Improvements of the French operational triple-PRT Doppler scheme

In 2005, a triple-PRT (Pulse Repetition Time) Doppler was introduced in the French operational network with the objective to provide de-aliased radial velocities (up to 60 m s-1) up to long range (250 km) and the constraint to keep the mean PRT at the same initial level (1 / 333 Hz for C-band radars). The Cartesian 1 km² radial velocity PPIs produced with that Doppler scheme have a Dealiasing Success Rate (DSR) of about 90% in the mean but 4 years of operations have shown that the DSR can be much less (60%) in clear-air or convective situations (cases of low-SNR and / or large spectrum widths). Therefore, the recommendation to the users has been so far to apply a 5x5 km² running filter on the raw PPIs. To improve the quality of these raw radial velocity PPIs, different solutions are tested.

The first is the increase of the PRFs : several radars have been upgraded since 2005 and can now support higher PRFs. Simulations “à la Zrnic” and data acquisition experiments have thus been conducted to design new values for the three PRTs and assess objectively the improvement with respect to the initial scheme.

With a maximum PRF of 550 Hz supported by the most recent French C-band radars, the best couple of ratio defining the two other PRFs was found to be (6/7, 3/4). The associated mean PRT is (1 / 471 Hz) and the resulting extended Nyquist velocity is equal to 44 m.s-1. For an extended Nyquist velocity close to 60 m.s-1 (like the one of the current scheme), the best couple of ratios is (8/9, 4/5); it is associated with a mean PRT of 1 / 498 Hz. The new PRF triplets were tested on the Trappes (near Paris) C-band polarimetric radar. On a convective situation with a squall line, the candidate (6/7, 3/4) led to a significantly increased DSR (95 % versus 84 % with the former triplet). This kind of test is going to be carried out again in order to assess quantitatively the improvement due to the new triplets.

Another part of the study consists in testing several polar projections : 1°x1km, 0.5°x300 m and 0.5°x300m with some oversampling (each complex pulse-pair is projected on one gate but also on its 8 neighbours). In the first results, consistently with expectations the 1°x1km and oversampled 0.5°x300 m proved to be better than the 1 km2 Cartesian grid for radial velocities at distances exceeding 60 km.