PSL Experimental Climate/Weather Forecasts

PSL generates experimental forecasts on time-scales of weeks to seasons as part of its research mission. The forecasts and associated information such as verifications are available on the following webpages. It is planned that these forecasts will eventually be transitioned to operational mode.
DescriptionSample Product
Model-Analogs (MA) and Linear Inverse Model (LIM) forecasts for Months 1-24
Experimental forecasts of numerous tropical fields, including precipitation, outgoing longwave radiation (OLR), sea surface temperature (SST), and sea surface height (SSH). Current Month 6 MA precipitation forecast and Niño3.4 Months 1-24 forecast from all models.
seasonal Forecast
Arctic sea-ice experimental forecast
Experimental, sea ice forecasts produced from a fully coupled ~9km ice(CICE5)-ocean(POP2)-land(CLM4.5)-atmosphere(WRF3.5) model called RASM-ESRL. The model is initialized with the NOAA Global Forecast System (GFS) analyses, CRYOSAT2 sea ice thickness, and Advanced Microwave Scanning Radiometer 2 (AMSR2) sea ice concentrations. The model is forced at the lateral boundaries by 3-hourly GFS forecasts of winds, temperature, and water vapor.
5day sea ice snow
Subseasonal Atlantic Hurricane Forecasts
Users can view the subseasonal forecasts of accumulated cyclone energy (ACE) in the Atlantic and west Atlantic basins at lead-times of 15-28 days (weeks 3-4). These forecasts are produced daily for lead-time validation periods starting in the beginning of July 2021 and going through the end of November. The forecast for a given date is the ensemble mean of the GEFSv12 (NOAA) and ECMWF’s (European Centre for Medium-Range Weather Forecasts) models. All of the sustained wind speed forecasts across the model ensemble members that: 1) fall within our Atlantic basin region, and 2) have lead times of greater than or equal to 15 days are used to calculate values of ACE for both the GEFSv12 and ECMWF models. Next, the values of ACE are standardized for the two models with respect to the typical mean and standard deviation of the forecasts for that time of the year. Plots compare the forecasts to the climatology (1990-2019).
Extended Winter Season Precipitation Forecasts for CONUS
Users can view probabilistic forecasts of seasonaly averaged wintertime precipitation (Nov-Mar) for the CONUS from 1982/83 to 2020/21. Forecasts are based on a statistical model using SST as a predictor and there is a 2nd set that uses a weighted ensemble of 4 different statistical models. Historical skill (the reforecast skill across CONUS for our statistical models versus the skill of the North American Multi-Model Ensemble (NMME) forecast) is also provided for the whole CONUS and selected subregions.
Global Tropical SST forecasts
A empirical dynamical technique, linear inverse modeling (LIM), is used to calculate forecasts regions of the tropical Pacific Ocean SSTs ("ENSO"), the tropical Atlantic Ocean SSTs, and the tropical Indian Ocean and the North Atlantic. LIM SST Forecasts are for 3 months to 12 months from the current date. Forecasts are presented as maps of global tropical SST forecasts and timeseries plots that include forecast history and verification.