A machine learning approach for classifying bird and insect radar echoes with S-band Polarimetric Weather Radar

The S-bandWSR-88D weather radar is sensitive enough to observe biological scatterers like birds and insects. However, their non-spherical shapes and frequent collocation in the radar resolution volume create challenges in identifying their echoes. We propose a method of extracting bird (or insect) features by coherently averaging dual polarization measurements from multiple radar scans, containing bird (insect) migration. Additional features are also computed to capture aspect and range dependence, and the variation of these echoes over local regions. Next, ridge classifier and decision tree machine learning algorithms are trained, first only with the averaged dual pol inputs and then different combinations of the remaining features are added. The performance of all models for both methods, are analyzed using metrics computed from the test data. Further studies on different patterns of birds/insects, including roosting birds, bird migration and insect migration cases, are used to further investigate the generality of our models. Overall, the ridge classifier using only dual polarization variables was found to perform consistently well across all these tests. Our recommendation is that this classifier can be used operationally on the US Next-Generation Radars (NEXRAD), as a first step in classifying biological echoes. It would be used in conjunction with the existing Hydrometeor Classification Algorithm (HCA), where the HCA would first separate biological from non-biological echoes, then our algorithm would be applied to further separate biological echoes into birds and insects. To the best of our knowledge, this study is the first to train a machine learning classifier that is capable of detecting diverse patterns of bird and insect echoes, based on dual polarization variables at each range gate.

Insights into wind turbine reflectivity and radar cross-section (RCS) and their variability using X-band weather radar observations

The increasing need of renewable energy fosters the expansion of wind turbine sites for power production throughout Europe with manifold effects, both on the positive and negative side. The latter concerns, among others, radar observations in the proximity of wind turbine (WT) sites. With the aim of better understanding the effects of large, moving scatterers like wind turbines on radar returns, MeteoSwiss performed two dedicated measurement campaigns with a mobile X-band Doppler polarimetric weather radar (METEOR 50DX) in the northeastern part of Switzerland in March 2019 and March 2020. Based on the usage of an X-band radar system, the performed campaigns are up to now unique. The main goal was to quantify the effects of wind turbines on the observed radar moments, to retrieve the radar cross-section (RCS) of the turbines themselves and to investigate the conditions leading to the occurrence of the largest RCS. Dedicated scan strategies, consisting of PPI (plan position indicator), RHI (range–height indicator) and fixed-pointing modes, were defined and used for observing a wind park consisting of three large wind turbines. During both campaigns, measurements were taken in 24/7 operation. The highest measured maxima of horizontal reflectivity (ZH) and RCS reached 78.5 dBZ and 44.1 dBsm, respectively. A wind turbine orientation (yawing) stratified statistical analysis shows no clear correlation with the received maximum returns. However, the median values and 99th percentiles of ZH show different enhancements for specific relative orientations. Some of them remain still for Doppler-filtered data, supporting the importance of the moving parts of the wind turbine for the radar returns. Further, we show, based on investigating correlations and an OLS (ordinary least square) model analysis, that the fast-changing rotor blade angle (pitch) is a key parameter, which strongly contributes to the variability in the observed returns.

Detection of the melting level with polarimetric weather radar

Accurate estimation of the melting level (ML) is essential in radar rainfall estimation to mitigate the bright band enhancement, classify hydrometeors, correct for rain attenuation and calibrate radar measurements. This paper presents a novel and robust ML-detection algorithm based on either vertical profiles (VPs) or quasi-vertical profiles (QVPs) built from operational polarimetric weather radar scans. The algorithm depends only on data collected by the radar itself, and it is based on the combination of several polarimetric radar measurements to generate an enhanced profile with strong gradients related to the melting layer. The algorithm is applied to 1 year of rainfall events that occurred over southeast England, and the results were validated using radiosonde data. After evaluating all possible combinations of polarimetric radar measurements, the algorithm achieves the best ML detection when combining VPs of ZH, ρHV and the gradient of the velocity (gradV), whereas, for QVPs, combining profiles of ZH, ρHV and ZDR produces the best results, regardless of the type of rain event. The root mean square error in the ML detection compared to radiosonde data is ∼200 m when using VPs and ∼250 m when using QVPs.

