/NCAR R1 daily minimum–maximum range for 1991–2020, labelled separately from the GEFS members. For each calendar date it spans the lowest and highest of the 30 daily 00Z longitude-mean winds at exactly 60°N or 60°S, taken from the accepted R1 reductions. Daily 00Z climatology is interpolated for display between dates; February 29 interpolates February 28 and March 1.

References & scales

Temperature anomalies subtract ERA5 2000–2025 monthly 00Z means, interpolated between month midpoints and onto each grid. No model-bias correction. This 26-year reference is not a 30-year climate normal.

Wind, polar-temperature and eddy-flux time series show 1991–2025 individual daily R1 histories when the extended history is available; each chart labels the actual historical years shown. Their climatological minima–maxima, 10th–90th percentiles and dashed median remain fixed to 1991–2020. The 2021–2025 traces do not enter those statistics and may fall outside the shaded range. Height-wave series currently use ERA5 2000–2025 for both traces and statistics; their extension is pending daily source coverage. Thin traces describe past variability, not forecast ensemble spread. February 29 is interpolated for display, gaps remain blank, and December connects only to the next available historical year.

Cross-section scales retain their fixed calibration from 9,497 R1 days in 2000–2025. Colorbar extensions mark values beyond the limits. QBO wind uses ±60 m/s and the NASA QBO palette.

Zonal wind anomaly

Shading is actual zonal-mean wind minus the ERA5 1991–2020 reference at each valid UTC, in m/s. Positive means more eastward in both hemispheres; contours show actual wind at 10 m/s intervals, with zero emphasized. This ERA5 reference is separate from the R1 reference used for the linked actual-wind time series.

The reference uses monthly means of daily 00Z analyses, with mean-preserving interpolation between month centres. It is not an observed daily climatology; other hours use the same 00Z-based reference. Leap days use a common-year solve (within 0.10 m/s). Forecast anomalies include model–ERA5 offsets and forecast drift; no bias correction is applied. The unequal color intervals emphasize small departures, with white within ±2 m/s and fixed limits of ±70 m/s.

Coverage is 1,000–1 hPa where both the actual field and every 1991–2020 daily 00Z baseline sample support above-ground values around the whole latitude circle, including reference interpolation endpoints. A mountain anywhere on that circle can mask its lower levels. Grey or hatching marks below-ground, unsupported or missing data; gaps are never filled.

Time–height panels use the exact 60°N/S row. Forecast context spans the preceding three calendar months of daily 00Z ERA5/ERA5T, then the model's own archived f000 analyses across the reanalysis latency gap, followed by its forecast. Source boundaries and gaps stay separate; reanalysis history is labelled pending when unavailable. Retrospectives use only their own daily 00Z ERA5 event dates. A history click moves the cursor; only an exact available forecast/event UTC moves the upper frame.

EP flux & signs

EP flux uses NOAA PSL methods and aostools, omitting vertical eddy momentum transport; it is unreliable near the equator. EP forcing is one contribution to wind acceleration.

Arrow direction uses latitude/log-pressure coordinates. Purple means downward flux, not proof of reflection. Nonlinear length compression preserves direction; the fixed EP-flux reference is a display anchor, not an absolute flux scale.

Westerly wind and northward eddy flux are positive. EP-acceleration contours: positive dashed, negative solid. Unsupported terrain/derivative regions stay blank.

Time series & time–height

Wind: 60°N/S. Temperature: area-weighted 60–90° caps. Both show 10, 50 and 100 hPa.

Eddy heat flux v′T′ (K m/s) and momentum flux u′v′ (m²/s²) average 45–75° with cosine weights, at 100, 30 and 10 hPa. Primes subtract the simultaneous zonal mean. Values are northward-positive covariances, not EP flux.

GFS history uses ERA5/ERA5T; GEOS uses GEOS-FP analyses. Available f000 analyses bridge recent dates. Boundaries and missing data are marked. Retrospectives use daily 00Z ERA5; each panel shows its own window.

Heat-flux time–height anomalies subtract the ERA5 2000–2025 31-day seasonal mean. Warm colors indicate enhanced poleward transport: positive in NH, negative in SH. Labels remain northward-positive. Forecast-page smoothing stays within source/gap boundaries (Gaussian σ 8 h, radius 24 h); retrospective daily values are unsmoothed.

Height waves 1–3

Fourier amplitudes on complete 60°N/S longitude rings, 100–1 hPa; ERA5 geopotential is divided by 9.80665. One-sided coefficients use 2/N normalization. Amplitudes are metres, not EP flux; A²/2 is longitudinal height variance. Ridge phase is east longitude modulo 360°/wave number. Time–height color limits: 2,600 / 1,000 / 450 m; time-series axes fit the displayed values. NASA method context.

Annular-mode proxy

Sign-reversed standardized 60–90° polar-cap height, using fixed ERA5 2000–2025 normalization; this is not EOF-based NAM/SAM. Panels show their shading range and unlabeled 2σ contours (negative dashed, positive solid). Context: Baldwin & Dunkerton (2001); Baldwin & Thompson (2009).

EOF NAM/SAM

Area-weighted zonal-height EOF1 over 20–90° at each supported pressure, using daily 00Z ERA5 2000–2025, a 31-day seasonal mean and fixed annual PC standard deviation. Positive index has negative polar-height loading; global-mean height is retained. Shading is ±4σ with thin, unlabeled 1σ contours. Recent forecast context uses GEOS assimilated analysis; source transitions and gaps remain separate. The index is not directly interchangeable with other providers' NAM/SAM definitions.

Regional views

Instantaneous v′T′ is poleward-positive in both hemispheres. Sections average 45–75°, from 1,000 to 1 hPa. Height-departure contours are every 200 m, negative dashed; polar-map contours show zonal wind. Terrain is excluded before departures; means require 80% longitude and weighted band coverage. Only complete archived runs are listed. Linked series require a retained zonal run.

Events

Major SSWs: Charlton & Polvani (2007), with NH dates checked against the NOAA compendium. Dates mark wind reversals or wave-spell ends; final warmings cover the search season. PWD/NWD adapt Reichler & Jucker (2022) to R1, with a provisional ±12.9-day threshold. Ranks compare catalogue indices, not impacts.

Literature-listed minor warmings use peak/reference dates; QBO disruption dates are approximate. Windows show context, not duration. QBO panels average 5°S–5°N with five-day smoothing. Antarctic references: Rao (2020) · Zi (2025) · Lim (2026).

Software

EP-vector scaling: Jucker (2021). Data handling: xarray, Hoyer & Hamman (2017).