Forecast system development activities
Focus Improvement of the IRI Net Assessment Large-scale, global information Seasonal averages of temperature and precipitation
Net Assessment: precipitation
Net Assessment: temperature
Current activities Pattern-based correction of model output. Correct systematic errors. Calibration based on historical forecasts. Observed SST vs. forecast SST. Information beyond terciles Climatology and forecast probability density functions. New methods for weighting models.
NDJ temperature variability patterns (EOFs) Model Obs. MOS 1 2
correlation Model MOS
Beyond terciles Near-Normal Below Normal Above Normal Historical distribution (climatological distribution) (33.3%, 33.3%, 33.3%) FREQUENCY Forecast distribution (15%, 32%, 53%) The uncertainty can be expressed (quantitatively) in a number of ways: 1) Probabilities of discrete events Confidence level is varied. Interval length is fixed. 2) Error bars / confidence intervals Confidence level is fixed. Interval length is varied. 3) Probability distribution on a continuous scale Breakpoints of categories are determined by historical observations. The probabilities of this distribution are the climatological probabilities. Forecast distribution (say of the ensemble members at a point, or over a region) represent a shift in the range of possibilities. Now categorical probabilities are not equal – they differ from climatology. NORMALIZED RAINFALL Tercile probabilities provide a limited description of the forecast distribution Users may require other properties of the forecast distribution.
DJF(Nov) t2m -- counting
Example 1991
DJF(Nov) t2m -- Bayesian
DJF(Nov) t2m – pattern “CPT”
DJF(Nov) t2m – pattern “honest”
Example 1991
DJF(Nov) t2m – pattern “new”
Example 1991
Flexible format map room
Flexible format map room
Parametric pdfs Gaussian for seasonal temperature Power transformed Gaussian (Box-Cox) for seasonal precipitation
No skew
Some skew
More skew
Future work Add precipitation forecasts to flexible map room. Improved multi-model weights NOAA funded proposal with COLA/GMU and CPC.