CastCheck v0.3.1 · data through 2026-08-30

Daily verification · methodology v0.3.1

How far off was each weather model?

Over all available history, across 23 U.S. stations, the most accurate raw instantaneous temperature forecast 1 day ahead is GraphCast (IFS init): mean absolute error 2.86 °F [2.81, 2.92], bias −1.65 °F, on n = 198 scored days.

Scope: raw model output on the native 0.25° grid. The headline variable is the instantaneous 2 m temperature at 00/06/12/18 UTC, scored against the observed 2 m temperature at the same instant (ASOS routine METAR) — the same four instants on both sides, so nothing in the number depends on a model's own diurnal amplitude. Errors are computed in °C and shown in °F (methodology v0.3.1).

Window
all available history
Variable
instantaneous temperature
Init
00Z
Interpolation
nearest
Data through
2026-08-30
Updated
Next update

Leader, lead day 1

2.86°F

[2.81, 2.92] MAE

GraphCast (IFS init)

Persistence baseline

5.08°F

[4.51, 5.63]

yesterday's observation — the bar every model has to clear

Scored days

198

23 stations · 12 systems

in all available history, lead day 1

Systems ranked

2/12

next update

below n = 30: published, greyed, unranked

Mean absolute error with 95 % confidence intervals

Lead day 1 · all available history · 23 stations pooled · 00Z · nearest · instantaneous temperature. Whiskers are moving-block bootstrap intervals; models whose whiskers overlap are not distinguishable at this sample size. Baselines are drawn in grey and never ranked.

MAE by model with 95 % confidence intervals, lead day 1, all, 00Z, nearest0246810mean absolute error, °FGraphCast (IFS init)GraphCast (IFS init): 2.86 °F2.86GraphCast (GFS init)GraphCast (GFS init): 3.25 °F3.25FourCastNet v2 (GFS init)FourCastNet v2 (GFS init): 2.4 °F2.4ECMWF AIFS SingleECMWF AIFS Single: 2.51 °F2.51FourCastNet v2 (IFS init)FourCastNet v2 (IFS init): 2.68 °F2.68ECMWF IFS HRESECMWF IFS HRES: 2.83 °F2.83Aurora (IFS init)Aurora (IFS init): 2.88 °F2.88Pangu-Weather (GFS init)Pangu-Weather (GFS init): 2.97 °F2.97Pangu-Weather (IFS init)Pangu-Weather (IFS init): 3.02 °F3.02NCEP GFSNCEP GFS: 3.34 °F3.34Aurora (GFS init)Aurora (GFS init): 3.45 °F3.45Persistence (baseline)Persistence (baseline): 5.08 °F5.08

Source: CastCheck 0.1.0, methodology v0.3.1 · ECMWF open data, NOAA/NCEP GFS, NOAA/CIRA AIWP, NWS ASOS · data through 2026-08-30 · CC BY 4.0. Hover a bar for its value.

Lead day 1target = init + 1 d

permalink

All available history, all 23 stations pooled, 00Z initialization, nearest interpolation, instantaneous temperature against the observed 2 m temperature at the same instant (ASOS routine METAR). Lower MAE is better; bias is positive when the model is too warm. Every number links to its permanent page.

Skill is 1 − MAE ÷ MAE(persistence) computed on the days both have a value — the small print in that column is that denominator and the size of the intersection, so the number can be checked against the persistence row rather than contradicting it. The persistence row's own n is its whole record, marked all days. Skill, debiased (out-of-sample) repeats the score after removing a bias estimated on the 30 scored days before each day and applied forward, never on the day itself. An interval reads — when the bootstrap could not run: fewer than 28 scored days or fewer than 4 blocks. Hit-rate intervals are Wilson score intervals, not bootstraps.

