Route 08 · For machines

The Wire

Everything this site publishes, in two files meant to be read by something other than a person: one plain-text dump for a broadcast desk or a grep, and one JSON feed for a model or an agent. Same format the rest of the Forecast Network carries.

08ROUTE

Wire

1 file

the whole outlook, plain text, pipe-delimited

Outlook rows

six regions x two fields x every lead

Endpoints

every route's own JSON, indexed

Encoding

ASCII

no typography anywhere in the feed or the wire

The two files

Both regenerate with every build

Plain text

/wire.txt

The whole outlook in one greppable file. Sections are marked ##; tables are pipe-delimited with a header row that names the unit of every column. Written for a broadcast producer, a shell pipeline or a model context window, which is why it is ASCII with no typography in it at all.

JSON

/assets/feed.json

The headline, the broadcast lines, the full regional outlook, the verification summary and an index of every other endpoint. Shaped like the feeds the rest of the network carries, so a reader that already handles one handles this.

Today’s lines

From the feed, verbatim

C3S multi-system puts S. America temperature at 74 percent likely above normal for November 2026 - the strongest tilt in this run.
Its own reforecast beat climatology over 96 percent of that box; the three systems differ by 11 points over the United States in February 2027.
C3S multi-system takes 9 of 10 whole-month scorings on reforecast RPSS over land
Skill decays with lead: over the United States, 54-61 percent of the box beat climatology at November 2026, 46-89 percent by March 2027

Derived, never written: the headline is the largest departure from the climatological third at the first whole month, from whichever system route 06 finds verifies best. It changes when the data does.

Endpoints

Every route’s own data

Path	route	Holds
/assets/feed.json	08	This feed: headline, broadcast lines, the regional outlook, and this index
/wire.txt	08	The whole outlook as plain text, section-marked and pipe-delimited
/assets/seas5/latest.json	01	Route 01 pointer: newest initialisation and every cycle published
/assets/probs/probs.json	10	Route 10: SEAS5 tercile probabilities, raw and trend-adjusted, with RPSS and ACC by lead
/assets/multi/multi.json	03	Route 03: ten C3S systems, the consensus, and each system's own reforecast score
/assets/nmme/nmme.json	02	Route 02: four NMME models, the consensus, and each model's own reforecast score
/assets/analog/analog.json	09	Route 09: this site's own analog forecast, its analog years and its cross-validated skill
/assets/compare/compare.json	11	Route 11: the matched verification of routes 09 and 10 over the same years
/assets/cities/cities.json	07	Route 07: 40 city briefings, each with the skill of its own gridcell attached
/assets/regimes/regimes.json	12	Route 12: the seven Euro-Atlantic weather regimes, SEAS5's monthly frequencies against its own climatology, with ACC and RPSS
/assets/vortex/vortex_skill.json	13	Route 13: Brier skill, ROC area and DJF-U60 anomaly correlation for the stratospheric vortex, per initialisation month over 24 reforecast winters
/assets/dailyfc/latest.json	16	Route 16 pointer: newest daily MEMBER initialisation. 51 members on a daily axis - percentiles, spread and absolute-threshold chances; no reforecast behind it, so no anomalies, no terciles and no skill score
/assets/sfs/sfs.json	15	Route 15: NOAA SFS Beta, TWELVE monthly leads from 31 members against a 1991-2025 reforecast - tercile probabilities raw and trend-adjusted, RPSS and ACC at every lead, the same on each WHOLE three-month season the start covers (DJF/MAM/JJA/SON, in skill_season and terciles_season), UNSCORED snowfall probabilities in `snow` and `snow_season` (ERA5 here has no snowfall, so they carry no RPSS and no ACC), and the Nino-3.4 plume from the model's own ocean (experiment beta1)
/assets/nino/nino.json	03	Route 03 ENSO: Nino 3.4, 3, 4, 1+2, relative Nino 3.4 and the IOD from the ten Copernicus systems - every member, each system's mean and ACC/RMSE vs ERA5, the one-vote consensus, and the strongest months in the ERA5 record
/assets/sfs/enso.json	15	Route 15: Nino 3.4, 3, 4, 1+2, relative Nino 3.4 and the IOD from the SFS ocean - beta1 and beta2, 31 members x 12 leads, each lead's ACC/RMSE vs ERA5 beside persistence and the ensemble's spread
/assets/sfs/beta2/sfs.json	15	Route 15: the same document for SFS experiment beta2, a different model on the same 31x12 live and 11x35 reforecast design; published only for starts whose calendar month beta2 has a reforecast for

Route 01 also publishes a per-initialisation catalog at /assets/seas5/<cycle>/catalog.json, listed in latest.json. It carries the colour lookup table every raster is encoded with, so the pictures are an invertible encoding rather than just pictures.

Terms of use

Short version: take it

Take it. Attribute it to forecastseason.com and, where you quote a figure, say which system it came from — every row names one, and the three do not agree. If you quote a probability, quote the skill beside it: a tercile probability without the reforecast behind it is the thing this site was built to stop publishing.

The underlying forecasts are ECMWF, Copernicus C3S and the NMME, under their own licences; the analog engine and every number derived here are ours. Both files regenerate with each build, and the routes behind them refresh monthly on the 9th.

Next exits

Route 06 — the scorecard behind the feed Route 01 — the raster tree and its catalog Route 07 — the 40 briefings About — what this site is