Comparison of Single-PRT, Dual-PRF, and Triple-PRT Weather Radar Measurements

We have carried out weather radar measurements using three different pulsing schemes with the C-band dual-polarization pulsed Doppler radar equipped with an RVP900 signal processor located in the University of Helsinki Kumpula Campus. The schemes include standard single-PRT measurements, a dual-PRF setup, and a triple-PRT system consisting of repeated patterns of three unequal pulse intervals. The scans are run consecutively in a repeated sequence separated by an interval of a few minutes enabling direct comparison of the results. The single-PRT and dual-PRF analysis are performed using the original Vaisala software, while the triple-PRT analysis was done on a second computer using our WnD software.

The use of non-uniform pulsing intervals presents an elegant solution to the so-called Doppler dilemma which restricts the measurement of high-speed targets with a pulsed Doppler radar at large distances. We employ a triple-PRT setup capable of handling velocities up to 50 meters per second within a distance of 260 kilometers. Our choice is to send repeated patterns consisting of three pulses separated by intervals of 1750, 2000, and 2500 microseconds (7:8:10 ratio).

We present side by side comparisons of the ground clutter removal of the three systems. The single-PRT setup employs the widely used GMAP algorithm, while the dual-PRF scheme uses a standard notch filter. Our triple-PRT system removes ground clutter using an adaptive filtering technique based on dual-polarization data, including a reflectivity correction based on a Gaussian precipitation model. In addition to the basic single-polarization quantities, such as reflectivity, velocity, width and suppression rates, we also show dual-polarization products.

The data collection is gathered in March and June 2012, including various different weather conditions, such as strong winds, snowfall, rainfall, low-altitude melting layers, temperature inversion, anomalous propagation, and biological scatterers. The cases contain both strong and weak instances of ground clutter signals for several elevations of the radar beam.

The results show that the triple-PRT clutter filter performs equally well as the single-PRT method at low precipitation velocities. For speeds comparable to or exceeding the unambiguous velocity range of the uniform pulsing scheme, the triple-PRT setup performs significantly better in regard to, for example, the reflectivity estimates. However, the most significant improvement of the clutter removal is due to the adaptive application of the filter.