This paper presents clutter detection and mitigation for polarimetric phased array weather radar measurements using machine learning. Three approaches of naive Bayes classifier (NBC), multilayer perceptron (MLP), and convolutional neural network (CNN) are used for clutter detection in the cylindrical polarimetric phased array radar measurements. Results show that CNN achieves the best performance in clutter detection, followed by MLP and NBC. This is because CNN utilizes spatial information of the input images, which has different features for clutter from that for weather. It is also shown that the combination of physics-based discriminants of power ratio and raw radar measurements is more effective in clutter detection than the direct use of raw radar measurements. In addition, CNN is employed for clutter mitigation and its performance is compared with the traditional speckle filter technique. It is demonstrated that CNN outperforms the speckle filter and incorporation of power ratio in the training process could further improve CNN’s performance in clutter mitigation.

This letter presents differences in spectrum width estimates between electronic scans and mechanical scans with a cylindrical polarimetric phased array radar (CPPAR). Theoretical formulation of spectrum width estimates considering the transverse wind, shear, and turbulence within the radar’s resolution volume is described. Cross correlation function between two resolution volumes is derived based on wave scattering by randomly distributed scatterers. Spectrum width estimates of ground clutter and weather scatterers with the CPPAR in electronic scans and mechanical scans are presented and compared. It is shown that instead of antenna rotation, the transverse wind, shear, and turbulence within the radar’s resolution volume, which were previously ignored for narrow beams, contribute the most to the differences in spectrum width estimates between mechanical scans and electronic scans for the CPPAR with a wide beam.

This article presents similarities and differences in clutter detection between electronic scans and mechanical scans with a cylindrical polarimetric-phased array radar (CPPAR). Theoretical explanations of clutter features in electronic scans that are different from those in mechanical scans are explored and verified by observations with the CPPAR. Clutter detection results with the CPPAR, based on the co polar correlation coefficient, dual-scan cross-correlation coefficient, power ratio, and their combinations in the electronic scan and mechanical scan modes are presented and compared.

This letter combines a time-domain modeling and simulation method for evaluating the impacts of system modules on polarimetric data quality of phased array weather radars with both theoretical analysis and actual measurements. In the presented phased array radar system simulator (PASIM), the distributed weather returns are modeled by randomly distributed scatterers, and Next-Generation Radar (NEXRAD) Level-II data or user-defined weather scenarios are utilized as weather truth fields. Based on a specially designed patch element, a dual-polarization phased array mobile demonstration system (Ten Panel Demonstrator or TPD) is simulated. In addition, the biases of differential reflectivity, copolar correlation coefficient, and differential phase along beam direction away from the broadside in principal plane and nonprincipal plane are used as data quality metrics. Moreover, theoretical analysis, system simulation, and actual TPD proof of concept measurements in a stratiform precipitation and a convective precipitation are presented, respectively, then similarities and discrepancies between simulations and measurements are compared and explained.

This letter presents initial weather measurements with a cylindrical polarimetric phased array radar (CPPAR) demonstrator developed at The University of Oklahoma. The overall system specifications, waveform design, beam pattern measurement, and beam-to-beam calibration of the CPPAR demonstrator are presented. The weather observations of convective precipitation are provided, employing a single-beam mechanical scan and commutating beam electronic scan. Measurement results from these two scan modes are compared, and the error statistics are derived and discussed. A new feature of the CPPAR commutating beam electronic scan in clutter detection is observed and explained.

In this study, as part of an effort to study snowfall characteristics and quantify winter precipitation in East China, we investigated the microphysical properties of snowfall, including size, shape, density, and terminal velocity using a 2-D video disdrometer (2-DVD) and a weighing precipitation gauge in Nanjing (NJ), East China during the winters of 2015-2019. We obtained larger snow density and terminal velocity values than those reported in the literature for this region. Higher snow density could account for higher snowflake terminal velocity, after removing the effects of observation altitude and surface temperature. We then fit the snow particle size distributions (PSDs) to the gamma model and explored the interrelationships among the model parameters and snowfall rate (SR). The relationship between radar reflectivity factor (Ze) and SR was derived based on snow PSD measurements and the snow density relation. Using this Ze-SR relationship, the estimated liquid-equivalent SRs are obtained from S-band NJ radar data collected during several snowfall events. Radar-inferred SRs showed reasonable agreement with those measured on the ground, with a mean absolute error of 16% for the collected snowfall events in NJ.