Lead day 1, all available history, 00Z, nearest, instantaneous temperature.
RankModelMAE °FBias °F±3 °F SkillSkill, debiased (out-of-sample) nvs leader
1GraphCast (IFS init)graphcast_ifs2.86[2.81, 2.92]−1.65[−1.74, −1.57]60% +0.44[+0.36, +0.50]vs 5.10 (n=198)+0.65n=183 19822.8 stns · 13 QC★ lowest MAE in this view
2GraphCast (GFS init)graphcast_gfs3.25[3.05, 3.47]−1.99[−2.30, −1.66]54% +0.47[+0.38, +0.54]vs 6.11 (n=34)+0.65n=19 3422.5 stns= not distinguishable from the leader after the Holm correction within this table
—FourCastNet v2 (GFS init)fourcastnet_gfs2.40—−0.56—77% ——vs 3.04 (n=1)—n=0 115 stns
—ECMWF AIFS Singleaifs_single2.51[2.37, 2.66]−0.83[−0.94, −0.70]68% +0.18[+0.13, +0.23]vs 3.06 (n=29)+0.37n=14 2922.7 stns · 2 QC
—FourCastNet v2 (IFS init)fourcastnet_ifs2.68—−0.52—67% ——vs 3.51 (n=2)—n=0 219 stns
—ECMWF IFS HRESifs_hres2.83[2.69, 2.96]−0.16[−0.35, 0.06]64% +0.07[+0.00, +0.14]vs 3.06 (n=29)+0.24n=14 2922.7 stns · 2 QC
—Aurora (IFS init)aurora_ifs2.88—−1.80—58% ——vs 3.51 (n=2)—n=0 219 stns
—Pangu-Weather (GFS init)pangu_gfs2.97—−0.10—60% ——vs 3.04 (n=1)—n=0 115 stns
—Pangu-Weather (IFS init)pangu_ifs3.02—−1.29—59% ——vs 3.51 (n=2)—n=0 219 stns
—NCEP GFSgfs3.34[3.25, 3.43]+0.81[0.66, 1.00]55% −0.09[−0.17, −0.02]vs 3.06 (n=29)+0.24n=14 2922.7 stns · 2 QC
—Aurora (GFS init)aurora_gfs3.45—−2.19—47% ——vs 3.04 (n=1)—n=0 115 stns
—Persistence (baseline)persistence · baseline5.08[4.51, 5.63]−0.15[−0.41, 0.12]47% —— 215all days · 14 QC

+ model too warm − model too cold bias interval includes zero ★ lowest MAE = not distinguishable from the leader ▼ worse than the leader ▲ better than the leader — all three Holm-corrected within this table, over the family of comparisons against the leader; the uncorrected verdict for every pair is published in the pairwise table of each permanent link and in pairwise_latest.csv.gzn < 30 greyed and unranked CI is a 95 % moving-block bootstrap interval on the group's own days, and reads — when it could not be computed skill reads — when fewer than 10 days are common to the model and the baseline: the ratio of two means over a handful of shared days is not a number worth printing

Lead day 3target = init + 3 d

permalink

Same sampling as lead day 1. Skill is 1 − MAE ÷ MAE(persistence) on the days both have a value.

Lead day 3, all available history, 00Z, nearest, instantaneous temperature.
RankModelMAE °FBias °FSkillnvs leader
1GraphCast (IFS init)graphcast_ifs3.18[3.09, 3.28]−1.71[−1.85, −1.57]+0.57[+0.51, +0.62]vs 7.35 (n=196)19622.8 stns · 13 QC★ lowest MAE in this view
2GraphCast (GFS init)graphcast_gfs3.86[3.68, 4.01]−2.19[−2.82, −1.52]+0.58[+0.51, +0.64]vs 9.15 (n=33)3322.7 stns · 2 QC= not distinguishable from the leader after the Holm correction within this table
—ECMWF AIFS Singleaifs_single2.74—−0.72—+0.29—vs 3.85 (n=27)2722.6 stns · 2 QC
—ECMWF IFS HRESifs_hres3.11—+0.09—+0.19—vs 3.85 (n=27)2722.6 stns · 2 QC
—NCEP GFSgfs3.69—+1.06—+0.04—vs 3.85 (n=27)2722.6 stns · 2 QC
—Persistence (baseline)persistence · baseline7.43[6.54, 8.31]−0.42[−1.08, 0.29]—215all days · 14 QC

Lead day 5target = init + 5 d

permalink

Same sampling as lead day 1. Skill is 1 − MAE ÷ MAE(persistence) on the days both have a value.

