High Resolution Forecasting and Observation System Experiments in the CASA Dallas-Fort Worth Urban Testbed

The Collaborative Adaptive Sensing of the Atmosphere (CASA) Dallas-Fort Worth Urban Testbed has been built as an example of a testbed built on the Network of Networks (NoN) concept. In the NoN concept state, local and private sector networks are combined with conventional National Weather Service (NWS) and Federal Aviation Administration (FAA) observations to maximize utility of the entire dataset. The DFW Testbed combines Federal observing resources such as radar systems from tri-agency WSR-88D radars, two FAA TDWR radars, seven (eight planned) CASA X-band radars, surface systems from the ASOS and AWOS, state mesonets, citizen and private weather sector surface observations (e.g., CWOP and Earth Networks), commercial systems (GST MoPED and Understory), commercial aviation observations from MDCRS, boundary layer wind profilers and radiometers.

These observing systems, combined through the NWS National Mesonet MADIS, and individual collection efforts provide valuation input for real-time high resolution nowcasting and short term forecasts, as well as Observing System Experiments (OSEs) to gauge the relative contributions of the various observing systems.

Real-time high resolution analyses and forecasts are being performed at the University of Oklahoma by the Center for Analysis and Prediction of Storms. Analysis are being done at 400-m horizontal grid spacing using the Advanced Regional Prediction System (ARPS) 3-D Variational (3DVAR) with cloud and hydrometeor analysis. Nowcasts are produced at 1-km horizontal resolution initialized with 3DVAR and cloud analysis using Incremental Analysis Updating with Variable-Dependent Timing (IAU-VDT) in the ARPS forecast model. In this way 2-hour forecasts are produced every 30-minutes with just 20-25 minutes latency on fewer than 200 CPU cores.

Important cases from the real-time system are being used to select cases for Observing System Experiments (OSEs) in three different frameworks: 1) using the operational configuration, 2) the Weather Research and Forecasting (WRF) model, and 3) the GSI-EnKF system. Among the cases being studied for data impacts include the Garland-Rowlett tornado of 2015 and the April 11, 2016 Hail Storm. Results of the data impacts for each of these cases are presented along with some recent real-time forecasting results.