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) · GraphCast (IFS init) · lead day 0

ALL · graphcast_ifs · 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, GRAP_v100, starting 2026-01-01 — 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
tmax00Zbilinear30d28 2 QC4.57 [4.46, 4.68]0.38-4.49 [-4.61, -4.37]5.07 [4.94, 5.20]7% [5%, 9%]14%26%——2026-08-01 → 2026-08-30
tmax00Zbilinear90d28 2 QC4.57 [4.42, 4.75]0.38-4.49 [-4.67, -4.34]5.07 [4.91, 5.26]7% [4%, 9%]14%26%——2026-08-01 → 2026-08-30
tmax00Zbilinear365d34 2 QC4.53 [4.40, 4.69]0.38-4.45 [-4.62, -4.32]5.08 [4.94, 5.23]7% [5%, 9%]14%27%——2026-01-01 → 2026-08-30
tmax00Zbilinearall34 2 QC4.53 [4.39, 4.70]0.38-4.45 [-4.62, -4.31]5.08 [4.94, 5.24]7% [5%, 9%]14%27%——2026-01-01 → 2026-08-30
tmax00Znearest30d28 2 QC4.77 [4.64, 4.94]0.40-4.41 [-4.58, -4.24]5.33 [5.17, 5.50]7% [5%, 9%]15%23%——2026-08-01 → 2026-08-30
tmax00Znearest90d28 2 QC4.77 [4.61, 5.03]0.40-4.41 [-4.66, -4.20]5.33 [5.13, 5.59]7% [5%, 9%]15%23%——2026-08-01 → 2026-08-30
tmax00Znearest365d34 2 QC4.71 [4.53, 4.93]0.40-4.40 [-4.62, -4.23]5.30 [5.14, 5.51]8% [5%, 10%]15%25%——2026-01-01 → 2026-08-30
tmax00Znearestall34 2 QC4.71 [4.54, 4.94]0.40-4.40 [-4.60, -4.23]5.30 [5.13, 5.53]8% [5%, 10%]15%25%——2026-01-01 → 2026-08-30
tmin00Zbilinear30d28 2 QC2.01 [1.91, 2.09]0.39+0.95 [0.79, 1.13]2.68 [2.52, 2.83]36% [33%, 39%]61%77%——2026-08-01 → 2026-08-30
tmin00Zbilinear90d28 2 QC2.01 [1.92, 2.12]0.39+0.95 [0.71, 1.14]2.68 [2.50, 2.85]36% [33%, 39%]61%77%——2026-08-01 → 2026-08-30
tmin00Zbilinear365d34 2 QC2.17 [1.96, 2.51]0.43+1.03 [0.81, 1.24]3.00 [2.60, 3.58]35% [30%, 38%]59%75%——2026-01-01 → 2026-08-30
tmin00Zbilinearall34 2 QC2.17 [1.95, 2.49]0.43+1.03 [0.78, 1.26]3.00 [2.58, 3.55]35% [31%, 38%]59%75%——2026-01-01 → 2026-08-30
tmin00Znearest30d28 2 QC2.06 [1.92, 2.16]0.40+0.97 [0.79, 1.15]2.77 [2.56, 2.94]37% [33%, 41%]59%75%——2026-08-01 → 2026-08-30
tmin00Znearest90d28 2 QC2.06 [1.92, 2.18]0.40+0.97 [0.71, 1.16]2.77 [2.56, 2.96]37% [32%, 41%]59%75%——2026-08-01 → 2026-08-30
tmin00Znearest365d34 2 QC2.24 [1.99, 2.63]0.43+1.05 [0.82, 1.25]3.09 [2.66, 3.67]35% [28%, 39%]57%73%——2026-01-01 → 2026-08-30
tmin00Znearestall34 2 QC2.24 [1.97, 2.62]0.43+1.05 [0.79, 1.27]3.09 [2.64, 3.65]35% [29%, 39%]57%73%——2026-01-01 → 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 GraphCast (IFS init) at ALL, lead day 0, 00Z bilinear tmax+10+50-5-102026-08-01: -3.6 °F2026-08-02: -4.23 °F2026-08-03: -5.27 °F2026-08-04: -4.17 °F2026-08-05: -5.12 °F2026-08-06: -3.99 °F2026-08-07: -4.33 °F2026-08-08: -4.13 °F2026-08-09: -3.93 °F2026-08-10: -3.94 °F2026-08-11: -4.59 °F2026-08-12: -5.24 °F2026-08-13: -4.67 °F2026-08-14: -4.98 °F2026-08-15: -3.97 °F2026-08-16: -4.3 °F2026-08-17: -4.56 °F2026-08-18: -4.51 °F2026-08-19: -4.48 °F2026-08-20: -4.23 °F2026-08-21: -4.66 °F2026-08-22: -4.16 °F2026-08-23: -4.58 °F2026-08-24: -4.83 °F2026-08-25: -4.31 °F2026-08-28: -5.16 °F2026-08-29: -5.44 °F2026-08-30: -4.28 °F2026-08-012026-08-30°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 (30 days, newest first) · download CSV
Climatological dayError °F
2026-08-30-4.28
2026-08-29-5.44
2026-08-28-5.16
2026-08-27—
2026-08-26—
2026-08-25-4.31
2026-08-24-4.83
2026-08-23-4.58
2026-08-22-4.16
2026-08-21-4.66
2026-08-20-4.23
2026-08-19-4.48
2026-08-18-4.51
2026-08-17-4.56
2026-08-16-4.30
2026-08-15-3.97
2026-08-14-4.98
2026-08-13-4.67
2026-08-12-5.24
2026-08-11-4.59
2026-08-10-3.94
2026-08-09-3.93
2026-08-08-4.13
2026-08-07-4.33
2026-08-06-3.99
2026-08-05-5.12
2026-08-04-4.17
2026-08-03-5.27
2026-08-02-4.23
2026-08-01-3.60

