AI Weather Quest — quintile-probability chart archive
Lower-quintile (< 20%) probability maps for four AI Weather Quest entries, pulled from the
ECMWF openCharts API so the same forecast can be compared model-by-model on identical base times,
forecast windows and variables.
Assembled alongside the S2S2D 2026 poster Operationalizing ECMWF AIFS Ensemble Forecasts for
Subseasonal Prediction in East Africa (Team Fahamu, ICPAC).
Models
| prefix |
model |
team |
note |
Fahamu04_ |
fp16FahamuAIFSv2 |
Fahamu (ICPAC) |
ours — AIFS ENS 2.0, FP16, 16-way chunking |
AIFS01gaia_ |
AIFSgaia |
AIFS (ECMWF) |
ECMWF’s own AIFS entry |
LP01lpm_ |
LPM |
LP |
independent entry |
CMAandFDU02FengshunAdjust_ |
FengshunAdjust |
CMAandFDU |
post-processed ("Adjust") variant |
What is here
96 PNGs = 4 models × 4 base times (09, 16, 23, 30 July 2026) × 3 variables × 2 forecast windows,
all at the < 20% quintile, global projection. Filenames are
<prefix>_<var>_<basedate>_win<N>_q20.png, so the same panel across models differs only in the
prefix. Per-model manifests with the ECMWF description strings are in manifests/.
How they were fetched
GET https://charts.ecmwf.int/opencharts-api/v1/products/s2s-competition-<PRODUCT>-quintile-<var>/
?base_time=2026-07-09T00:00:00Z # ISO-8601 — the compact form in the web UI is rejected
&valid_time=2026-08-02T18:00:00Z # window END at 18:00Z: base+24d = win1, base+31d = win2
&projection=opencharts_global&quintile=0.2
-> data.link.href = direct PNG (2000×1800)
Reading these maps
These are probability fields, not skill scores. A sharper, more structured field is not
automatically a better one — a confident forecast that is wrong scores worse than a diffuse one.
Use them to check that a model produces physically coherent structure; use the
leaderboard RPSS to judge skill.
Intercomparison — accumulated precipitation (tp, pr on the leaderboard)
09 Jul 2026 — window 1 — days 19–25 (+594 h)
|
|
fp16FahamuAIFSv2 (Fahamu (ICPAC))
 |
AIFSgaia (AIFS (ECMWF))
 |
LPM (LP)
 |
FengshunAdjust (CMAandFDU)
 |
09 Jul 2026 — window 2 — days 26–32 (+762 h)
|
|
fp16FahamuAIFSv2 (Fahamu (ICPAC))
 |
AIFSgaia (AIFS (ECMWF))
 |
LPM (LP)
 |
FengshunAdjust (CMAandFDU)
 |
16 Jul 2026 — window 1 — days 19–25 (+594 h)
|
|
fp16FahamuAIFSv2 (Fahamu (ICPAC))
 |
AIFSgaia (AIFS (ECMWF))
 |
LPM (LP)
 |
FengshunAdjust (CMAandFDU)
 |
16 Jul 2026 — window 2 — days 26–32 (+762 h)
|
|
fp16FahamuAIFSv2 (Fahamu (ICPAC))
 |
AIFSgaia (AIFS (ECMWF))
 |
LPM (LP)
 |
FengshunAdjust (CMAandFDU)
 |
23 Jul 2026 — window 1 — days 19–25 (+594 h)
|
|
fp16FahamuAIFSv2 (Fahamu (ICPAC))
 |
AIFSgaia (AIFS (ECMWF))
 |
LPM (LP)
 |
FengshunAdjust (CMAandFDU)
 |
23 Jul 2026 — window 2 — days 26–32 (+762 h)