Lead day 5, all available history, 00Z, nearest, instantaneous temperature.
RankModelMAE °FBias °FSkillnvs leader
1GraphCast (IFS init)graphcast_ifs3.88[3.66, 4.10]−1.66[−1.91, −1.43]+0.50[+0.45, +0.54]vs 7.77 (n=196)19622.8 stns · 12 QC★ lowest MAE in this view
2GraphCast (GFS init)graphcast_gfs5.06[4.62, 5.63]−1.18[−2.48, 0.16]+0.50[+0.44, +0.55]vs 10.21 (n=33)3322.7 stns · 2 QC= not distinguishable from the leader after the Holm correction within this table
—ECMWF AIFS Singleaifs_single2.98—−0.78—+0.26—vs 4.00 (n=25)2522.6 stns · 2 QC
—ECMWF IFS HRESifs_hres3.33—+0.11—+0.17—vs 4.00 (n=25)2522.6 stns · 2 QC
—NCEP GFSgfs3.86—+0.67—+0.04—vs 4.00 (n=25)2522.6 stns · 2 QC
—Persistence (baseline)persistence · baseline7.83[6.92, 8.71]−0.66[−1.66, 0.35]—215all days · 13 QC

Lead day 7target = init + 7 d

permalink

Same sampling as lead day 1. Skill is 1 − MAE ÷ MAE(persistence) on the days both have a value.

Lead day 7, all available history, 00Z, nearest, instantaneous temperature.
RankModelMAE °FBias °FSkillnvs leader
1GraphCast (IFS init)graphcast_ifs4.88[4.51, 5.30]−1.81[−2.31, −1.35]+0.43[+0.37, +0.47]vs 8.49 (n=194)19422.7 stns · 12 QC★ lowest MAE in this view
2GraphCast (GFS init)graphcast_gfs6.77[5.56, 8.57]+0.09[−1.98, 3.09]+0.38[+0.23, +0.47]vs 10.87 (n=33)3322.7 stns · 2 QC= not distinguishable from the leader after the Holm correction within this table
—ECMWF AIFS Singleaifs_single3.28—−1.17—+0.23—vs 4.23 (n=23)2322.6 stns · 2 QC
—ECMWF IFS HRESifs_hres3.64—+0.24—+0.14—vs 4.23 (n=23)2322.6 stns · 2 QC
—NCEP GFSgfs4.19—+0.37—+0.01—vs 4.23 (n=23)2322.6 stns · 2 QC
—Persistence (baseline)persistence · baseline8.41[7.39, 9.34]−0.81[−2.03, 0.36]—215all days · 14 QC

The same four samples as a daily maximum and minimum

Lead day 1, all available history, 00Z, nearest. Here the forecast's max/min of the four samples is scored against the observation's max/min of the same four instants. Like for like: whatever the four-sample definition misses, it misses on both sides. The comparison against the true NWS daily extremes is a different question and lives on the station and permanent-link pages.

Sampled daily maximumtmax_s

RankModelMAE °FBias °FSkillnvs leader
1GraphCast (IFS init)graphcast_ifs2.87[2.78, 2.98]−1.86[−1.98, −1.73]+0.43vs 5.06 (n=198)198★
2GraphCast (GFS init)graphcast_gfs3.21[2.98, 3.47]−2.53[−2.70, −2.34]+0.42vs 5.56 (n=34)34=
—ECMWF AIFS Singleaifs_single2.23[2.17, 2.29]−0.24[−0.37, −0.10]+0.20vs 2.80 (n=29)29
—FourCastNet v2 (GFS init)fourcastnet_gfs2.57—−1.88——vs 3.22 (n=1)1
—ECMWF IFS HRESifs_hres2.58[2.47, 2.71]−0.04[−0.38, 0.32]+0.08vs 2.80 (n=29)29
—FourCastNet v2 (IFS init)fourcastnet_ifs2.66—−1.22——vs 3.48 (n=2)2
—Pangu-Weather (GFS init)pangu_gfs2.91—−0.23——vs 3.22 (n=1)1
—Pangu-Weather (IFS init)pangu_ifs3.03—−0.85——vs 3.48 (n=2)2
—Aurora (IFS init)aurora_ifs3.34—−2.94——vs 3.48 (n=2)2
—NCEP GFSgfs3.57[3.32, 3.78]+1.81[1.45, 2.25]−0.28vs 2.80 (n=29)29
—Aurora (GFS init)aurora_gfs4.64—−4.50——vs 3.22 (n=1)1
—Persistence (baseline)persistence5.04[4.49, 5.58]−0.13[−0.41, 0.18]—215all days

