Combining mesoscale, nowcast, and CFD model output in near real-time for protecting urban areas and buildings from releases of hazardous airborne materials S. Swerdlin, T. Warner, J. Copeland, D. Hahn, J. Sun, R. Sharman, Y. Liu, J. Knievel, A. Crook, M. Raines National Center for Atmospheric Research swerdlin@ucar.edu J. Weil University of Colorado, Boulder
Concept Combine weather models at several scales to provide detailed urban wind field awareness, including very fine (1-5 m) scales for critical assets Develop hazardous material sensor network and algorithms to detect releases and predict plume transport and diffusion Integrate these two systems for real-time operations Provide support and training to emergency managers Multi-use: monitor and reverse-locate sources of industrial pollution; support fire fighting and flood management
Computing urban wind fields For operational system being developed by NCAR in Washington, DC, using three models to compute rooftop fields RTFDDA: : MM5-based Real-Time Four- Dimensional Data Assimilation VDRAS: Variational Doppler RADAR Assimilation System VLAS: Variational LIDAR Assimilation System Blend these onto a common grid Use blend to provide initial and lateral boundary conditions for city- and building- aware models
Rooftop model 1: MM5-based RT-FDDA LIDAR SURFACE OBS e.g., new 12 h forecast every hour at 500 m res, using real-time obs RT- FDDA time TAMDAR Forecast SATELLITE RADAR UPPER AIR QuickSCAT scatterometer
Rooftop models 2&3: VDRAS and RADAR LIDAR VLAS RADAR/LIDAR assimilation system. Uses 4 Dimensional Variational Assimilation (4DVAR) to calculate microscale 3D analysis of boundary layer winds. Radial winds Desired 3-D winds
Example: VDRAS coupled to plume model VDRAS wind vectors show convergence line below formation of thunderstorm cells
Example 1: VDRAS coupled to plume model (cont) Emergency response application. Two simulated releases 30 minutes apart: plume model coupled to VDRAS winds 1557 LT release Release height 10 m 1 kg inert, nonbuoyant gas 15 June 1998 Wash. D.C. 1629 LT release
RT-FDDA VDRAS VLAS skimming flow models L = 10-1000 km L = 10-100 km Spatial-temporal blending scheme L = 1-10 km skimming flow master grid (updated every 5 mins)
skimming flow master grid (covers large urban area) Provides 3-D, time-varying, initial, and lateral boundary conditions to buildingaware models, every 5 minutes Urban canopy flow: 10 x 10 km tiles Building flow: 1.5 x 1.5 km tiles LANL s QUIC-Urb, updated every 5 mins L = 10-20 m L = 2-5 m CFDRC s CFD-Urban, updated every 10 mins
Example 2: VLAS applied at the neighborhood scale Notional plume CTI Doppler lidar
VLAS wind vectors in Washington, D.C.
Closely separated simulated releases have distinct patterns
Simulated releases from same location, at 5-min intervals
Sensitivity of simulated plume prediction to atmospheric stability conditions Neutral/Convective atmosphere Stable atmosphere
Using high-resolution winds to compute threat zone in near real-time x e.g., agent released at X would require 1 min to impact the target 1 km X t 3 m t 1 m X
Dual use for LIDARs: : mapping plumes Continuously collected LIDAR output examined for presence of interesting plumes
Conclusion Scheme of creating multi-resolution rooftop blend, and using this to provide background and forcing conditions for building-aware models seems to be effective More verification study is needed to determine skill of coupled NWP, plume-model system