Model Verification
Saved forecasts, ensemble members and observations in one comparison.
Model verification · KLGA
Saved forecasts compared with observations
Selections apply to both the table and exported plot. Models with no verified cases for these filters are omitted. All scores use the selected station; precipitation is not silently substituted from a different airport.
| Model | Forecast cases | Valid hours | MAE °F | Bias °F | RMSE °F |
|---|
Loading saved verification…
Models appear when saved forecasts have matching observations in the selected period and lead range. Daily scores require a complete day captured in advance and its climate report.
Uses the measure, period, lead range and sorting selected above. Created in your browser from saved verification statistics; no model downloads. Changing filters clears the preview.
Hourly temperatures use routine METAR within 15 minutes of forecast valid time. Daily high/low use 24 complete forecast hours saved before the standard-time day starts, compared with the same station’s completed climate report. Missing reports remain unverified. MAE = mean absolute error; bias = forecast minus observed; RMSE = square root of mean squared error. Each case has equal weight; counts differ by model. Cycle captures are labeled separately: their initialization is provider-reported and their contents may combine runs. Newly logged forecasts are never backdated. Scores refresh in background approximately every 30 minutes; cumulative CSVs update every six hours. REFS members and means share this comparison with all available models. Precipitation/timing scores appear only where those metrics have been logged.