Hydrometeor classification of quasi-vertical profiles of polarimetric radar measurements using a top-down iterative hierarchical clustering method

Correct, timely and meaningful interpretation of polarimetric weather radar observations requires an accurate understanding of hydrometeors and their associated microphysical processes along with well-developed techniques that automatize their recognition in both the spatial and temporal dimensions of the data. This study presents a novel technique for identifying different types of hydrometeors from quasi-vertical profiles (QVPs). In this new technique, the hydrometeor types are identified as clusters belonging to a hierarchical structure. The number of different hydrometeor types in the data is not predefined, and the method obtains the optimal number of clusters through a recursive process. The optimal clustering is then used to label the original data. Initial results using observations from the National Centre for Atmospheric Science (NCAS) X-band dual-polarization Doppler weather radar (NXPol) show that the technique provides stable and consistent results. Comparison with available airborne in situ measurements also indicates the value of this novel method for providing a physical delineation of radar observations. Although this demonstration uses NXPol data, the technique is generally applicable to similar multivariate data from other radar observations.

Advanced Real-Time Monitoring of Rainfall Using Commercial Satellite Broadcasting Service: A Case Study

Correct regulation of meteoric surface and subsurface flow waters is a fundamental goal for the sustainable development of the territories. A new system, aimed at real-time monitoring of the rainfall and of the cumulated rainfall, is introduced and discussed in the present paper. The system implements a Sensor Network based on the IoT paradigm and can cover safety-critical “hot spots” with a relatively small number of sensors, strategically placed, in areas not covered by traditional weather radars and rain gauges, and lowering the costs of deployment and maintenance with respects to these devices. A real application case, based on the implementation of the pilot plant at the Monte Scarpino landfill (Genoa, Italy), is presented and discussed. The system performances are assessed on the basis of comparisons with data provided by a polarimetric weather radar and by a traditional rain gauge.

Distinguishing between Warm and Stratiform Rain Using Polarimetric Radar Measurements

Modeled statistical differential reflectivity–reflectivity (i.e., ZDR–Ze) correspondences for no bright-band warm rain and stratiform bright-band rain are evaluated using measurements from an operational polarimetric weather radar and independent information about rain types from a vertically pointing profiler. It is shown that these relations generally fit observational data satisfactorily. Due to a relative abundance of smaller drops, ZDR values for warm rain are, on average, smaller than those for stratiform rain of the same reflectivity by a factor of about two (in the logarithmic scale). A ZDR–Ze relation, representing a mean of such relations for warm and stratiform rains, can be utilized to distinguish between warm and stratiform rain types using polarimetric radar measurements. When a mean offset of observational ZDR data is accounted for and reflectivities are greater than 16 dBZ, about 70% of stratiform rains and approximately similar amounts of warm rains are classified correctly using the mean ZDR–Ze relation when applied to averaged data. Since rain rate estimators for warm rain are quite different from other common rain types, identifying and treating warm rain as a separate precipitation category can lead to better quantitative precipitation estimations.

Multisensor Characterization of the Incandescent Jet Region of Lava Fountain-Fed Tephra Plumes

Explosive basaltic eruptions eject a great amount of pyroclastic material into the atmosphere, forming columns rising to several kilometers above the eruptive vent and causing significant disruption to both proximal and distal communities. Here, we analyze data, collected by an X-band polarimetric weather radar and an L-band Doppler fixed-pointing radar, as well as by a thermal infrared (TIR) camera, in relation to lava fountain-fed tephra plumes at the Etna volcano in Italy. We clearly identify a jet, mainly composed of lapilli and bombs mixed with hot gas in the first portion of these volcanic plumes and here called the incandescent jet region (IJR). At Etna and due to the TIR camera configuration, the IJR typically corresponds to the region that saturates thermal images. We find that the IJR is correlated to a unique signature in polarimetric radar data as it represents a zone with a relatively high reflectivity and a low copolar correlation coefficient. Analyzing five recent Etna eruptions occurring in 2013 and 2015, we propose a jet region radar retrieval algorithm (JR3A), based on a decision-tree combining polarimetric X-band observables with L-band radar constraints, aiming at the IJR height detection during the explosive eruptions. The height of the IJR does not exactly correspond to the height of the lava fountain due to a different altitude, potentially reached by lapilli and blocks detected by the X-band weather radar. Nonetheless, it can be used as a proxy of the lava fountain height in order to obtain a first approximation of the exit velocity of the mixture and, therefore, of the mass eruption rate. The comparisons between the JR3A estimates of IJR heights with the corresponding values recovered from TIR imagery, show a fairly good agreement with differences of less than 20% in clear air conditions, whereas the difference between JR3A estimates of IJR height values and those derived from L-band radar data only are greater than 40%. The advantage of using an X-band polarimetric weather radar in an early warning system is that it provides information in all weather conditions. As a matter of fact, we show that JR3A retrievals can also be obtained in cloudy conditions when the TIR camera data cannot be processed.