Error distribution

Distribution of the signed daily error in °F for GraphCast (IFS init) at ALL, lead day 0-10 to -9 °F: 0 days-9 to -8 °F: 0 days-8 to -7 °F: 0 days-7 to -6 °F: 0 days-6 to -5 °F: 5 days-5 to -4 °F: 22 days-4 to -3 °F: 7 days-3 to -2 °F: 0 days-2 to -1 °F: 0 days-1 to 0 °F: 0 days0 to 1 °F: 0 days1 to 2 °F: 0 days2 to 3 °F: 0 days3 to 4 °F: 0 days4 to 5 °F: 0 days5 to 6 °F: 0 days6 to 7 °F: 0 days7 to 8 °F: 0 days8 to 9 °F: 0 days9 to 10 °F: 0 daysdashed: P50 |e| 4.4 · P90 |e| 5.15 °F-100+10 °Fn = 34 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
-10.00-9.000
-9.00-8.000
-8.00-7.000
-7.00-6.000
-6.00-5.005
-5.00-4.0022
-4.00-3.007
-3.00-2.000
-2.00-1.000
-1.00+0.000
+0.00+1.000
+1.00+2.000
+2.00+3.000
+3.00+4.000
+4.00+5.000
+5.00+6.000
+6.00+7.000
+7.00+8.000
+8.00+9.000
+9.00+10.000

Month by month

00Z bilinear daily maximum, newest first.

Monthn MAE °FBias °F
2026-08284.49 -4.49
2026-0164.28 -4.28

Availability

ModelScored period DaysCoverage
GraphCast (IFS init)2026-01-01 → 2026-08-30 34
GraphCast (IFS init)2026-01-01 → 2026-08-30 34

Compared with other models

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

Other modelCommon days MAE difference °FDistinguishable
ECMWF AIFS Single 28 +1.18 [1.08, 1.28] yes
Aurora (GFS init) 2 -0.86 [-1.08, -0.64] yes
Aurora (IFS init) 3 -0.01 [-0.04, 0.04] no
FourCastNet v2 (GFS init) 2 +0.58 [0.23, 0.92] yes
FourCastNet v2 (IFS init) 3 +0.48 [-0.09, 0.76] no
NCEP GFS 28 +1.95 [1.81, 2.13] yes
GraphCast (GFS init) 2 -0.66 [-0.90, -0.42] yes
ECMWF IFS HRES 23 +1.34 [1.22, 1.41] yes
Pangu-Weather (GFS init) 2 +1.20 [0.83, 1.57] yes
Pangu-Weather (IFS init) 3 +1.45 [1.04, 1.79] yes

Cite this page

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

LongCastCheck (2026). Station-level verification of raw weather-model 2 m temperature forecasts: ALL (All stations (mean of daily station errors)), model graphcast_ifs, 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/graphcast_ifs/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