|
|
fp16FahamuAIFSv2 (Fahamu (ICPAC))
 |
AIFSgaia (AIFS (ECMWF))
 |
LPM (LP)
 |
FengshunAdjust (CMAandFDU)
 |
30 Jul 2026 — window 1 — days 19–25 (+594 h)
|
|
fp16FahamuAIFSv2 (Fahamu (ICPAC))
 |
AIFSgaia (AIFS (ECMWF))
 |
LPM (LP)
 |
FengshunAdjust (CMAandFDU)
 |
30 Jul 2026 — window 2 — days 26–32 (+762 h)
|
|
fp16FahamuAIFSv2 (Fahamu (ICPAC))
 |
AIFSgaia (AIFS (ECMWF))
 |
LPM (LP)
 |
FengshunAdjust (CMAandFDU)
 |
Intercomparison — near-surface temperature (t2m, tas on the leaderboard)
09 Jul 2026 — window 1 — days 19–25 (+594 h)
|
|
fp16FahamuAIFSv2 (Fahamu (ICPAC))
 |
AIFSgaia (AIFS (ECMWF))
 |
LPM (LP)
 |
FengshunAdjust (CMAandFDU)
 |
09 Jul 2026 — window 2 — days 26–32 (+762 h)
|
|
fp16FahamuAIFSv2 (Fahamu (ICPAC))
 |
AIFSgaia (AIFS (ECMWF))
 |
LPM (LP)
 |
FengshunAdjust (CMAandFDU)
 |
16 Jul 2026 — window 1 — days 19–25 (+594 h)
|
|
fp16FahamuAIFSv2 (Fahamu (ICPAC))
 |
AIFSgaia (AIFS (ECMWF))
 |
LPM (LP)
 |
FengshunAdjust (CMAandFDU)
 |
16 Jul 2026 — window 2 — days 26–32 (+762 h)
|
|
fp16FahamuAIFSv2 (Fahamu (ICPAC))
 |
AIFSgaia (AIFS (ECMWF))
 |
LPM (LP)
 |
FengshunAdjust (CMAandFDU)
 |
23 Jul 2026 — window 1 — days 19–25 (+594 h)
|
|
fp16FahamuAIFSv2 (Fahamu (ICPAC))
 |
AIFSgaia (AIFS (ECMWF))
 |
LPM (LP)
 |
FengshunAdjust (CMAandFDU)
 |
23 Jul 2026 — window 2 — days 26–32 (+762 h)
|
|
fp16FahamuAIFSv2 (Fahamu (ICPAC))
 |
AIFSgaia (AIFS (ECMWF))
 |
LPM (LP)
 |
FengshunAdjust (CMAandFDU)
 |
30 Jul 2026 — window 1 — days 19–25 (+594 h)
|
|
fp16FahamuAIFSv2 (Fahamu (ICPAC))
 |
AIFSgaia (AIFS (ECMWF))
 |
LPM (LP)
 |
FengshunAdjust (CMAandFDU)
 |
30 Jul 2026 — window 2 — days 26–32 (+762 h)
|
|
fp16FahamuAIFSv2 (Fahamu (ICPAC))
 |
AIFSgaia (AIFS (ECMWF))
 |
LPM (LP)
 |
FengshunAdjust (CMAandFDU)
 |
Intercomparison — mean sea level pressure (mslp, mslp on the leaderboard)
09 Jul 2026 — window 1 — days 19–25 (+594 h)
|
|
fp16FahamuAIFSv2 (Fahamu (ICPAC))
 |
AIFSgaia (AIFS (ECMWF))
 |
LPM (LP)
 |
FengshunAdjust (CMAandFDU)
 |
09 Jul 2026 — window 2 — days 26–32 (+762 h)
|
|
fp16FahamuAIFSv2 (Fahamu (ICPAC))
 |
AIFSgaia (AIFS (ECMWF))
 |
LPM (LP)
 |
FengshunAdjust (CMAandFDU)
 |
16 Jul 2026 — window 1 — days 19–25 (+594 h)
|
|
fp16FahamuAIFSv2 (Fahamu (ICPAC))
 |
AIFSgaia (AIFS (ECMWF))
 |
LPM (LP)
 |
FengshunAdjust (CMAandFDU)
 |
16 Jul 2026 — window 2 — days 26–32 (+762 h)
|
|
fp16FahamuAIFSv2 (Fahamu (ICPAC))
 |
AIFSgaia (AIFS (ECMWF))
 |
LPM (LP)
 |
FengshunAdjust (CMAandFDU)
 |
23 Jul 2026 — window 1 — days 19–25 (+594 h)
|
|
fp16FahamuAIFSv2 (Fahamu (ICPAC))
 |
AIFSgaia (AIFS (ECMWF))
 |
LPM (LP)
 |
FengshunAdjust (CMAandFDU)
 |
23 Jul 2026 — window 2 — days 26–32 (+762 h)
|
|
fp16FahamuAIFSv2 (Fahamu (ICPAC))
 |
