Raw model output on the native 0.25° grid — no MOS, no bias correction, no post-processing. Scores understate operational forecast quality. Read the fairness statement →

CastCheck methodology v0.2 · data through 2026-08-30

All stations (mean of daily station errors) · Pangu-Weather (GFS init) · lead day 0

ALL · pangu_gfs · lead 0

Target day = initialization date + 0 days. This page is a permanent link: the numbers change as history accumulates, the address and the methodology version do not.

Scores use the latest model version segment, PANG_v100, starting 2026-08-28 — days before that ran a different version of the model and are excluded rather than silently pooled. Of the scored days, 2 carry a QC flag on the observation.

JSON for this page CSV of the daily errors Cite this page

Every number for this combination

All four windows × both initializations × both interpolations × both variables. Errors are computed in °C and shown in °F; CI is a 95 % moving-block bootstrap interval. MAE debiased and skill, debiased remove the constant offset of the window first, so a station that the model always reads a degree too warm is not scored as if it had no skill — compare the two skill columns to see how much of the error is a fixed offset. n QC next to n counts days whose observation carries a QC flag.

VariableInitMethod Windown MAE °F MAE debiased Bias °F RMSE °F ±1 °F±2 °F±3 °F Skill Skill, debiased Period
tmax00Zbilinear30d2 2 QC3.74 [3.61, 3.87]0.17-3.07 [-3.24, -2.90]4.30 [4.10, 4.50]4% [0%, 9%]20%48%——2026-08-29 → 2026-08-30
tmax00Zbilinear90d2 2 QC3.74 [3.61, 3.87]0.17-3.07 [-3.24, -2.90]4.30 [4.10, 4.50]4% [0%, 9%]20%48%——2026-08-29 → 2026-08-30
tmax00Zbilinear365d2 2 QC3.74 [3.61, 3.87]0.17-3.07 [-3.24, -2.90]4.30 [4.10, 4.50]4% [0%, 9%]20%48%——2026-08-29 → 2026-08-30
tmax00Zbilinearall2 2 QC3.74 [3.61, 3.87]0.17-3.07 [-3.24, -2.90]4.30 [4.10, 4.50]4% [0%, 9%]20%48%——2026-08-29 → 2026-08-30
tmax00Znearest30d2 2 QC3.92 [3.84, 3.99]0.14-3.24 [-3.39, -3.10]4.59 [4.31, 4.85]7% [0%, 13%]20%46%——2026-08-29 → 2026-08-30
tmax00Znearest90d2 2 QC3.92 [3.84, 3.99]0.14-3.24 [-3.39, -3.10]4.59 [4.31, 4.85]7% [0%, 13%]20%46%——2026-08-29 → 2026-08-30
tmax00Znearest365d2 2 QC3.92 [3.84, 3.99]0.14-3.24 [-3.39, -3.10]4.59 [4.31, 4.85]7% [0%, 13%]20%46%——2026-08-29 → 2026-08-30
tmax00Znearestall2 2 QC3.92 [3.84, 3.99]0.14-3.24 [-3.39, -3.10]4.59 [4.31, 4.85]7% [0%, 13%]20%46%——2026-08-29 → 2026-08-30
tmin00Zbilinear30d2 2 QC2.59 [2.45, 2.73]1.03+0.63 [-0.40, 1.65]3.42 [3.18, 3.65]26% [22%, 30%]54%70%——2026-08-29 → 2026-08-30
tmin00Zbilinear90d2 2 QC2.59 [2.45, 2.73]1.03+0.63 [-0.40, 1.65]3.42 [3.18, 3.65]26% [22%, 30%]54%70%——2026-08-29 → 2026-08-30
tmin00Zbilinear365d2 2 QC2.59 [2.45, 2.73]1.03+0.63 [-0.40, 1.65]3.42 [3.18, 3.65]26% [22%, 30%]54%70%——2026-08-29 → 2026-08-30
tmin00Zbilinearall2 2 QC2.59 [2.45, 2.73]1.03+0.63 [-0.40, 1.65]3.42 [3.18, 3.65]26% [22%, 30%]54%70%——2026-08-29 → 2026-08-30
tmin00Znearest30d2 2 QC2.69 [2.59, 2.78]1.07+0.54 [-0.53, 1.61]3.57 [3.35, 3.78]28% [17%, 39%]54%70%——2026-08-29 → 2026-08-30
tmin00Znearest90d2 2 QC2.69 [2.59, 2.78]1.07+0.54 [-0.53, 1.61]3.57 [3.35, 3.78]28% [17%, 39%]54%70%——2026-08-29 → 2026-08-30
tmin00Znearest365d2 2 QC2.69 [2.59, 2.78]1.07+0.54 [-0.53, 1.61]3.57 [3.35, 3.78]28% [17%, 39%]54%70%——2026-08-29 → 2026-08-30
tmin00Znearestall2 2 QC2.69 [2.59, 2.78]1.07+0.54 [-0.53, 1.61]3.57 [3.35, 3.78]28% [17%, 39%]54%70%——2026-08-29 → 2026-08-30

+ model too warm · − model too cold · bias interval includes zero · ★ lowest MAE · = not distinguishable from the leader · ▼ significantly worse · n < 30 greyed and unranked · CI is a 95 % moving-block bootstrap interval.

