Since 2005, the Meteorological Development Laboratory (MDL) has been collaborating with the
National Center for Atmospheric Research (NCAR)
to demonstrate the functionality and capability of NCAR’s AutoNowCaster (ANC) system within National Weather Service operations.
Initially, this work focused on the degree to which human forecasters’ interactions with ANC improved the thunderstorm nowcast
accuracy above and beyond that provided by ANC alone. The Dallas/Fort Worth Weather Forecast Office (WFO) played a large role
in the first five years of this collaboration. Today, the program focuses on assessing ANC’s ability to provide guidance in
the form of gridded thunderstorm nowcasts. For that purpose, MDL is running ANC over the Golden Triangle, an area that
encompasses Chicago’s, New York City’s, and Atlanta’s Air Traffic Management hubs. This demonstration involves the Chicago WFO,
whose forecasters can interact with ANC. The results of this experiment are intended to provide guidance on the utility of ANC
in generating convective products that address NextGen requirements. The goal is to develop a concept of operations where ANC
feeds the NextGen/4-D weather cube.
ANC runs at the MDL and generates 1-hour forecasts of convective initiation, growth, and decay every five to six minutes.
Currently, these forecast products are sent in real time to the Chicago and Melbourne WFOs. Forecasters at these WFOs can
interact with ANC using the Advanced Weather Interactive Processing System (AWIPS) Display 2 Dimensions (D2D) interface to
create surface boundaries. The forecasters’ inputs are instantly sent to ANC via the AWIPS Local Data Acquisition and
Dissemination system. The resulting forecast products--which incorporate the forecasters’ inputs--are then disseminated
back to the WFOs. These products are also sent to the Chicago Center Weather Service Unit via web pages.
The ANC system is a suite of automated applications that produces 0-1 hour thunderstorm nowcasting
predictor fields derived from observation-based feature detections, numerical weather prediction
model output, and human forecaster input. Data ingested by ANC include WSR-88D level 2, satellite,
sounding, and surface observations. The software applications in the ANC environment include
algorithms for boundary-storm interactions, cumulus cloud detection and growth, boundary-relative
shear profile, boundary-relative updraft strength and storm trends and tracks. A fuzzy-logic
application is used to combine the weighted outputs from the various analysis algorithms to produce
time- and space-specific forecasts of thunderstorm initiation, growth and decay. A more comprehensive
and detailed overview of ANC is described in a paper titled
NCAR Auto-Nowcast System.