A new approach to local modeling

Versions of atmospheric meso-scale computer models have been available at National Weather Service Forecast Offices for several years allowing offices the ability to modify various model initialization and run-time parameters. The output from these efforts has received positive feedback from forecasters especially with features such as the sea breeze. However altering the options available within locally controlled models is often not user-friendly and does not incorporate real-time forecaster interaction. The Weather Forecast Office (WFO) in Tulsa, Oklahoma, has developed a system to change that paradigm. A graphical user interface (GUI) has been created to allow forecasters to interact with a local model and manipulate a wide variety of initialization and parameterization schemes yielding a plethora of model solutions.

Ideally, forecasters should be able to identify critical forecast elements within a specified time period, and then select the model initialization, physics and dynamics parameters which are most likely to best predict those critical elements. Additionally, an ensemble approach can be obtained by running the local model multiple times, incorporating both initialization and/or parameterization scheme alterations. This ability to alter physics and dynamic parameterization schemes differs from many current ensemble approaches. The output, available in near real-time, may provide insight that will allow a forecaster to gain confidence in forecast values for the critical elements.

The definition of critical elements will range across common ‘base' parameters such as temperature, dewpoint and wind speed and direction, along with modeled solutions of convective initiation and evolution, cloud cover, and precipitation forecasts. The ability for forecasters to focus model solution(s) on an appropriate element across a specified time range is expected to have positive effects within the appropriate gridded forecasts that compose the National Digital Forecast Database. More importantly, this added detail to the local gridded database will aid in decision support operations in both planned and emergency events, such as hazardous spill mitigation, wild fires and large outdoor venues.

The GUI will be shown with the variety of interactions explained. Additionally images will be presented showing how output can vary amongst both model initialization and parameterization schemes. Examples of importing the model output into the Graphical Forecast Editor will be shown, along with initial efforts toward model verification across critical forecast elements. The process of forecaster training and implementation into routine operations with WFO Tulsa will also be discussed.