New configuration for the modernized Canadian weather-radar network

Environment and Climate Change Canada (EECC) is undertaking a program to renew its radar network, including changes to hardware, data collection and data processing. Up until at least 2023, ECCC expects to replace a minimum of 20 operational C-band Doppler weather radars with new S-band polarimetric radars. There are options that, if invoked, would see replacement of additional radars and eventually the whole network would operate at S band. An implication of this process is that we will have a mixed-frequency network for several years. During this time, the weather-radar network will place increased emphasis on quantitative applications like numerical weather prediction and hydrology while continuing to support existing applications. We are tasked with formulating a new radar configuration to meet the demands of radar renewal. The boundary conditions for this are challenging, as the organization will be acquainting itself with a new vendor (SELEX), a new wavelength (S band), and a new transmitter type (klystron) while maintaining a level of consistency with the old magnetron C-band radars during the period of transition.

The existing 10-minute scan strategy was formulated for use with a signal processor that was decommissioned over 20 years ago. It consists of a top-down 24-sweep volume containing single-PRF uncorrected reflectivity collected over five minutes. Maximum range is 256 km; the lowest sweep (normally 0.3°) is configured to be lower (0.1°) during winter. Four dedicated Doppler sweeps follow in ascending order, with 112.5-120 km maximum ranges. The priorities on the non-Doppler volume are to address range/detectability and vertical structure for nowcasting applications, such as summer thunderstorms and tornadoes, and shallow snow squalls during winter.

We propose a new configuration for the radars that is formulated to continue to meet existing requirements while accommodating new ones. Trade-offs abound, and there remain uncertainties that need to be clarified up until the first new S-band radar is installed and tested, which may necessitate modifications to our proposal and further testing/validation. The new configuration has been developed through several iterations with ECCC’s Radar User Group (RUG), comprised of operational forecasters, research scientists, application developers, and operations personnel.

The configuration proposal is called PVOL6, which stands for “six-minute polar volume”. It is a top-down volume containing 17 sweeps, and a lowest elevation angle of 0.4° and highest tilt of 24.4°. In terms of range, the maximum ranges are 125 km for the top six sweeps, 160 km for the following two sweeps, and 240 km for the lowest nine sweeps. The top ten sweeps use a single PRF, whereas the lowest seven are dual-PRF at 4:3 ratio. The priority for radial wind speeds in the low sweeps is an unambiguous velocity interval of ±48 m/s. In order to achieve acceptable data quality when acquiring dual-polarization moments, the antenna speed is 11°/s for the lowest seven sweeps, higher when acquiring single-PRF data. For these lower sweeps, this gives us sample sizes of 20 and 27 at 450 and 600 Hz PRFs, respectively, with an azimuthal resolution of 0.5° for each PRF. With the higher single-PRF sweeps, sample sizes increase to 24 at 600 Hz PRF, 28 at 900 Hz PRF, and 37 at 1200 Hz PRF, with an azimuthal resolution of 1°. Conventional and dual-polarization moments are collected with all 17 sweeps. Both uncorrected and corrected reflectivities are included.

We had considered more complex scan strategies involving combinations of two-minute or five-minute sub-volumes, where these sub-volumes could be interleaved to achieve complete volumes. But the total acquisition time required to collect such a complete volume would have introduced uncertainties with the vertical alignment of the precipitation systems, which we deemed would have had a negative impact on data/product quality. So we instead focussed on the simpler sequential PVOL6, with the option to explore an extra low-level sweep at minute three if this can be achieved without unacceptable compromise to the rest of the scan strategy.

As the name indicates, we propose that the PVOL6 data acquisition go from a ten to a six-minute cycle. We also recommend a single configuration as opposed to one each for warm and cold seasons. We recommend that this cycle be regular, thereby consistent. A bonus is that this timing matches the nominal six-minute acquisition cycle of NEXRAD. This is a baseline configuration that we expect will require regional and local tweaks for meaningful deployment at e.g. mountain and other sites where the baseline would not be the optimal solution.

We also propose a matching PVOL6 at C band for the existing radars. There are a few notable differences between PVOL6S and PVOL6C, not just when it comes to the range-Doppler dilemma at each wavelength. The same 17 sweeps in PVOL6S are also acquired with PVOL6C, except at C band they are all collected using a single PRF that is 1200 Hz for the top six sweeps, 900 Hz for the following two, and 600 Hz for the remaining nine sweeps. Then follow three dedicated Doppler sweeps at dual-PRF (750 and 1000 Hz) out to 150 km range. The advantages of this approach are that we harmonize configurations and data acquisition at S and C bands to the same six-minute cycle, while raising the quality of data from the existing C-band radars to support both old and new applications. An impact analysis on mechanical stress and wear to the C-band antenna systems will be needed to determine if this proposed harmonization may be realized.