Science Deliverables to Canadian Weather Radar Network Renewal

From 2017 to 2023, 33 new weather radars have replaced the instruments in the Canadian weather radar network. This is a generational change in several ways. It is a change of wavelength from C to S band. It is a change of transmitter type, from magnetron to klystron. The change in hardware solution presents new challenges to the organization insofar as shifting from an in-house design at C band, based on acquired industrial components, to directly relying on an external vendor for complete commercial off-the-shelf radars and support. Radar technician training is a priority to establish qualified and quality personnel to maintain the equipment and ensure continued operation to meet data quality and availability service levels. The new solution involves the introduction of operational polarimetry network wide. There are significant changes in the way Canadian weather radar data are acquired, quality controlled, processed downstream, represented (formatted), and made available. Last but certainly not least, the overall change is so large that the people involved will most likely only experience it once in their careers.

The Canadian Weather Radar Replacement Program (CWRRP) has successfully replaced the radar instruments in the national weather radar network. 29 of the radars are replacements at existing sites. Four completely new sites have been added: in the Montréal area, a new radar at Blainville, Québec, replaces the McGill University S-band radar; in northern Alberta near Fort McMurray, expanding radar network coverage; at Halfmoon Peak, northwest of Vancouver, British Columbia, replacing the Mt. Sicker radar on Vancouver Island; and at an offline site at the Centre for Atmospheric Research Experiments facility north of Toronto, Ontario, that is used for training and testing.

The operational scan strategy contains 17 sweeps in top-down order with a regular update cycle of six minutes. The top 11 sweeps, starting at 24.4°, have 1° x 500 m bins and single PRF. The lower six sweeps, ending at 0.4°, have 0.5° x 500 m bins and 4:3 dual PRF. Ten moments are collected with all sweeps, including Doppler and dual-polarization moments; there are no separate sweeps optimized for reflectivity or radial winds. Maximum range of the lower sweeps has been 240 km, that is now being extended to 330 km. There is a variant of this scan strategy for use at mountain sites, where a priority is coverage in nearby valleys with population centres. Since the deployment of the first new S-band radar in 2017, several changes to signal processing have been introduced, all of which have had net positive impacts to data quality, e.g. minor changes like range sampling and larger changes like introducing random phase. A second signal processor, used passively, has recently been introduced at King City, Ontario, that allows us to explore additional signal processing optimizations, either in real time or offline, and output moment data for downstream processing. Annual signal processing workshops with the vendor have helped improve ECCC’s understanding of system capabilities and our ability to exploit them.

Calibration monitoring is currently performed using several methods. Returns from light rain in operational moment data are used opportunistically to monitor ZDR bias and stability over time. A recent enhancement to this approach is to add an analysis of returns from snow, to improve the utility of this approach in Canadian conditions. Another opportunistic use of moment data is the exploitation of solar hits to monitor power levels and ZDR bias of the receive chain, along with antenna pointing accuracy. Inter-radar matching compares data from radars whose coverage areas overlap sufficiently. Monitoring of power calibration stability is done by analyzing normalized returns from select clutter targets. These methods can be automated, and their outputs used to populate an operational network monitoring console.

Analysis of the quality of radar sites is performed offline using monthly statistics that reveal the presence and impacts of topography, vegetation, structures, and infrastructure. In Canada, the radio frequency interference (RFI) landscape contains more negative impacts at S band than at C band. Analysis of the impacts of wind farms is being updated with the transition to S band. Adaptation of existing legacy products for use with the new S-band data has been combined with the introduction of new dual-polarization based products: particle classification and quantitative precipitation estimation, along with improved dual-polarization based quality controls. Operational production and data management have been rationalized and centralized. Canadian S-band data, in the form of polar sweeps and volumes, are available in ODIM_H5 format while representation is being prepared to use the new WMO standard FM301/CfRadial2.

Radar data assimilation of reflectivity composites is running operationally, leveraging dual-polarization based quality control and a monitoring framework that uses precipitation occurrence observations from METARs to objectively assess the impacts of quality control on processed data. Assimilating these data using the latent heat nudging technique and the 2.5 km High Resolution Deterministic Prediction System gives improved precipitation forecast skill at nowcasting time scales, while improving the model state variables out to 48 hours lead time. Research is underway on assimilating radial wind velocity data.

With CWRRP ending this year, the larger scope of Canadian weather radar network renewal is expected to continue in the coming years. Improved basic data acquisition, quality assurance, quality control, product generation, and data availability remain ongoing priorities to be addressed collectively across ECCC.