Experimental, site-calculated

Westerly Wind Burst Monitor

A regional detector run on satellite-based equatorial Pacific winds, with the thresholds, code and sensitivity tests shown.

Read this first. This is an experimental calculation made by this site, not an official NOAA product and not a forecast. It is separate from the coupling checklist and its score on the dashboard, and nothing here feeds that score. For NOAA’s own view, see the CPC 850-hPa wind-anomaly chart.
Latest observation: 13 Nov 2026Calculated: 16 Nov 2026, 01:55 UTC
Latest observation (13 Nov 2026)
No active burst
Burst index −0.2 m/s against a 5 m/s threshold.
Last burst
10–13 Oct
4 days, peak 5.5 m/s, near 159°W.
Bursts found in 2026
7
Four appear with every background setting tested; three do not.

Summary

Experimental, site-calculated. This is not an official NOAA product and not a forecast, and it is separate from the dashboard coupling checklist and its score. Latest observation: 13 Nov 2026 (calculated 16 Nov 2026, 01:55 UTC). Status on that day: No active burst, burst index −0.2 m/s against a 5 m/s threshold. The most recent completed burst ran 10 to 13 Oct 2026 (4 days, peak 5.5 m/s near 159°W).

Run on CCMP satellite-based winds, the detector finds 23 westerly wind bursts between 1 Jan 2017 and 13 Nov 2026 (by year, 2018: 2, 2019: 7, 2021: 3, 2023: 3, 2024: 1, 2026: 7). 7 of them are in 2026:

Method in brief. CCMP v2.1 near-real-time 6-hourly 10 m zonal winds (Remote Sensing Systems, served by NOAA PIFSC OceanWatch), 5°S–5°N and 130°E–150°W at 2-degree spacing. Daily means are compared with a smoothed 2016–2025 climatology; a 10-degree sliding box has its own previous-61-day mean removed; a box is active at 5 m/s or more. A burst is a connected track of at least 3 consecutive days in which each day’s active region lies within 4° of longitude of the track’s region the day before. Version 0.3.0.

Validation limits. The 5 m/s intensity, 10-degree extent and more-than-2-day duration are taken from literature summaries, not re-derived from the original papers. The 61-day background removal, the baseline, the continuity rule and the box size are this site’s own choices, and they matter: across the 5 background windows tested the all-years total ranges from 21 to 25 events, and thresholds of 4, 5, 6 m/s give 69, 23, 8 events. The results have not been checked against TAO buoy winds or a published burst catalogue, and the real-data download has not been independently reproduced. CCMP is a blended analysis with a model background, and near-real-time values can be revised.

Burst index, last six months

Experimental, site-calculated. Strongest 10°-longitude box (5°S–5°N) of daily zonal-wind anomaly, after subtracting that box’s own previous-61-day mean. A burst is a connected track: at least 3 consecutive days with an active box at or above 5 m/s, each day’s active region lying within 4° of longitude of the same track’s region on the previous day.

Day inside a burstAt or above 5 m/s but not (yet) part of a connected 3-day burstBelow 5 m/s

Where the wind anomaly sits

Zonal wind anomaly averaged over 5°S–5°N by longitude and day (Experimental, site-calculated). Red is westerly anomaly, meaning weaker trade winds or actual westerlies; blue is stronger trades. This map keeps the slow background, so in a strong El Niño much of the basin looks red even without a burst.

2026 bursts and how firm they are

“Backgrounds” counts how many of five background windows (30, 45, 61, 75 and 90 days) also find the burst, matching events by overlapping dates and overlapping longitudes, not by date alone. Bursts found by only some windows are the ones to treat as tentative.

StartEndDaysPeak (m/s)Peak longitudeActive boxes centredBackgrounds
Show 2017 to 2025 bursts (same rule)
StartEndDaysPeak (m/s)Peak longitude

How sensitive are the counts?

Burst counts from 1 Jan 2017 to 13 Nov 2026, changing one setting at a time (all else at the default: 61-day background, 5 m/s, 4° continuity). “0 deg” means the connected patch must overlap the previous day’s patch in the same boxes.

Setting changedAll years202620192025

Local proxy versus regional detector

This is why the site does not judge bursts from one station. A single point misses bursts that happen elsewhere along the equator and can flag local disturbances. The comparison uses one 2-degree CCMP grid cell near Tarawa, not station data, so treat it as an illustration.

Method, data and limits

Data. CCMP v2.1 near-real-time 6-hourly ocean-surface winds from Remote Sensing Systems (Mears et al. 2019, Journal of Geophysical Research: Oceans), which blend satellite radiometer and scatterometer winds with a model background from NCEP GFS. Served by NOAA PIFSC OceanWatch ERDDAP, dataset ccmp-daily-v2-1-NRT, available for public use with citation. We use the zonal component at 2-degree spacing, 5°S–5°N, 130°E–150°W.

Steps. Daily means; anomaly against a smoothed 2016–2025 daily climatology; a sliding box 10° of longitude by about 10° of latitude; each box minus its own previous-61-day mean; a box is active at 5 m/s or more; each day’s active boxes are grouped into patches. A track follows one patch per day and continues only when that patch lies within 4° of longitude of the track’s patch the day before; a patch joins at most one track, so two separate patches cannot be stitched into one event. A burst is a track lasting at least 3 consecutive calendar days. Its peak and longitude range come from that track’s own patches. The most recent day is labelled Candidate burst when a track ending on it has lasted only 1 or 2 days, Burst in progress only when that track has reached 3 days, and Unavailable when the day has no valid value.

What is published and what is ours. The 5 m/s intensity, 10° zonal extent and more-than-2-days duration are the criteria summarised in the literature (Harrison and Vecchi 1997; Vecchi and Harrison 2000; Seiki and Takayabu 2007; Puy et al. 2016). We have not re-derived them from those papers. The 61-day background removal, the 2016–2025 baseline, the continuity rule and the 10° by 10° box are this site’s own choices, and the sensitivity table above shows what they do.

Reproduce it. Version 0.3.0 of the script wwb_detect.js (an ES module; save it as wwb_detect.mjs) downloads the data and writes the file this page reads, wwb-experimental.json. With Node 18 or later: node wwb_detect.mjs --out out --cache cache. Each quarter is one query of this form (brackets shown unencoded):

https://oceanwatch.pifsc.noaa.gov/erddap/griddap/ccmp-daily-v2-1-NRT.csv?
uwnd[(2026-07-01T00:00:00Z):1:(last)][(-4.875):8:(5.125)][(130.125):8:(210.125)]