Target Date Issue Date Lead Time

Forecast made for
located in
,
,

Probability of Exceedance

Probability Distribution

visit site

Temperature Flexible Seasonal Forecast

This seasonal forecasting system consists of probabilistic temperature seasonal forecasts based on the full estimate of the probability distribution.

Please refer to our licensing agreement for permission to use any IRI forecast material.

Probabilistic seasonal forecasts from multi-model ensembles through the use of statistical recalibration, based on the historical performance of those models, provide reliable information to a wide range of climate risk and decision making communities, as well as the forecast community. The flexibility of the full probability distributions allows to deliver interactive maps and point-wise distributions that become relevant to user-determined needs.

The default map shows globally the seasonal temperature forecast probability (colors between 0 and 1) of exceeding the 50th percentile of the distribution from historical 1981-2010 climatology. The quantitative value of that percentile is indicated by the contours. The forecast shown is the latest forecast made (e.g. Sep 2012) for the next season to come (e.g. Oct-Dec 2012). Five different seasons are forecasted and it is also possible to consult forecasts made previously. What makes the forecast flexible is that underlying the default map is the full probability distribution for the forecast and climatology. Therefore, the user can specify the historical percentile or a quantitative value (here temperature in ˚C) for probability of exceedance or non-exceedance. The climatological reference on which the forecast probability of (non-)exceeding is computed can be tailored by defining its starting and ending years.

Clicking on a point on the map will show the local culmulative distribution and probability distribution fucntions of the forecast (green) together with the climatological distribution (black).

New Colors Scales

Color scales have been changed for maps of probability of (non-) exceeding percentile thresholds. They have shifted from colors depicting intensity of the probabilities values (low/high probabilities associated with cold/warm colors) to colors indicating that the distribution forecasts colder (shades of blue) or hotter (shades of red) conditions than normal (moccasin).

Dataset Documentation

Distribution parameters: The means and standard deviations of the climatological observed reference and the forecast are available here. This is a parametrized dataset. The parameters are climStart and climEnd, respectively the starting year and the ending year defining the period of climatological observed reference. The default parameters are (1981) and (2010).

Observations: monthly temperature from Climate Anomaly Monitoring System

More Information

This forecasting system acts to correct spatial biases in the medians of ensemble forecasts for individual models. The model calibration is performed using multi-variate regression based on EOF structures, thus allowing for spatial corrections. Penalized regression methods help forecast reliability and select predictors effectively. In particular, Lasso regression, similar to ridge regression, eliminates more agressively poor predictors and retains much of the skill improvements of EOF regression.

This system has been based on several of the atmospheric general circulation models (AGCMs) that IRI uses in its “Net Assessment” seasonal forecast. The correction parameters in the system are estimated from retrospective forecasts that use forecast SST thereby explicitly incorporating SST forecast error and lead-time dependence.

The system produces a complete probability distribution, not just tercile probabilities so that probabilities for user-defined categories (based on user-defined climatology) are possible. Related to that, the distribution is quantitative so that, for example, probabilities of threshold exceedances can be queried.

Reference: Recalibrating and Combining Ensemble Predictions.Science and Technology Infusion Climate Bulletin, NOAA’s National Weather Service, 36th NOAA Annual Climate Diagnostics and Prediction Workshop. Michael K. Tippett, Tony Barnston, Lisa Goddard, Simon Mason, Malaquias Pena Mendez, Huug van den Dool

Instructions

Helpdesk

Contact help@iri.columbia.edu with any technical questions or problems with this Map Room.