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 →

All stations (mean of daily station errors) · FourCastNet v2 (GFS init) · lead day 2

ALL · fourcastnet_gfs · lead 2

Target day = initialization date + 2 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, FOUR_v200, 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, 1 carry a QC flag on the observation.

data through 2026-08-30
updated 2026-08-31T03:52:06+00:00
next update 2026-08-31T11:00:00+00:00
methodology v0.2

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.

Variable	Init	Method	Window	n	MAE °F	Bias °F	RMSE °F	±1 °F	±2 °F	±3 °F	Skill	Skill, debiased	Period
tmax	12Z	bilinear	30d	1 1 QC	3.68 [3.68, 3.68]	-3.68 [-3.68, -3.68]	4.52 [4.52, 4.52]	17% [17%, 17%]	26%	43%	-0.09	—	2026-08-30 → 2026-08-30
tmax	12Z	bilinear	90d	1 1 QC	3.68 [3.68, 3.68]	-3.68 [-3.68, -3.68]	4.52 [4.52, 4.52]	17% [17%, 17%]	26%	43%	-0.09	—	2026-08-30 → 2026-08-30
tmax	12Z	bilinear	365d	1 1 QC	3.68 [3.68, 3.68]	-3.68 [-3.68, -3.68]	4.52 [4.52, 4.52]	17% [17%, 17%]	26%	43%	-0.09	—	2026-08-30 → 2026-08-30
tmax	12Z	bilinear	all	1 1 QC	3.68 [3.68, 3.68]	-3.68 [-3.68, -3.68]	4.52 [4.52, 4.52]	17% [17%, 17%]	26%	43%	-0.09	—	2026-08-30 → 2026-08-30
tmax	12Z	nearest	30d	1 1 QC	3.78 [3.78, 3.78]	-3.78 [-3.78, -3.78]	4.64 [4.64, 4.64]	13% [13%, 13%]	26%	48%	-0.12	—	2026-08-30 → 2026-08-30
tmax	12Z	nearest	90d	1 1 QC	3.78 [3.78, 3.78]	-3.78 [-3.78, -3.78]	4.64 [4.64, 4.64]	13% [13%, 13%]	26%	48%	-0.12	—	2026-08-30 → 2026-08-30
tmax	12Z	nearest	365d	1 1 QC	3.78 [3.78, 3.78]	-3.78 [-3.78, -3.78]	4.64 [4.64, 4.64]	13% [13%, 13%]	26%	48%	-0.12	—	2026-08-30 → 2026-08-30
tmax	12Z	nearest	all	1 1 QC	3.78 [3.78, 3.78]	-3.78 [-3.78, -3.78]	4.64 [4.64, 4.64]	13% [13%, 13%]	26%	48%	-0.12	—	2026-08-30 → 2026-08-30
tmin	12Z	bilinear	30d	1 1 QC	2.69 [2.69, 2.69]	+0.66 [0.66, 0.66]	3.80 [3.80, 3.80]	30% [30%, 30%]	52%	74%	+0.18	—	2026-08-30 → 2026-08-30
tmin	12Z	bilinear	90d	1 1 QC	2.69 [2.69, 2.69]	+0.66 [0.66, 0.66]	3.80 [3.80, 3.80]	30% [30%, 30%]	52%	74%	+0.18	—	2026-08-30 → 2026-08-30
tmin	12Z	bilinear	365d	1 1 QC	2.69 [2.69, 2.69]	+0.66 [0.66, 0.66]	3.80 [3.80, 3.80]	30% [30%, 30%]	52%	74%	+0.18	—	2026-08-30 → 2026-08-30
tmin	12Z	bilinear	all	1 1 QC	2.69 [2.69, 2.69]	+0.66 [0.66, 0.66]	3.80 [3.80, 3.80]	30% [30%, 30%]	52%	74%	+0.18	—	2026-08-30 → 2026-08-30
tmin	12Z	nearest	30d	1 1 QC	2.74 [2.74, 2.74]	+0.60 [0.60, 0.60]	3.92 [3.92, 3.92]	26% [26%, 26%]	61%	74%	+0.16	—	2026-08-30 → 2026-08-30
tmin	12Z	nearest	90d	1 1 QC	2.74 [2.74, 2.74]	+0.60 [0.60, 0.60]	3.92 [3.92, 3.92]	26% [26%, 26%]	61%	74%	+0.16	—	2026-08-30 → 2026-08-30
tmin	12Z	nearest	365d	1 1 QC	2.74 [2.74, 2.74]	+0.60 [0.60, 0.60]	3.92 [3.92, 3.92]	26% [26%, 26%]	61%	74%	+0.16	—	2026-08-30 → 2026-08-30
tmin	12Z	nearest	all	1 1 QC	2.74 [2.74, 2.74]	+0.60 [0.60, 0.60]	3.92 [3.92, 3.92]	26% [26%, 26%]	61%	74%	+0.16	—	2026-08-30 → 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 FourCastNet v2 (GFS init) at ALL, lead day 2, 00Z bilinear tmax: no scored days yet.

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 (0 days, newest first) · download CSV

Climatological day	Error °F

Error distribution

Distribution of the signed daily error in °F for FourCastNet v2 (GFS init) at ALL, lead day 2: no scored days yet.

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.

Availability

Model	Scored period	Days	Coverage
FourCastNet v2 (GFS init)	2026-08-30 → 2026-08-30	1
FourCastNet v2 (GFS init)	2026-08-30 → 2026-08-30	1

Cite this page

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

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

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