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 4

ALL · graphcast_ifs · lead 4

Target day = initialization date + 4 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, 1 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
tmax00Zbilinear30d25 1 QC5.32 [4.94, 5.68]0.63-5.05 [-5.44, -4.71]6.02 [5.60, 6.38]5% [3%, 7%]13%23%-0.38+0.562026-08-05 → 2026-08-29
tmax00Zbilinear90d25 1 QC5.32 [4.88, 5.75]0.63-5.05 [-5.49, -4.67]6.02 [5.51, 6.45]5% [3%, 7%]13%23%-0.38+0.562026-08-05 → 2026-08-29
tmax00Zbilinear365d31 1 QC5.22 [4.83, 5.66]0.76-4.94 [-5.39, -4.53]5.99 [5.56, 6.39]6% [4%, 9%]15%25%-0.07+0.652026-01-05 → 2026-08-29
tmax00Zbilinearall31 1 QC5.22 [4.81, 5.60]0.76-4.94 [-5.35, -4.49]5.99 [5.55, 6.36]6% [4%, 10%]15%25%-0.07+0.652026-01-05 → 2026-08-29
tmax00Znearest30d25 1 QC5.49 [5.11, 5.90]0.65-5.00 [-5.45, -4.58]6.26 [5.83, 6.67]6% [4%, 8%]15%23%-0.42+0.552026-08-05 → 2026-08-29
tmax00Znearest90d25 1 QC5.49 [5.07, 6.00]0.65-5.00 [-5.52, -4.52]6.26 [5.73, 6.77]6% [3%, 8%]15%23%-0.42+0.552026-08-05 → 2026-08-29
tmax00Znearest365d31 1 QC5.38 [4.97, 5.85]0.78-4.91 [-5.41, -4.44]6.20 [5.74, 6.66]7% [4%, 10%]17%25%-0.10+0.642026-01-05 → 2026-08-29
tmax00Znearestall31 1 QC5.38 [4.94, 5.79]0.78-4.91 [-5.37, -4.43]6.20 [5.72, 6.61]7% [4%, 10%]17%25%-0.10+0.642026-01-05 → 2026-08-29
tmax12Zbilinear30d25 1 QC5.53 [5.14, 5.96]0.63-5.30 [-5.77, -4.93]6.26 [5.80, 6.68]5% [3%, 7%]12%21%-0.43+0.562026-08-05 → 2026-08-29
tmax12Zbilinear90d25 1 QC5.53 [5.06, 6.03]0.63-5.30 [-5.83, -4.88]6.26 [5.69, 6.74]5% [3%, 7%]12%21%-0.43+0.562026-08-05 → 2026-08-29
tmax12Zbilinear365d30 1 QC5.47 [5.05, 5.94]0.67-5.25 [-5.71, -4.87]6.26 [5.83, 6.71]6% [4%, 7%]13%23%-0.16+0.692026-01-05 → 2026-08-29
tmax12Zbilinearall30 1 QC5.47 [5.05, 5.91]0.67-5.25 [-5.68, -4.86]6.26 [5.80, 6.66]6% [4%, 7%]13%23%-0.16+0.692026-01-05 → 2026-08-29
tmax12Znearest30d25 1 QC5.71 [5.29, 6.19]0.64-5.25 [-5.76, -4.80]6.48 [5.99, 6.96]6% [4%, 8%]13%22%-0.48+0.552026-08-05 → 2026-08-29
tmax12Znearest90d25 1 QC5.71 [5.22, 6.29]0.64-5.25 [-5.83, -4.72]6.48 [5.88, 7.08]6% [4%, 8%]13%22%-0.48+0.552026-08-05 → 2026-08-29
tmax12Znearest365d30 1 QC5.62 [5.20, 6.15]0.68-5.21 [-5.73, -4.75]6.45 [5.97, 6.99]6% [4%, 8%]13%23%-0.19+0.682026-01-05 → 2026-08-29
tmax12Znearestall30 1 QC5.62 [5.17, 6.12]0.68-5.21 [-5.69, -4.77]6.45 [5.97, 6.94]6% [4%, 8%]13%23%-0.19+0.682026-01-05 → 2026-08-29
tmin00Zbilinear30d25 1 QC2.38 [2.21, 2.55]0.61+0.63 [0.33, 0.92]3.12 [2.90, 3.33]31% [28%, 35%]54%69%+0.24+0.232026-08-05 → 2026-08-29
tmin00Zbilinear90d25 1 QC2.38 [2.19, 2.60]0.61+0.63 [0.31, 0.96]3.12 [2.89, 3.39]31% [28%, 35%]54%69%+0.24+0.232026-08-05 → 2026-08-29
tmin00Zbilinear365d31 1 QC2.59 [2.26, 3.12]0.72+0.88 [0.43, 1.50]3.48 [2.98, 4.18]30% [26%, 34%]52%66%+0.38+0.582026-01-05 → 2026-08-29
tmin00Zbilinearall31 1 QC2.59 [2.27, 3.16]0.72+0.88 [0.41, 1.53]3.48 [2.95, 4.28]30% [26%, 34%]52%66%+0.38+0.582026-01-05 → 2026-08-29
tmin00Znearest30d25 1 QC2.39 [2.18, 2.58]0.61+0.65 [0.33, 0.93]3.17 [2.91, 3.41]31% [27%, 36%]56%68%+0.24+0.232026-08-05 → 2026-08-29
tmin00Znearest90d25 1 QC2.39 [2.16, 2.62]0.61+0.65 [0.30, 0.97]3.17 [2.90, 3.46]31% [26%, 37%]56%68%+0.24+0.232026-08-05 → 2026-08-29
tmin00Znearest365d31 1 QC2.63 [2.25, 3.20]0.71+0.88 [0.44, 1.46]3.57 [3.01, 4.32]29% [24%, 34%]52%65%+0.37+0.582026-01-05 → 2026-08-29
tmin00Znearestall31 1 QC2.63 [2.26, 3.29]0.71+0.88 [0.42, 1.48]3.57 [2.99, 4.44]29% [24%, 34%]52%65%+0.37+0.582026-01-05 → 2026-08-29
tmin12Zbilinear30d25 1 QC2.29 [2.14, 2.44]0.65+0.40 [0.08, 0.66]2.98 [2.77, 3.18]31% [27%, 35%]55%70%+0.27+0.182026-08-05 → 2026-08-29
tmin12Zbilinear90d25 1 QC2.29 [2.12, 2.47]0.65+0.40 [0.05, 0.70]2.98 [2.76, 3.21]31% [27%, 35%]55%70%+0.27+0.182026-08-05 → 2026-08-29
tmin12Zbilinear365d30 1 QC2.43 [2.18, 2.79]0.67+0.53 [0.17, 0.93]3.22 [2.85, 3.75]30% [26%, 34%]54%68%+0.39+0.602026-01-05 → 2026-08-29
tmin12Zbilinearall30 1 QC2.43 [2.18, 2.81]0.67+0.53 [0.13, 0.95]3.22 [2.81, 3.80]30% [26%, 34%]54%68%+0.39+0.602026-01-05 → 2026-08-29
tmin12Znearest30d25 1 QC2.31 [2.14, 2.48]0.64+0.43 [0.11, 0.68]3.04 [2.80, 3.25]31% [28%, 35%]54%71%+0.26+0.192026-08-05 → 2026-08-29
tmin12Znearest90d25 1 QC2.31 [2.12, 2.51]0.64+0.43 [0.06, 0.72]3.04 [2.78, 3.29]31% [27%, 36%]54%71%+0.26+0.192026-08-05 → 2026-08-29
tmin12Znearest365d30 1 QC2.48 [2.19, 2.93]0.65+0.54 [0.18, 0.90]3.30 [2.88, 3.87]29% [25%, 33%]53%68%+0.38+0.612026-01-05 → 2026-08-29
tmin12Znearestall30 1 QC2.48 [2.18, 2.95]0.65+0.54 [0.16, 0.91]3.30 [2.85, 3.93]29% [25%, 34%]53%68%+0.38+0.612026-01-05 → 2026-08-29

