The New French Operational Polarimetric Radar Rainfall Product Jordi Figueras i Ventura, Fadela Kabeche, Béatrice Fradon, Abdel-Amin Boumahmoud, Pierre Tabary Météo France, 42 Av Coriolis, 31057 Toulouse CEDEX, France, jordi.figueras@meteo.fr (Dated: 30 May 2012) Jordi Figueras i Ventura 1. Introduction Météo France installed its first polarimetric radar in Trappes, near Paris, in 2004. Currently 11 out of 24 radars of the Metropolitan France weather radar network are polarimetric, 10 at C-band and 1 at S-band. It is expected that by 2020 the entire radar network will be polarimetric. In addition, a gap filling network of X-band polarimetric radars is being deployed in the Alps, complementing a pre-existing polarimetric X-band radar owned by CNRS and operated by NOVIMET. Since 2004 continuous work has been performed to implement efficient procedures to monitor data quality (Gourley et al., 2006), correct the precipitation-induced attenuation (Gourley et al. 2007b), identify the bright band (Tabary et al. 2006), classify the hydrometeors (Gourley et al. 2007a) and assess the impact of ground-clutter contamination on the measurements (Friedrich et al. 2009). These efforts led to a first version of a polarimetric processing chain (see Figueras i Ventura et al (2012) for a detailed description). The information provided by the polarimetric chain is exploited for the first time at Météo France for quantitative precipitation estimation (QPE). The polarimetric chain has been fully operational in all Météo France polarimetric radars since February 2012. Currently under test is a new version of the precipitation estimation algorithm which further exploits the possibilities offered by polarimetry. This paper describes the main differences between the old and the new polarimetric radar rainfall product (section 2), it comments on the evaluation methodology (section 3) and discusses the main results obtained by each algorithm at the 3 frequency lengths operated by Météo France (section 4). Conclusions are drawn in section 5. 2. The new rainfall product The flow diagram of the old and new radar rainfall processing chains are shown in Fig. 1a) and Fig 1b) respectively. The conventional processing is described in detail in Tabary (2007). It has as input the reflectivity in 1 km 2 resolution Cartesian coordinates. It performs a ground clutter (GC) identification, corrects for partial beam blocking (PBB), estimates and corrects the vertical profile of reflectivity (VPR), corrects for the gas attenuation, syncronizes the multi-tilt scans to the end of the 5- min cycle using the advection field, performs the weighted linear combination of the synchronized multi-tilt scans, oversamples the composite to 1-min resolution and adds the samples to obtain the 5-min rainfall accumulation. An adjustment coefficient based on the rain gauge information is applied to the entire image (Tabary et al. 2011). The weighted linear combination is performed based on a quality index (QI) which is processed in parallel to the reflectivity. The QI considers the effects of PBB and the height of the measurement above the ground. Pixels identified as ground clutter with reflectivity above 8 dbz are assigned a QI 0 and are omitted from further analysis. The first polarimetric version simply adds two new modules to the conventional radar rainfall product: a correction for precipitation induced attenuation by using the differential phase and the filtration of clear air (CA) echoes according to the output of a basic polarimetric echo classification. a) b) Fig. 1Flow diagram of the processing chain: a) old chain b) new chain. Polarimetric modules are in grey italic
The new polarimetric radar product aims at improving the performance of the previous one in three major aspects: better echo classification, better precipitation estimation and a lower detectability threshold. It also paves the way towards higher spatial resolution products by performing most of the data processing in high-resolution (0.5 x240m) polar coordinates. In the new product, the echo classification is fully performed by the polarimetric chain. The necessary corrections on the reflectivity at each individual scan (partial beam blocking (PBB) correction, gas and precipitation induced attenuation corrections, noise filtering, etc.) are also performed by the polarimetric chain with the aim of having the best quality polarimetric variables entering into the hydrometeor classification. The rainfall rate is estimated differently according to the echo type. Solid precipitation (snow, ice, etc.) is estimated using a Z-R relationship where the reflectivity has been corrected for attenuation. In areas where polarimetry was not available (e.g. low SNR), reflectivity is also converted into rainfall rate using a Z-R relationship. For the quantification of rain a hybrid estimator is used. If K dp is above a certain threshold a K dp -R relationship is applied, otherwise a Z-R relationship is used instead. Notice that Z dr is not used in the QPE because its current stability and precision is considered insufficient (See Figueras i Ventura et al, 2012). The threshold is based on the K dp value because it is insensitive to attenuation, calibration errors and PBB. At the moment the Z-R relationship used for all the estimators is the Marshall-Palmer. Below the threshold level, K dp is considered to be too noisy to be used. The K dp -R relationship used at S and C bands is the Beard and