Calibration models statistically relate dynamical model reforecasts to observations over some historical period. These statistical relationships are then applied to correct real-time dynamical model forecasts. A bayesian joint probability method of calibration (EW + CRPS) is applied here, along with re-plotted ELR-calibrated forecasts courtesy of IRI. Contact Johnna Infanti (johnna.infanti@noaa.gov) or Justin Hicks (justin.hicks@noaa.gov) with any questions or comments.

North America | Global | Africa | Central America | Hispaniola | Hawaii

BJP Calibrated Skill Maps




Raw + Calibrated MME Forecast Anomalies/Probabilities

EW CRPS ELR

Weighted Unweighted Weighted Lag Unweighted Lag	Weighted Unweighted Weighted Lag Unweighted Lag	Weighted Unweighted Weighted Lag Unweighted Lag

Weighted Unweighted Weighted Lag Unweighted Lag	Weighted Unweighted Weighted Lag Unweighted Lag	Weighted Unweighted Weighted Lag Unweighted Lag

Weighted Unweighted Weighted Lag Unweighted Lag	Weighted Unweighted Weighted Lag Unweighted Lag	Weighted Unweighted Weighted Lag Unweighted Lag

Individual Model Anomalies (Raw)
North America | Global | Africa | Central America | Hispaniola | Hawaii

Individual Model Probabilities (BJR-Calibrated)
North America | Global | Africa | Central America | Hispaniola | Hawaii

Individual Model Probabilities (ELR-Calibrated)
North America | Global | Africa | Central America | Hispaniola | Hawaii

Individual Model Probabilities (2-Category)
North America | Global | Africa | Central America | Hispaniola | Hawaii

Individual Model Probabilities (3-Category)
North America | Global | Africa | Central America | Hispaniola | Hawaii

Verification (HSS, Expected Count)
Heidke Skill Score

Verification (ACC, Hit/Miss)
Anomaly Correlation Coefficient
Hit/Miss