(Upcoming season: Probabilistic Forecasts | Rainfall Rate Forecasts | Skill Scores)
ASMC’s experimental forecasts of precipitation for the ASEAN region are issued near the middle of each month. These forecasts are for the next 3-month period of average rainfall in mm/day for each 2.5x2.5 square-grid located within the ASEAN regional area.
To generate the 3-months seasonal forecasts, CPT (Climate Predictability Tool) is used together with ECHAM4.5 Precipitation Model Output as predictors and NASA's GPCP Rainfall Rate as predictands. The climatology training period for the datasets is from 1979 to 2009.
The forecasts, generated using a type of multiple linear regression technique, provide probabilities that the precipitation fall in the lowest one-third, middle one-third or the highest one-third of the climatological distribution of rainfall. This climatological distribution is determined by the observations for the season in question for a period of about 30 years (1979 to 2009) using the NASA’s GPCP Rainfall Rate. On top of the forecast probabilities, the actual rainfall rate forecasts are also given for each grid.
Skill scores of the regression model used to generate the forecast are also provided, based on hindcast simulations using prescribed observed rainfall rates.
Click on the links on the left to view the probabilistic forecasts, the rainfall rate forecasts, and the corresponding skill scores. The forecasts were made using models constructed by the Principal Components Regression technique and the constructed models were validated using the cross-validation method. A more detailed explanation on the tool and the techniques used is available here.
2. ABOUT CPT TOOL
CPT is a Windows application created by IRI (The International Research Institute for Climate and Society), used for constructing a seasonal climate forecast model, performing model validation, and producing forecasts given updated data.
Its design has been tailored for producing seasonal climate forecasts using model output statistic (MOS) corrections to climate predictions from general circulation model (GCM), or for producing forecasts using fields of sea-surface temperatures. Although the software is specifically tailored for these applications, it can be used in more general settings to perform canonical correlation analysis (CCA) or principal components regression (PCR) on any data, and for any application.
For more information on the datasets used (training period: 1979 to 2009), please visit:
For more information on the tool, please visit iri.columbia.edu/climate/tools/cpt.