PSD  »  Research Teams  »  Forecast and Modeling Development

Related Links

Experimental Products

GEFS Reforecast Version 2
HFIP GFS/EnKF Forecast Graphics
NCEP Ensemble Forecast

Forecast and Modeling Development Team

Leads: Jeff Whitaker and Robert Pincus

Forecasters rely on sophisticated computer models to provide an estimate of what the weather will look like tomorrow, next week, and beyond. With this information people can plan what to wear, what roads to avoid, or how to prepare for a hurricane that might be coming their way. Decision makers such as water managers and emergency planners can also decide what steps should be taken to prepare for water scarcity from droughts, or the hazards of flooding, for example. Forecasts have gotten substantially better in the past decade as the models and their ability to make use of observations has improved, but they are still not perfect. To keep communities safe and to support decision makers, continual research and technological advances are needed to improve the accuracy and extent of predictions.

PSD’s Forecast and Modeling Development Team performs research and development to improve NOAA operational forecast products. We improve the models that underpin forecasts, the data assimilation systems that provide starting points for forecasts, and the statistical systems that improve today’s forecasts using errors in past forecasts.

Given the uncertainty in weather-climate predictions, we are particularly interested in improving probabilistic forecasts. Such forecasts permit a spectrum of users to make informed decisions guided by the models. We are heavily involved in data assimilation and prediction with ensembles (multiple simulations of future weather). The focus on probability forecasts also motivates our work to more accurately represent the sources of uncertainty in forecasts. We consider and improve the methods to simulate how that uncertainty depends on today’s initial estimate of the environment, how the uncertainty depends on the errors caused by approximations built into the weather forecast models, and how the uncertainty estimates can be improved through statistical post-processing.

We also work to improve understanding and simulation of small-scale weather processes, especially those related to clouds, the transfer of electromagnetic radiation through the atmosphere, and energy exchanges between atmosphere and oceans.

Current Research Activities

Developing advanced methods for data assimilation that optimally merge past forecasts with new observations to estimate the state of the atmosphere, either now or in the past, using methods that account for each day’s particular weather. Building more accurate and efficient treatments of radiation in the atmosphere.
Improving representations of the uncertainty in weather forecasts due to approximations, interactions between the land and ocean surface, and other causes. Assimilating high-resolution observations for hurricane and severe-weather forecasts.
Assembling datasets that lead to reliable and skillful probabilistic forecast products for end-users, especially reanalyses and reforecasts. Exploring methods for representing clouds and convection that adapt smoothly to model scale.