Valid for Issued on Lead Time

Forecast made for

located in or near

Probability of Exceedance

Probability Distribution

visit site

Forecast Verification

Precipitation Flexible Seasonal NextGen Forecast

This Maproom shows full distribution rainfall forecast using a NextGen multi-model approach.

NextGen is a systematic general approach for designing, implementing, producing and verifying objective climate forecasts. It involves the identification of decision-relevant variables by the stakeholders; the analysis of the physical mechanisms, sources of predictability and suitable candidate predictors (in models and observations) for those key relevant variables. In those cases when prediction skill is high enough, NextGen helps select the best dynamical models for the region of interest through a process-based evaluation, and automatizes the generation and verification of tailored multi-model, statistically calibrated predictions at seasonal and sub-seasonal timescales.

The system takes advantage of the expertise of forecasters and local scientists at the country’s national meteorological service and universities, to maximize predictive skill and tailoring of the climate services generated by the process. Rather than focusing on probabilities of above normal, normal and below normal categories of total rainfall or mean temperature, NextGen also provides probabilities of exceeding (or not) particular thresholds of interest in the decision-making process, thus enabling users to forecast with the same system both mean and extreme values.

The models employed in this forecast are CanSIPSv2, COLA-RSMAS-CCSM4, GFDL-CM2p5-FLOR-A06, GFDL-CM2p5-FLOR-B01, NASA-GEOSS2S, NCEP-CFSv2, and the predictand is rainfall from BMD ENACTS. The default map shows, for the latest forecast made, the median value of the seasonal rainfall total forecast in the season. Users can use the Field menu to express the forecast in different ways, as follows:

• Rainfall: most likely seasonal total rainfall

• Anomaly: deviation in mm of the most likely seasonal total rainfall from yearly average of the most likely seasonal total rainfall predicted by the hindcast (1982-2010)

• Percent of Median: deviation in percentages of the most likely seasonal total rainfall from yearly median of the most likely seasonal total rainfall predicted by the hindcast (1982-2010)

• Dominant Tercile Probability: probability value of most probably forecast tercile category.

• Probability of non-/exceeding a Percentile: forecast probability of seasonal total rainfall to be below/above the historically observed (1982-2010) chosen percentile

• Probability of non-/exceeding a Precipitation amount: forecast probability of seasonal total rainfall to be below/above the chosen rainfall amount

The Layers button, showing when mousing over the map, will reveal in/active layers on the map. The Skill map is inactive by default and can be activated here, to be overlaid over the forecast map.

Clicking on the map will reveal information about the location clicked, as well as the full forecast distribution at that given location, compared with the historical distribution. Cumulative full distribution of the forecast (red) together with the climatological distribution (blue and black) for the forecast in view on the map shows under Probability of Exceedance, as well as the full probability distribution under Probability Distribution. If the SKill Layer is activated, a 3rd graph will show to present the comparison of hindcast performance versus actually observed seasonal total rainfall.

Dataset Documentation

Precipitation predictand:0.05˚ spatial resolution, precipitation in Bangladesh from BMD ENACTS.

Precipitation Forecasts:Forecasts and Hindcasts for indivudual models and NextGen aggregate from BMD.

Instructions

Helpdesks

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