Sampled daily minimumtmin_s

RankModelMAE °FBias °FSkillnvs leader
1GraphCast (IFS init)graphcast_ifs2.56[2.44, 2.69]−1.05[−1.17, −0.93]+0.42vs 4.40 (n=198)198★
2GraphCast (GFS init)graphcast_gfs3.21[2.98, 3.50]−1.00[−1.36, −0.56]+0.48vs 6.19 (n=34)34=
—Aurora (IFS init)aurora_ifs2.07—−0.45——vs 2.54 (n=2)2
—Pangu-Weather (IFS init)pangu_ifs2.13—−0.77——vs 2.54 (n=2)2
—FourCastNet v2 (GFS init)fourcastnet_gfs2.17—+0.62——vs 2.17 (n=1)1
—FourCastNet v2 (IFS init)fourcastnet_ifs2.21—+0.31——vs 2.54 (n=2)2
—Aurora (GFS init)aurora_gfs2.25—+0.06——vs 2.17 (n=1)1
—ECMWF IFS HRESifs_hres2.38[2.23, 2.53]−0.14[−0.34, 0.07]−0.02vs 2.34 (n=29)29
—ECMWF AIFS Singleaifs_single2.39[2.15, 2.58]−1.27[−1.51, −1.05]−0.02vs 2.34 (n=29)29
—Pangu-Weather (GFS init)pangu_gfs2.53—+0.17——vs 2.17 (n=1)1
—NCEP GFSgfs2.67[2.55, 2.80]+0.43[0.14, 0.78]−0.14vs 2.34 (n=29)29
—Persistence (baseline)persistence4.37[3.79, 4.92]−0.17[−0.46, 0.11]—215all days

Every model × every lead day

MAE in °F with the bias and n underneath, all available history, 00Z, nearest, instantaneous temperature. The sparkline is the same model's MAE across lead days 1–9 on a shared vertical scale.

Modeld0d1d2d3d4d5d6d7d8d9 lead 1–9
ECMWF AIFS Singleaifs_single2.41−0.88 · n=302.51−0.83 · n=292.64−0.80 · n=282.74−0.72 · n=272.82−0.76 · n=262.98−0.78 · n=253.15−0.99 · n=243.28−1.17 · n=233.57−1.47 · n=223.76−1.19 · n=21ECMWF AIFS Single MAE by lead day 1 to 9
Aurora (GFS init)aurora_gfs2.87−1.68 · n=23.45−2.19 · n=1———————— Aurora (GFS init) MAE by lead day 1 to 9
Aurora (IFS init)aurora_ifs2.74−1.67 · n=32.88−1.80 · n=23.14−1.75 · n=1——————— Aurora (IFS init) MAE by lead day 1 to 9
FourCastNet v2 (GFS init)fourcastnet_gfs2.69−1.06 · n=22.40−0.56 · n=1———————— FourCastNet v2 (GFS init) MAE by lead day 1 to 9
FourCastNet v2 (IFS init)fourcastnet_ifs2.76−1.46 · n=32.68−0.52 · n=22.35−0.09 · n=1——————— FourCastNet v2 (IFS init) MAE by lead day 1 to 9
NCEP GFSgfs3.03+0.66 · n=303.34+0.81 · n=293.44+0.85 · n=283.69+1.06 · n=273.67+0.87 · n=263.86+0.67 · n=253.98+0.74 · n=244.19+0.37 · n=234.43+0.27 · n=224.82+0.38 · n=21NCEP GFS MAE by lead day 1 to 9
GraphCast (GFS init)graphcast_gfs2.95−1.45 · n=353.25−1.99 · n=343.52−2.18 · n=333.86−2.19 · n=334.41−1.77 · n=335.06−1.18 · n=335.94−0.65 · n=336.77+0.09 · n=336.73+0.06 · n=337.62+1.69 · n=33GraphCast (GFS init) MAE by lead day 1 to 9
GraphCast (IFS init)graphcast_ifs2.65−1.40 · n=1992.86−1.65 · n=1983.02−1.72 · n=1973.18−1.71 · n=1963.45−1.73 · n=1963.88−1.66 · n=1964.38−1.65 · n=1954.88−1.81 · n=1945.52−1.92 · n=1936.38−1.74 · n=192GraphCast (IFS init) MAE by lead day 1 to 9
ECMWF IFS HRESifs_hres2.62−0.26 · n=302.83−0.16 · n=292.99−0.01 · n=283.11+0.09 · n=273.16+0.13 · n=263.33+0.11 · n=253.60+0.26 · n=243.64+0.24 · n=233.89−0.04 · n=224.24+0.63 · n=21ECMWF IFS HRES MAE by lead day 1 to 9
Pangu-Weather (GFS init)pangu_gfs2.77−0.35 · n=22.97−0.10 · n=1———————— Pangu-Weather (GFS init) MAE by lead day 1 to 9
Pangu-Weather (IFS init)pangu_ifs2.73−1.00 · n=33.02−1.29 · n=23.24−1.41 · n=1——————— Pangu-Weather (IFS init) MAE by lead day 1 to 9
Persistence (baseline)persistence—5.08−0.15 · n=2156.86−0.29 · n=2157.43−0.42 · n=2157.54−0.53 · n=2157.83−0.66 · n=2158.15−0.75 · n=2158.41−0.81 · n=2158.55−0.89 · n=2158.63−0.98 · n=215Persistence (baseline) MAE by lead day 1 to 9

