Predictability of boundary layer wind forecasts from four-dimensional variational data assimilation systems

In this talk we will quantify the inherent uncertainty of the Variational Doppler Radar and Lidar Assimilation System (VDRAS and VLAS) wind forecasts used as input to transport and dispersion (T&D) models. Model-generated forecast winds must be used as input to T&D models in areas of complex local surface forcing, such as urban areas. VDRAS and VLAS are adjoint based four-dimensional data assimilation systems that ingest Doppler radar, or lidar, radial wind observations along with in situ point measurements to produce a three-dimensional analysis and short-term forecast of meteorological fields. The wind field from VDRAS and/or VLAS analyses and forecasts can then be used to drive a variety of T&D applications.

A Doppler lidar has been making continuous radial wind observations over a limited portion of the National Capitol Region. Relatively little is known about predictability at very fine spatial scales, and this dataset provides a unique opportunity to examine the utility of fine-scale forecasts within an urban region, and to learn more about the predictability of boundary layers flows in these environments. In addition the Doppler radar at Sterling Virginia provides an additional dataset, though at a coarser resolution. To this end, real-time forecasts from VDRAS and VLAS are evaluated over the National Capitol Region and statistics demonstrating the models' predictability limits will be shown.