Development and Evaluation of Cloud-Based Weather Event Simulation Capabilities

Since 2015 staff members at the Cooperative Institute for Mesoscale Meteorological Studies (CIMMS) at the University of Oklahoma in conjunction with the Warning Decision Training Division (WDTD) of the National Weather Service (NWS) have deployed and continually refresh archived data playback capabilities for AWIPS-2 to NWS field offices and regional headquarters locations. This WES-2 Bridge software (Weather Event Simulator for AWIPS-2) is akin to a “flight simulator” for NWS meteorologists as it mimics the behavior of their operational AWIPS-2 system through the use of archived datasets from previous weather events.

Unfortunately, the deployment of AWIPS and WES-2 Bridge in the NWS Weather Forecast Office (WFO) environment presents significant challenges, which partly result from operational security requirements of AWIPS and ever-increasing sizes of data archives that outpace available storage capacity. Each forecast office locally archives the same data independently and inefficiently. Archived cases easily consume 250-500 GB (1 TB+ for a multi-day event). Disk space limits individual data cases, rather than scientific or operational needs. Significant time and expense is involved in transferring data to the simulator.

Secondarily, simulations are mostly performed in isolation because there is one simulator per WFO. However, the actual forecast and warning process is inherently distributed and collaborative among multiple forecasters in multiple locations. Many service assessments have documented the need to enhance communication and collaboration among all these collaborators; yet little operationally-representative and contextual-based training and exercises exist to develop and maintain collaborative skills.

To address both of these limitations, CIMMS staff are exploring proof-of-concept deployments of the WES-2 Bridge and accompanying AWIPS-2 software in various cloud-computing platforms. Specific tests involve recent builds of AWIPS-2 and WES-2 Bridge and associated archived datasets including GOES-16, MRMS, FLASH, forecast models, WSR-88D, and standard observational data. This paper reports preliminary results from these evaluation activities and presents potential solutions that cloud environments could offer to the challenges experienced by NWS users of the WES in training and research applications.