Where the errors are

Both maps show a fixed quantity per station. Neither shows which model wins where: on samples this short the per-station winner is mostly noise, and a map of winners would invite a comparison the data cannot carry.

ECMWF IFS HRES bias

Mean bias of the fixed reference model ECMWF IFS HRES (lead day 1, instantaneous temperature, all window, 00Z, nearest). One model everywhere, so the colours compare stations, not models. Dot area grows with the number of scored days; fill is the mean bias (warm = model too warm, cool = too cold, grey = interval includes zero). The frame is a latitude/longitude graticule on an Albers projection.

Mean bias in °F of ECMWF IFS HRES, lead day 1, instantaneous temperature, all window, 00Z, nearest125°W115°W105°W95°W85°W75°W65°W25°N30°N35°N40°N45°N50°NKATL Atlanta Hartsfield: bias −1.09 °F, n = 29KATL−KAUS Austin Bergstrom: bias +1.47 °F, n = 29KAUS+KBOS Boston Logan: bias −0.83 °F, n = 26KBOS−KDCA Washington Reagan: bias −1.46 °F, n = 29KDCA−KDEN Denver Intl: bias +2.21 °F, n = 28KDEN+KDFW Dallas-Fort Worth: bias +2.16 °F, n = 29KDFW+KEWR Newark Liberty: bias −2.52 °F, n = 29KEWR−KIAH Houston Bush: bias −0.65 °F, n = 29KIAH−KLAS Las Vegas Harry Reid: bias −2.92 °F, n = 28KLAS−KLAX Los Angeles Intl: bias −2.05 °F, n = 28KLAX−KMIA Miami Intl: bias −1.62 °F, n = 29KMIA−KMSP Minneapolis-St Paul: bias +0.48 °F, n = 29KMSP+KMSY New Orleans Intl: bias +0.78 °F, n = 29KMSY+KNYC New York Central Park: bias +0.06 °F, n = 29KNYC+KOKC Oklahoma City: bias +4.07 °F, n = 29KOKC+KORD Chicago O'Hare: bias −1.27 °F, n = 29KORD−KPHL Philadelphia Intl: bias −2.60 °F, n = 29KPHL−KPHX Phoenix Sky Harbor: bias +0.75 °F, n = 28KPHX+KSAN San Diego Lindbergh: bias −3.73 °F, n = 28KSAN−KSAT San Antonio Intl: bias +1.57 °F, n = 29KSAT+KSEA Seattle-Tacoma: bias +1.37 °F, n = 28KSEA+KSFO San Francisco Intl: bias +1.78 °F, n = 28KSFO+KTTN Trenton Mercer: bias +0.31 °F, n = 29KTTN+

Source: CastCheck · station coordinates NWS/NOAA · Albers conic projection · stations.csv. Hover a dot for its numbers.