AIFSgaia (AIFS (ECMWF))
 |
LPM (LP)
 |
FengshunAdjust (CMAandFDU)
 |
30 Jul 2026 — window 1 — days 19–25 (+594 h)
|
|
fp16FahamuAIFSv2 (Fahamu (ICPAC))
 |
AIFSgaia (AIFS (ECMWF))
 |
LPM (LP)
 |
FengshunAdjust (CMAandFDU)
 |
30 Jul 2026 — window 2 — days 26–32 (+762 h)
|
|
fp16FahamuAIFSv2 (Fahamu (ICPAC))
 |
AIFSgaia (AIFS (ECMWF))
 |
LPM (LP)
 |
FengshunAdjust (CMAandFDU)
 |
All files
| base time |
window |
variable |
fp16FahamuAIFSv2 |
AIFSgaia |
LPM |
FengshunAdjust |
| 09 Jul 2026 |
1 |
Accumulated precipitation (tp) |
png |
png |
png |
png |
| 09 Jul 2026 |
1 |
Near-surface temperature (t2m) |
png |
png |
png |
png |
| 09 Jul 2026 |
1 |
Mean sea level pressure (mslp) |
png |
png |
png |
png |
| 09 Jul 2026 |
2 |
Accumulated precipitation (tp) |
png |
png |
png |
png |
| 09 Jul 2026 |
2 |
Near-surface temperature (t2m) |
png |
png |
png |
png |
| 09 Jul 2026 |
2 |
Mean sea level pressure (mslp) |
png |
png |
png |
png |
| 16 Jul 2026 |
1 |
Accumulated precipitation (tp) |
png |
png |
png |
png |
| 16 Jul 2026 |
1 |
Near-surface temperature (t2m) |
png |
png |
png |
png |
| 16 Jul 2026 |
1 |
Mean sea level pressure (mslp) |
png |
png |
png |
png |
| 16 Jul 2026 |
2 |
Accumulated precipitation (tp) |
png |
png |
png |
png |
| 16 Jul 2026 |
2 |
Near-surface temperature (t2m) |
png |
png |
png |
png |
| 16 Jul 2026 |
2 |
Mean sea level pressure (mslp) |
png |
png |
png |
png |
| 23 Jul 2026 |
1 |
Accumulated precipitation (tp) |
png |
png |
png |
png |
| 23 Jul 2026 |
1 |
Near-surface temperature (t2m) |
png |
png |
png |
png |
| 23 Jul 2026 |
1 |
Mean sea level pressure (mslp) |
png |
png |
png |
png |
| 23 Jul 2026 |
2 |
Accumulated precipitation (tp) |
png |
png |
png |
png |
| 23 Jul 2026 |
2 |
Near-surface temperature (t2m) |
png |
png |
png |
png |
| 23 Jul 2026 |
2 |
Mean sea level pressure (mslp) |
png |
png |
png |
png |
| 30 Jul 2026 |
1 |
Accumulated precipitation (tp) |
png |
png |
png |
png |
| 30 Jul 2026 |
1 |
Near-surface temperature (t2m) |
png |
png |
png |
png |
| 30 Jul 2026 |
1 |
Mean sea level pressure (mslp) |
png |
png |
png |
png |
| 30 Jul 2026 |
2 |
Accumulated precipitation (tp) |
png |
png |
png |
png |
| 30 Jul 2026 |
2 |
Near-surface temperature (t2m) |
png |
png |
png |
png |
| 30 Jul 2026 |
2 |
Mean sea level pressure (mslp) |
png |
png |
png |
png |
Provenance and licence
Charts © 2026 European Centre for Medium-Range Weather Forecasts (ECMWF), source
https://www.ecmwf.int, licensed CC BY 4.0 with the
ECMWF Terms of Use. Redistributed here
unmodified, with attribution, as that licence permits.
Leaderboard: https://aiweatherquest.ecmwf.int/leaderboards/ ·
Pipeline: https://github.com/icpac-igad/ea-aifs