Chuang and it is used for K dp above 1 km -1. Studies at X-band by Kabeche et al. (2012) found that the Brandes K dp -R relationship is more suitable at such frequency and that above 0.5 km -1 the K dp is sufficiently clean to be usable. The rainfall rate output of the polarimetric chain is converted into 1 km 2 Cartesian coordinates using the nearest neighbour approach and back to reflectivity using the inverse Marshall-Palmer Z-R relationship. This artificial reflectivity is the input to the modules of the legacy processing chain. When computing the QI, the new product also takes into account the precipitation-induced path integrated attenuation (PIA). Moreover, pixels where precipitation has been estimated using K dp are considered not sensitive to PBB or PIA. 3. Data Selection and Evaluation Methodology The three radar rainfall products: the single polarization (hereafter CONV), the first version of the polarimetric chain (hereafter DBP1) and the second version of the polarimetric chain (hereafter DBP2) have been evaluated off-line using several precipitation episodes. The final Météo France product is adjusted using rain gauges but in order to strictly evaluate the performance of the processing chain data before the rain gauge adjustment (RG adj) has also been analysed. The 5-min rainfall accumulation estimated by the radar is hourly accumulated using a very strict criteria whereby if one of the 12 pixels in the time series is not valid within the hour the hourly accumulation is discarded. Unlike the other products, DBP2 invalidates pixels containing ground clutter with very low reflectivity. For this reason a slightly different criterion has been taken to validate the hourly accumulation. In that case, if a maximum of 2 pixels are invalid but they have a reflectivity threshold below 5 dbz the invalid pixels are set to have a 0 mm rainfall accumulation and the hourly accumulation is considered valid. The hourly rainfall accumulations are compared against hourly rain gauges. The Météo France rain gauge network consists of tipping bucket gauges with a bucket resolution of 0.2 mm, i.e. the minimum hourly rainfall accumulation that can be measured is 0.2 mm. All rain gauge data are routinely quality-controlled as described in Figueras i Ventura et al. 2012. The radar-rain gauge comparison is done by matching each rain gauge with the corresponding radar pixel. The quality of the algorithms is evaluated using the normalized bias (NB) between the rain gauge and the radar rainfall accumulation, the correlation (corr), the root mean square error (RMS) and the Nash-Sutcliffe model efficiency coefficient (Nash). The present study is focused on a warm period, where the benefits of the polarimetric QPE are more relevant. The selection of events is performed objectively using three criteria. Firstly, the daily average ground temperature close to the radar must be high enough so that the radar beam was below the freezing level height at 60 km range. A standard atmosphere temperature decrease of 6 /km is used to estimate the altitude of the freezing level height. Secondly, a significant amount of rain must be present in the vicinity of the radar. The amount of rain is determined by calculating the average daily rainfall accumulation of all the rain gauges within a 60 km radius area. Events with an average rainfall accumulation higher than 5 mm are considered. Finally, all the selected events should have been observed using the same exploitation mode. For S and C band, data from the year 2010 is analysed. At C band four different radars out of the nine present that year from different regions and with different characteristics are used. They are considered to be representative of the entire radar network. For the evaluation at S band events observed by the Nimes radar, the sole S band radar in the network in 2010, are analysed. The X band is analysed using data from Maurel, which has a raw data processor similar to the one used by the radars in the operational network. Data from Maurel is available since the second half of 2011. 4. Results The results are stratified according to three thresholds on the rain-gauge hourly accumulations: >0.2 mm (all rainfall accumulations, AR), >1 mm (moderate and high hourly accumulations, MR) and >5mm (intense hourly accumulations, IR). The results are obtained in the area 60 km around the radar, which have the best hydrological visibility. 1) Results at S-band The global results obtained by each rainfall product are represented in Fig. 2. As it can be seen in Fig 2.a) the CONV
product slightly underestimates precipitation (NB=-0.13 for AR). The underestimation is more pronounced at IR (NB=-0.27). As it is shown in Fig. 2.b), the RG adj is able to globally correct for such underestimation but IR remain underestimated (NB=-0.13). The use of the DBP1, which basically corrects for precipitation-induced attenuation results in an improvement of the NB and corr as shown in Fig. 2.c). Such improvement is even more relevant at IR (NB reduced from 0.27 to 0.18) as shown in Fig. 2.d). The RG adj is less dramatically beneficial for the DBP1 and the final results are comparable with those obtained by the CONV product. Nevertheless minor improvements respect to CONVadj should be signaled. As represented in Fig. 2.e), DBP2 obtains the best score in all categories. Particularly remarkable is the NB at IR which is reduced by 1/3 compared to that of CONV. The application of the