NRL Four-dimensional Variational Radar Data Assimilation for Improved Near-term and Short-term Storm Prediction

In need of providing near-term (0-6h) and short-term (0-72h) storm forecasts for environmental assessment in support of US Navy operations, the Naval Research Laboratory (NRL) developed a new variational radar data assimilation system for the Navy’s Coupled Ocean/Atmosphere Mesoscale Prediction System (COAMPS). The system uses a four-dimensional variational (4DVar) technique for Doppler radar radial velocity data assimilation. To avoid the need for an adjoint/tangent linear model for the complex COAMPS microphysics, a three-dimensional variational (3DVar) data assimilation algorithm for radar reflectivity observations is also included within the 4DVar system with its own costfunction minimization procedures. The radar data assimilation is within the framework of the COAMPS-4DVar that was developed for operational COAMPS data assimilation. The system has been tested with quality-controlled data from the land-based Weather Surveillance Radar 88 Doppler (WSR-88D) radar network in the US and the US Navy’s shipboard Doppler radars (de-classified data). All of the test experiments were conducted over time periods lasting from a couple of days to a week using a cycled assimilation of radar observations along with conventional and satellite data. Results show notable improvement in both storm location and intensity for cases using the radar data assimilation. The storm forecast improvements are maintained from a few hours to about a day, depending on storm scales and radar data coverage. In addition, other model parameters, such as winds, temperature, and moisture, also show improvements in the forecasts with lead times from a few hours to a couple of days.