Daily error, last 90 days

Daily forecast minus observed error in °F for Pangu-Weather (GFS init) at ALL, lead day 0, 00Z bilinear tmax+5+2.50-2.5-52026-08-29: -3.24 °F2026-08-30: -2.9 °F2026-08-292026-08-30°F, forecast − observed
Forecast − observed, °F, 00Z bilinear daily maximum. The dashed line is zero; the thick line is a centred 7-day mean, drawn only where the whole window is present. Gaps are days with no score, never interpolation. Hover a point for its date and value.
The same series as a table (2 days, newest first) · download CSV
Climatological dayError °F
2026-08-30-2.90
2026-08-29-3.24

Error distribution

Distribution of the signed daily error in °F for Pangu-Weather (GFS init) at ALL, lead day 0-5 to -4.5 °F: 0 days-4.5 to -4 °F: 0 days-4 to -3.5 °F: 0 days-3.5 to -3 °F: 1 day-3 to -2.5 °F: 1 day-2.5 to -2 °F: 0 days-2 to -1.5 °F: 0 days-1.5 to -1 °F: 0 days-1 to -0.5 °F: 0 days-0.5 to 0 °F: 0 days0 to 0.5 °F: 0 days0.5 to 1 °F: 0 days1 to 1.5 °F: 0 days1.5 to 2 °F: 0 days2 to 2.5 °F: 0 days2.5 to 3 °F: 0 days3 to 3.5 °F: 0 days3.5 to 4 °F: 0 days4 to 4.5 °F: 0 days4.5 to 5 °F: 0 daysdashed: P50 |e| 3.07 · P90 |e| 3.21 °F-50+5 °Fn = 2 days
All scored days for 00Z bilinear daily maximum, 20 equal bins. The dashed marks are the 50th and 90th percentiles of the absolute error, mirrored on both sides. Hover a bar for its bin and count.
Bin counts as a table
From °FTo °FDays
-5.00-4.500
-4.50-4.000
-4.00-3.500
-3.50-3.001
-3.00-2.501
-2.50-2.000
-2.00-1.500
-1.50-1.000
-1.00-0.500
-0.50+0.000
+0.00+0.500
+0.50+1.000
+1.00+1.500
+1.50+2.000
+2.00+2.500
+2.50+3.000
+3.00+3.500
+3.50+4.000
+4.00+4.500
+4.50+5.000

Month by month

00Z bilinear daily maximum, newest first.

Monthn MAE °FBias °F
2026-0823.07 -3.07

Availability

ModelScored period DaysCoverage
Pangu-Weather (GFS init)2026-08-29 → 2026-08-30 2
Pangu-Weather (GFS init)2026-08-29 → 2026-08-30 2

Compared with other models

Paired bootstrap on the days both models cover (90d, 00Z, bilinear, daily maximum). Negative means Pangu-Weather (GFS init) has the lower MAE. "Distinguishable" means the 95 % interval excludes zero.

Other modelCommon days MAE difference °FDistinguishable
ECMWF AIFS Single 2 +0.24 [0.02, 0.47] yes
Aurora (GFS init) 2 -2.06 [-2.21, -1.92] yes
Aurora (IFS init) 2 -1.24 [-1.64, -0.84] yes
FourCastNet v2 (GFS init) 2 -0.62 [-0.65, -0.60] yes
FourCastNet v2 (IFS init) 2 -0.44 [-0.54, -0.34] yes
NCEP GFS 2 +0.81 [0.58, 1.05] yes
GraphCast (GFS init) 2 -1.86 [-1.99, -1.73] yes
GraphCast (IFS init) 2 -1.20 [-1.57, -0.83] yes
ECMWF IFS HRES 2 +0.06 [-0.08, 0.20] no
Pangu-Weather (IFS init) 2 +0.07 [-0.07, 0.21] no

Cite this page

ShortCastCheck, ALL pangu_gfs lead 0, methodology v0.2, accessed 2026-08-31, https://castcheck.zifanzhang.com/station/ALL/model/pangu_gfs/lead/0/

LongCastCheck (2026). Station-level verification of raw weather-model 2 m temperature forecasts: ALL (All stations (mean of daily station errors)), model pangu_gfs, lead day 0. Methodology version 0.2, schema version 0.1, data through 2026-08-30. Accessed 2026-08-31. https://castcheck.zifanzhang.com/station/ALL/model/pangu_gfs/lead/0/

Accessed2026-08-31 (data through 2026-08-30; page built 2026-08-31T03:52:06+00:00)

JSON CSV all scores methodology v0.2 licence