model too cold interval includes zero model too warm dot area ∝ scored days

ECMWF IFS HRES bias as a table
StationModelMAE °FBias °Fn
KATL Atlanta HartsfieldECMWF IFS HRES2.93−1.0929
KAUS Austin BergstromECMWF IFS HRES2.38+1.4729
KBOS Boston LoganECMWF IFS HRES2.64−0.8326
KDCA Washington ReaganECMWF IFS HRES3.02−1.4629
KDEN Denver IntlECMWF IFS HRES3.89+2.2128
KDFW Dallas-Fort WorthECMWF IFS HRES2.28+2.1629
KEWR Newark LibertyECMWF IFS HRES3.29−2.5229
KIAH Houston BushECMWF IFS HRES2.39−0.6529
KLAS Las Vegas Harry ReidECMWF IFS HRES4.09−2.9228
KLAX Los Angeles IntlECMWF IFS HRES2.30−2.0528
KMIA Miami IntlECMWF IFS HRES2.79−1.6229
KMSP Minneapolis-St PaulECMWF IFS HRES2.20+0.4829
KMSY New Orleans IntlECMWF IFS HRES2.42+0.7829
KNYC New York Central ParkECMWF IFS HRES2.30+0.0629
KOKC Oklahoma CityECMWF IFS HRES4.60+4.0729
KORD Chicago O'HareECMWF IFS HRES2.37−1.2729
KPHL Philadelphia IntlECMWF IFS HRES3.29−2.6029
KPHX Phoenix Sky HarborECMWF IFS HRES2.79+0.7528
KSAN San Diego LindberghECMWF IFS HRES3.76−3.7328
KSAT San Antonio IntlECMWF IFS HRES1.98+1.5729
KSEA Seattle-TacomaECMWF IFS HRES2.59+1.3728
KSFO San Francisco IntlECMWF IFS HRES2.72+1.7828
KTTN Trenton MercerECMWF IFS HRES2.22+0.3129

All-model mean bias

Mean bias averaged over every scored model at each station (lead day 1, instantaneous temperature, all window, 00Z, nearest). A station that is cold for all of them is a station property — elevation, coastline, a grid cell that is partly sea — not a model ranking. Dot area grows with the number of scored days; fill is the mean bias (warm = model too warm, cool = too cold). The frame is a latitude/longitude graticule on an Albers projection.

All-model mean bias in °F, lead day 1, instantaneous temperature, all window, 00Z, nearest125°W115°W105°W95°W85°W75°W65°W25°N30°N35°N40°N45°N50°NKATL Atlanta Hartsfield: bias −2.23 °F, n = 197KATL−KAUS Austin Bergstrom: bias +0.26 °F, n = 197KAUS+KBOS Boston Logan: bias −0.96 °F, n = 185KBOS−KDCA Washington Reagan: bias −1.64 °F, n = 197KDCA−KDEN Denver Intl: bias +0.28 °F, n = 196KDEN+KDFW Dallas-Fort Worth: bias −0.90 °F, n = 194KDFW−KEWR Newark Liberty: bias −1.61 °F, n = 196KEWR−KIAH Houston Bush: bias −0.40 °F, n = 198KIAH−KLAS Las Vegas Harry Reid: bias −5.23 °F, n = 196KLAS−KLAX Los Angeles Intl: bias −0.48 °F, n = 197KLAX−KMIA Miami Intl: bias −0.87 °F, n = 196KMIA−KMSP Minneapolis-St Paul: bias −2.04 °F, n = 198KMSP−KMSY New Orleans Intl: bias −1.69 °F, n = 198KMSY−KNYC New York Central Park: bias +0.24 °F, n = 197KNYC+KOKC Oklahoma City: bias +0.95 °F, n = 197KOKC+KORD Chicago O'Hare: bias −1.36 °F, n = 198KORD−KPHL Philadelphia Intl: bias −1.49 °F, n = 198KPHL−KPHX Phoenix Sky Harbor: bias −3.25 °F, n = 197KPHX−KSAN San Diego Lindbergh: bias −2.37 °F, n = 196KSAN−KSAT San Antonio Intl: bias −0.74 °F, n = 198KSAT−KSEA Seattle-Tacoma: bias −0.24 °F, n = 197KSEA−KSFO San Francisco Intl: bias −0.26 °F, n = 197KSFO−KTTN Trenton Mercer: bias +2.10 °F, n = 194KTTN+

