Assimilation of radar reflectivity data in the Met Office convective-scale forecast system

The Met Office has developed a convective-scale NWP based nowcasting system, which uses a 1.5 km resolution version of the Met Office Unified Model with hourly-cycling 4D-Var data assimilation. The demonstration system assimilates Doppler winds in 4D-Var, and reflectivity data via latent heat nudging, in addition to surface, aircraft and satellite observations.

The nowcasting system is being developed to assimilate radar reflectivity data directly in the Met Office incremental 4D-Var system. This has involved the design of a radar reflectivity operator, and its linearization and adjoint, observation monitoring and quality control, and enhancements to the linear and adjoint physics of the perturbation forecast model, including the introduction of 3D rainrate increments to the perturbation forecast model. Linearisation tests have been used to determine appropriate conversion rates of ice and liquid water cloud increments to rainrate increments for the nowcasting system.

The control variables used in the current system are velocity potential, stream function, unbalanced pressure and a single humidity variable representing total water. Recent developments to the covariance statistics system will allow investigation of the benefit of specific control variables for hydrometeors.

The current formulation of the reflectivity operator uses a simple Z-R relation to forward model reflectivity from rainrate. Minimising the linearization error in the forward model and avoiding biases in the assimilation has been a key aspect of designing the system. Assimilating Z in dBZ units can lead to a dry bias, so alternative formulations have been investigated. Research has been undertaken to allow the use of observations with large innovations, and where the background is dry. The use of ice observations is being investigated.

The Met Office C-band weather radar network is currently being upgraded. All radars are now Doppler capable, and when the project is completed, all radars will have dual-polarization capability. The extra polarimetric information will support great improvements in data quality, and allow the development of an improved observation operator.

The Met Office is collaborating closely with the University of Reading on techniques which will improve the quality of radar reflectivity observations and their use in the data assimilation system. In heavy rain cases, which are the most critical for the nowcasting application, attenuation of the radar beam by hydrometeors can lead to severe underestimation of precipitation rates. A method to constrain the attenuation correction has been developed at the University of Reading and the Met Office is collabarating with University of Reading partners to implement the technique.

This paper will present a summary of the work completed, results from the first assimilation trials, and future plans.