Evaluation of polarimetric Quantitative Precipitation Estimation Products obtained with the French operational X-, C- and S-band polarimetric radars

The investment of Météo France on operational radar polarimetry started with the installation, in 2004, of the first polarimetric radar of the network in Trappes, near Paris. Since then, a significant amount of work has been performed and the positive results obtained have led to the extension of dual-polarization to other radars of the network, which now counts 15 operational polarimetric radars at X, C and S-band (11 at C-band, 3 at S-band and 2 at X-band) out of a total of 33 (including 7 radars overseas). A first dual-polarization processing chain (called “V1”), correcting for precipitation-induced path integrated attenuation (using the differential phase) and for clear air echoes, has been introduced operationally on all polarimetric radars at the beginning of 2012. Clear benefits have been demonstrated on intense precipitation (significant reduction of the negative bias).

In parallel, several studies have been carried out both off-line and in real time to further improve the Quantitative Precipitation Estimation (QPE) and to design and implement a new hydrometeor classification algorithm, still with the constraint to be applicable to X, C and S-band radars. These studies resulted in a second version of polarimetric processing chain (called “V2”).

The main innovation of the second version of polarimetric processing chain lies in the use of an hybrid Z-K_DP estimator in rain and hail, where a standard Z-R relationship (Marshall-Palmer) is used at low K_DP (typically 1°km^-1) and a R-K_DP relationship is used at higher K_DP. K_DP is estimated through a linear regression performed over the range profile of Φ_DP, which is filtered using a running 6 km median filter in a previous step. Moreover, the new dual-polarization processing chain uses a fuzzy logic hydrometeor classification algorithm that takes into account data-based membership functions, measurement conditions and temperature information. The formulation of the algorithm is unique for X, C, S-band radars and employs wavelength-adapted bivariate membership functions for (Z_H, Z_DR), (Z_H, K_DP) and (Z_H, ρ_HV) established using real data collected by the French polarimetric radars and T-matrix simulations. As for V1, work had to be done to properly interface the second version of polarimetric processing chain (V2) with the many modules composing the complex QPE algorithm (ground-clutter identification, partial beam blocking correction, vertical profile of reflectivity correction, estimation of a quality index, estimation and application of a radar – rain gauge calibration factor, …). Finally, what-if questions had to be addressed, such as : what if one TR tube falls down and the radar has to be run is single-polarization only ?

The evaluation was done at X, C and S-band by comparing hourly QPE obtained with V1 and V2 against rain gauges on a large number different [radar;event] couples. The evaluation shows a clear improvement of the QPE produced by V2 compared to V1. The level of improvement is more important at X-band than at C-band and more important at C-band than at S-band.