Source: CastCheck · station coordinates NWS/NOAA · Albers conic projection · stations.csv. Hover a dot for its numbers.

model too cold interval includes zero model too warm dot area ∝ scored days

All-model mean bias as a table
StationModelsMAE °FBias °Fn
KATL Atlanta Hartsfield11 models3.18−2.23197
KAUS Austin Bergstrom11 models3.10+0.26197
KBOS Boston Logan8 models2.48−0.96185
KDCA Washington Reagan11 models2.97−1.64197
KDEN Denver Intl8 models3.27+0.28196
KDFW Dallas-Fort Worth11 models2.41−0.90194
KEWR Newark Liberty11 models2.92−1.61196
KIAH Houston Bush11 models2.27−0.40198
KLAS Las Vegas Harry Reid8 models5.67−5.23196
KLAX Los Angeles Intl8 models2.21−0.48197
KMIA Miami Intl11 models2.43−0.87196
KMSP Minneapolis-St Paul11 models3.29−2.04198
KMSY New Orleans Intl11 models2.60−1.69198
KNYC New York Central Park11 models2.59+0.24197
KOKC Oklahoma City11 models3.46+0.95197
KORD Chicago O'Hare11 models2.44−1.36198
KPHL Philadelphia Intl11 models2.80−1.49198
KPHX Phoenix Sky Harbor8 models4.29−3.25197
KSAN San Diego Lindbergh8 models3.32−2.37196
KSAT San Antonio Intl11 models2.61−0.74198
KSEA Seattle-Tacoma8 models2.40−0.24197
KSFO San Francisco Intl8 models2.46−0.26197
KTTN Trenton Mercer11 models3.42+2.10194

Data availability

Each model is scored only over its own available period (2026-01-01 → 2026-08-30); the windows above are therefore not identical across models. Pairwise comparisons on the permanent-link pages use common days only.

ModelPeriodScored daysCoverage
ECMWF AIFS Single2026-08-02 → 2026-08-3029
2026-01-012026-08-30
Aurora (GFS init)2026-08-30 → 2026-08-301
2026-01-012026-08-30
Aurora (IFS init)2026-08-29 → 2026-08-302
2026-01-012026-08-30
FourCastNet v2 (GFS init)2026-08-30 → 2026-08-301
2026-01-012026-08-30
FourCastNet v2 (IFS init)2026-08-29 → 2026-08-302
2026-01-012026-08-30
NCEP GFS2026-08-02 → 2026-08-3029
2026-01-012026-08-30
GraphCast (GFS init)2026-01-02 → 2026-08-3034
2026-01-012026-08-30
GraphCast (IFS init)2026-01-02 → 2026-08-30198
2026-01-012026-08-30
ECMWF IFS HRES2026-08-02 → 2026-08-3029
2026-01-012026-08-30
Pangu-Weather (GFS init)2026-08-30 → 2026-08-301
2026-01-012026-08-30
Pangu-Weather (IFS init)2026-08-29 → 2026-08-302
2026-01-012026-08-30
Persistence (baseline)2026-01-01 → 2026-08-30215
2026-01-012026-08-30

Stations23

full station table
  • KNYC — New York Central Park
  • KEWR — Newark Liberty
  • KPHL — Philadelphia Intl
  • KTTN — Trenton Mercer
  • KBOS — Boston Logan
  • KDCA — Washington Reagan
  • KATL — Atlanta Hartsfield
  • KMIA — Miami Intl
  • KORD — Chicago O'Hare
  • KMSP — Minneapolis-St Paul
  • KDFW — Dallas-Fort Worth
  • KIAH — Houston Bush
  • KAUS — Austin Bergstrom
  • KSAT — San Antonio Intl
  • KMSY — New Orleans Intl
  • KOKC — Oklahoma City
  • KDEN — Denver Intl
  • KPHX — Phoenix Sky Harbor
  • KLAS — Las Vegas Harry Reid
  • KLAX — Los Angeles Intl
  • KSAN — San Diego Lindbergh
  • KSFO — San Francisco Intl
  • KSEA — Seattle-Tacoma