+ 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 4, 00Z bilinear tmax+10+50-5-102026-08-05: -4.47 °F2026-08-06: -4.73 °F2026-08-07: -5.67 °F2026-08-08: -4.5 °F2026-08-09: -4.14 °F2026-08-10: -3.97 °F2026-08-11: -4.54 °F2026-08-12: -6.24 °F2026-08-13: -3.65 °F2026-08-14: -5.21 °F2026-08-15: -4.99 °F2026-08-16: -3.78 °F2026-08-17: -4.69 °F2026-08-18: -5.13 °F2026-08-19: -5.93 °F2026-08-20: -5.32 °F2026-08-21: -4.88 °F2026-08-22: -5.54 °F2026-08-23: -5.95 °F2026-08-24: -6.01 °F2026-08-25: -4.96 °F2026-08-26: -5.49 °F2026-08-27: -5.69 °F2026-08-28: -6.26 °F2026-08-29: -4.5 °F2026-08-052026-08-29°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 (25 days, newest first) · download CSV
Climatological dayError °F
2026-08-29-4.50
2026-08-28-6.26
2026-08-27-5.69
2026-08-26-5.49
2026-08-25-4.96
2026-08-24-6.01
2026-08-23-5.95
2026-08-22-5.54
2026-08-21-4.88
2026-08-20-5.32
2026-08-19-5.93
2026-08-18-5.13
2026-08-17-4.69
2026-08-16-3.78
2026-08-15-4.99
2026-08-14-5.21
2026-08-13-3.65
2026-08-12-6.24
2026-08-11-4.54
2026-08-10-3.97
2026-08-09-4.14
2026-08-08-4.50
2026-08-07-5.67
2026-08-06-4.73
2026-08-05-4.47

Error distribution

Distribution of the signed daily error in °F for GraphCast (IFS init) at ALL, lead day 4-10 to -9 °F: 0 days-9 to -8 °F: 0 days-8 to -7 °F: 0 days-7 to -6 °F: 4 days-6 to -5 °F: 10 days-5 to -4 °F: 12 days-4 to -3 °F: 4 days-3 to -2 °F: 1 day-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| 4.96 · P90 |e| 6.01 °F-100+10 °Fn = 31 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
-7.00-6.004
-6.00-5.0010
-5.00-4.0012
-4.00-3.004
-3.00-2.001
-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
2026-08255.05 -5.05
2026-0164.49 -4.49

Availability

ModelScored period DaysCoverage
GraphCast (IFS init)2026-01-05 → 2026-08-29 31
GraphCast (IFS init)2026-01-05 → 2026-08-29 31
GraphCast (IFS init)2026-01-05 → 2026-08-29 30
GraphCast (IFS init)2026-01-05 → 2026-08-29 30

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 25 +1.62 [1.27, 1.90] yes
NCEP GFS 25 +2.06 [1.75, 2.42] yes
ECMWF IFS HRES 19 +1.81 [1.40, 2.14] yes
Persistence (baseline) 25 +1.46 [0.60, 2.28] yes

Cite this page

ShortCastCheck, ALL graphcast_ifs lead 4, methodology v0.2, accessed 2026-08-31, https://castcheck.zifanzhang.com/station/ALL/model/graphcast_ifs/lead/4/

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 4. 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/4/

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