Real-Time Modeling and Data Assimilation
Figure 1. Horizontal cross section from a simulation by the unmodified WRF Model version 2.0.3.1, valid at 2300 UTC 14 July 1998 on Domain 3 of 4. Horizontal divergence at the lowest model level is contoured (positive in red and negative in blue), and terrain elevation (m AMSL) is shaded. The solid black lines mark the perimeter of the Great Salt Lake and the border between Utah and Nevada. This shows the existence of noise in the boundary-layer wind field.
In FY06, RAL modelers continued to work with the U.S. Army Test and Evaluation Command (ATEC), Defense Threat Reduction Agency (DTRA), Defense Advanced Research Project Agency (DARPA), and other public and industrial agencies to study and improve the 4-Dimensional Weather (4DWX) system's Real-Time Four-Dimensional Data Assimilation and forecast (RTFDDA) system. RTFDDA is a unique scheme for assimilating all available observations from a number of data feeds and producing a new analysis and forecast every one to three hours. A highlight of this work in FY06 was the transition of RTFDDA from MM5 to the Advanced Research core of the Weather Research and Forecast (WRF) model.
FY06 Accomplishments
Figure 2. Horizontal cross section from a simulation by the WRF Model version 2.0.3.1 modified to include explicit, monotonic, 6th-order diffusion, valid at 2300 UTC 14 July 1998 on Domain 3 of 4. Horizontal divergence at the lowest model level is contoured (positive in red and negative in blue), and terrain elevation (m AMSL) is shaded. The solid black lines mark the perimeter of the Great Salt Lake and the border between Utah and Nevada. This shows the lack of noise in the boundary-layer wind field.
RAL modelers have led the effort to implement the RTFDDA system in WRF. Other components of 4DWX (e.g., a series of preprocessors, algorithms for data quality control, a series of post processors, and secondary applications) have also been integrated into WRF. In addition, several special categories such as playa, white sand, and lava, which characterize the environments around some of the western ATEC ranges, were added to the model's land surface database. Another improvement to the model resulted from observing local wind circulations that are sometimes induced by heterogeneity in land surface around the test ranges. RAL scientists discovered that, when wind is weak and the boundary layer is well-mixed, numerical noise can grow and dominate forecasts made by the current version of the WRF model (Fig. 1). To reduce the problem, an explicit 6th-order horizontal diffusion scheme was added to the model; this improvement will be available to the community in an upcoming official release of the WRF Model (Fig. 2). Implementing RTFDDA within WRF allows RTFDDA users to take advantage of recent advances in numerical weather prediction and ultimately streamlines the transfer of future scientific breakthroughs from developers to users. Since the fall of 2005, two WRF-based RTFDDA systems have been installed at Dugway Proving Ground, UT and Aberdeen Test Center, MD. They are running in parallel to the MM5-based RTFDDA to facilitate evaluation and refinement of the WRF-RTFDDA system. In spite of missing a few additional refinements that are part of the operational MM5-RTFDDA, WRF-RTFDDA verification statistics are proving very close to those of its MM5 counterpart.