RG adj, though, does not significantly improved the results, as shown in Fig. 2.f). Fig. 2 Results at S-band 2) Results at C-band The global results at C-band are represented in Fig. 3. As it can be seen in Fig. 3.a) the radar largely underestimates precipitation (-0.32 for AR and up to 0.47 for IR). Corr is rather poor as well for IR. This large underestimation can be attributed to precipitation-induced and radome-induced attenuation but miscalibration of the reflectivity should not be discarded. The RG adj significantly reduces the NB (down to 0.10 for AR and 0.28 for IR) and it has a positive impact also on the corr (and increase of 0.04 in the score) (see Fig. 3.b). The positive impact of the attenuation correction of DBP1 is readily visible in Fig. 3.c). Bias for AR is reduced to 0.25 but the most positive impact is, as expected in IR, where both the NB and the corr are dramatically improved (NB down to 0.34 and correlation up to 0.70). With the RG adj a similar score in terms of bias to CONVAdj is obtained but the correlation is further improved (See Fig. 3.d). Again, the best score is obtained by DBP2, with the most remarkable improvement on the IR (NB=-0.19 and corr up to 0.79, see Fig. 3.e). The improvement due to RG adj is less significant, particularly that of IR (see Fig. 3.f).
Fig. 3 Results at C-band 3) Results at X band The global results obtained at X-band are represented in Fig. 4. The radar of Maurel is situated in a mountainous environment. The estimation of surface precipitation in such environment in rendered more difficult due to PBB and the fact that measurements are performed high above the ground. In addition, rain gauges tend to be placed down in the valleys, orographic enhancement effects not measurable by the radar should not be discarded in such conditions. This is the likely cause for the extremely low correlation exhibited by the CONV method shown in Fig. 4.a). Part of the very large negative bias of the measurement can be attributed to radar miscalibration. In a separate study, a radar-to-radar comparison in collocated pixels was performed between the radar at Mont Maurel and that at Mont-Vial (Frasier et al. 2012). The study concluded that the Mont Maurel radar was underestimating reflectivity by 2 db respect to the Mont-Vial radar. Another cause for the underestimation of precipitation is wet-radome attenuation, the effect of which is more significant at X-band that at a lower frequency. Again, the RG adj significantly reduces the NB although it fails to improve the corr (see Fig. 4.b). As shown in Fig. 4.c) the attenuation correction of DBP1 results in a reduction of roughly 0.2 in NB and an improve of 0.1 in the corr. The use of the RG adj improves the normalized bias by roughly 0.2 respect to CONVAdj (see Fig. 4.d). The best results by far are obtained by DBP2 (see Fig. 4.e). The use of K dp, which is insensitive to wet radome attenuation and partial beam blocking, largely reduces the NB and it significantly increases the corr. The RG adj further reduces the bias (see Fig. 4.f). The bias obtained by DBP2Adj is comparable to that obtained at the other frequency bands.
Fig. 4 Results at X-band 5. Conclusion This paper has presented the new operational polarimetric radar rainfall product, which is meant to be tested in shadow mode alongside the existing product by June 2012. Currently, there is an operational product with two operational modes: one used by polarimetric radars (half of the network at the moment) and one used by conventional ones. The new product has several innovative features. Among others: it makes a wider use of polarimetry for echo classification and precipitation estimation and it attempts to improve the detectability level by providing a better noise level estimation and by considering the impact of weaker ground clutter echoes. The new product is objectively evaluated against the previous product (in its polarimetric and non-polarimetric modes) by comparing hourly rainfall accumulations estimated using radar and hourly rain gauge accumulations. The results demonstrate the benefits of polarimetry. The first version of the polarimetric radar rainfall product, which essentially consists in filtering out clear air echoes and correcting for precipitation-induced attenuation, already shows a remarkable improvement of the scores, particularly for intense precipitation. The second version of the polarimetric radar rainfall product, which makes use of K dp to estimate the rainfall rate further improves the results. Nevertheless the results show some margin for improvement. It is evident that C and X band radars in the French network tend to underestimate reflectivity. This may be partially due to wet radome attenuation (more severe at larger frequencies) but miss-calibration of the radar constant should not be excluded. Monitoring using the self-consistency method and intercomparison between collocated radar gates may improve the reflectivity calibration. Although version two of the polarimetric radar rainfall product presents a better bias for intensive precipitations, it still largely underestimates them. Uncertainties in the coefficients of the R-K dp relationship may be partially blamed for that but it is hypothesized that most of the underestimation is due to the smoothing of small but intensive convective cells produced by the spatial filtering of the differential phase. An adaptive filter length would certainly improve the results.
The results of the new product on warm period precipitation are considered to be satisfactory. However, much more effort has to be placed on the processing of winter precipitation and hail cases. In this respect, the improved hydrometeor classification achieved by the use of polarimetry will certainly help. Robust relationships between the liquid water content of the various solid hydrometeor types and the polarimetric variables have to be established. It should also be noticed that the relationship between rain and polarimetric variables is not univocal and the used relationships are representative only of the most common drop size distribution in the region. Consequently, large deviations are possible. The real time estimation of the DSD could be possible using the information provided by the differential reflectivity but so far issues such as its stability, the correction for attenuation and PBB, etc. have prevented its use in quantitative applications. Should the mentioned issues be tackled, an estimation of liquid water content on the ground based on the vertical profile of liquid water content could be provided, together with the minimum detectable liquid water content and the hydrometeor type on the ground. Regardless of all the envisioned improvements on the rainfall estimation by individual radars, the issue of the optimal blending of data from different radars to obtain a national composite has not been discussed in this article. An specific study on how to combine data from polarimetric and non-polarimetric radars and working at different frequency lengths is currently being conducted at Météo France. Acknowledgment The financial support for this study was provided by the European Union, the Provence-Alpes-Côte d Azur Region, and the French Ministry of Ecology, Energy, Sustainable Development and Sea through the RHYTMME project. References Figueras i Ventura J., Boumahmoud A.-A., Fradon B., Dupuy P., Tabary P., 2012: Long-term monitoring of French polarimetric radar data quality and evaluation of several polarimetric quantitative precipitation estimators in ideal conditions for operational implementation at C-band. Quart. Jour. of the Royal Meteo. Soc., Online early release Frasier S., Beck J., Kabeche F., Figueras i Ventura J., Al-Sakka H., Fradon B., Boumahmoud A.-A., Bousquet O., Tabary P., 2012: Assessment of reflectivity observations by a heterogeneous network of X- and K-band radars. In Proceedings of 7 th Conference on Radar in Meteorology and Hydrology, Toulouse, France, 24-29 June 2012 Friedrich K., Germann U., Tabary P., 2009: Influence of Ground Clutter Contamination on Polarimetric Radar Parameters. J. Atmos. Oceanic Technol., 26, 251-269 Kabeche F., Figueras i Ventura J., Fradon B., Boumahmoud A.-A., Frasier S., Tabary P., 2012: Design and test of an X-band optimal rain rate estimator in the frame of the RHYTMME Project. In Proceedings of 7 th Conference on Radar in Meteorology and Hydrology, Toulouse, France, 24-29 June 2012 Gourley J.J., Tabary P., Parent du Chatelet J., 2006: Data Quality of the Meteo-France C-band Polarimetric Radar. J. Atmos. Oceanic Technol., 23, 1340-1356 Gourley J.J., Tabary P., Parent du Chatelet J., 2007a: A Fuzzy Logic Algorithm for the Separation of Precipitating from Nonprecipitating Echoes Using Polarimetric Radar Observations. J. Atmos. Oceanic Technol., 24, 1439-1451 Gourley J.J., Tabary P., Parent du Chatelet J., 2007b: Empirical Estimation of Attenuation from Differerential Propagation Phase Measurements at C Band. J. Appl. Meteor. Climatol., 46, 306-317 Tabary P., Le Henaff G., Vulpiani G., Parent-du-Châtelet J., Gourley J.J., 2006: Melting layer characterization and identification with a C- band dual-polarization radar: A long-term analysis. In Proceedings of 4th European Conference on Radar in Meteorology and Hydrology, Barcelona, Spain, 18-22 Sept. 2006. Servei Meteorològic de Catalunya: Barcelona Tabary P., 2007: The New French Operational Radar Rainfall Product. Part I: Methodology. Wea. Forecasting., 22, 393-408 Tabary P., Boumahmoud A.-A., Andrieu H., Thompson R.J., Illingworth A.J., Le Bouar E., Testud J., 2011: Evaluation of two integrated polarimetric Quantitative Precipitation Estimation (QPE) algorithms at C-band. J. of Hydrology., 405, 248-260