Our aim, in this paper, is to develop a clutter detection algorithm to provide more representative weather radar observations. The new discriminant function based on the phase fluctuation index (PFI) is introduced to achieve a better performance for clutter detection algorithms. Statistical properties of the PFI for pure weather and ground clutter are presented. A Bayesian classifier is used to make an optimal decision to detect clutter mixed with weather echoes. The performance improvements are demonstrated by applying the PFI detection algorithm to radar data collected by a WSR-88D polarimetric weather radar. Our proposed clutter detection algorithm is compared to several other detection algorithms and reveals the PFI algorithm yields the highest probability of detection.

Future weather radar systems will need to provide rapid updates within a flexible multifunctional overall radar network. This naturally leads to the use of electronically scanned phased array antennas. However, the traditional multifaced planar antenna approaches suffer from having radiation patterns that are variant in both beam shape and polarization as a function of electronic scan angle; even with practically challenging angle-dependent polarization correction, this places limitations on how accurately weather can be measured. A cylindrical array with commutated beams, on the other hand, can theoretically provide patterns that are invariant with respect to azimuth scanning with very pure polarizations. This paper summarizes recent measurements of the cylindrical polarimetric phased array radar demonstrator, a system designed to explore the benefits and limitations of a cylindrical array approach to these future weather radar applications.

A hybrid method of combining linear programming (LP) and physical constraints is developed to estimate specific differential phase (KDP) and to improve rain estimation. The hybrid KDPestimator and the existing estimators of LP, least squares fitting, and a self-consistent relation of polarimetric radar variables are evaluated and compared using simulated data. Simulation results indicate the new estimator's superiority, particularly in regions where backscattering phase (δhv) dominates. Furthermore, a quantitative comparison between auto-weather-station rain-gauge observations and KDP-based radar rain estimates for a Meiyu event also demonstrate the superiority of the hybrid KDPestimator over existing methods.

On phased-array radars, scanning is done by stepping the beam from one direction to the next direction and dwelling long enough at each direction to achieve acceptable errors of estimates. Combining data from the three directions is suggested to obtain superresolution similar to that available on the national network of weather radar (Weather Surveillance Radar-1988 Doppler or WSR-88D). Spectral analysis of such data is addressed, and it is demonstrated that the Doppler spectra of simply concatenated time series have very strong sidebands due to the discontinuity of the signals from the three beam positions. This artifact degrades the performance of the spectral clutter filters and other methods that rely on spectral processing to enhance the weather signal. Special adjustments of the signals at each range location before concatenating (splicing) are proposed to mitigate the effects of discontinuities in time and thus improve clutter filtering. The adjustment is such that the total information contained in the signal can be preserved in subsequent processing. Spectral quality of the concatenated signals is quantified via results from simulations. Samples of spectra obtained with the National Weather Radar Testbed are presented to substantiate the predictions. A ground clutter detector/filter accepted by the National Weather Service is applied to the conditioned time series data, and the ensuing fields of reflectivity factor and Doppler velocity are compared to the fields from which clutter had not been removed.

Accurate radar remote sensing requires a radar system with high cross-polarization isolation, highly matched dual-polarization patterns, and low sidelobes. A cylindrical polarimetric phased-array radar (CPPAR), which has polarization purity and scan-invariant beam properties, has recently been introduced to the weather and air surveillance communities. To achieve low sidelobes and matched beams, pattern synthesis using an optimization method is presented. Results reported herein support the idea that CPPARs can be designed and implemented for accurate weather measurements. Furthermore, some uncertainty analysis is performed to show the effects of the amplitude and phase errors on the radiation patterns of the PAR.

Planar or cylindrical phased arrays are two candidate antennas for future polarimetric weather radar. These two candidate antennas have distinctly different attributes when used to make quantitative measurements of the polarimetric properties of precipitation. Of critical concern is meeting the required polarimetric performance for all directions of the electronically steered beam. The copolar and cross-polar radiation patterns and polarimetric parameter estimation performances of these two phased array antennas are studied and compared with that obtained using a dual-polarized parabolic reflector antenna. Results obtained from simulation show that the planar polarimetric phased array radar has unacceptable polarimetric parameter biases that require beam to beam correction, whereas biases obtained with the cylindrical polarimetric phased array radar are much lower and comparable to that obtained using the parabolic reflector antenna.

The scan-to-scan correlation method to discriminate weather signals from ground clutter, described in this letter, takes advantage of the fact that the correlation time of radar echoes from hydrometeors is typically much shorter than that from ground objects. In this letter, the scan-to-scan correlation method is applied to data from the WSR-88D, and its results are compared with those produced by the WSR-88D's ground clutter detector. A subjective comparison with an operational clutter detection algorithm used on the network of weather radars shows that the scan-to-scan correlation method produces a similar clutter field but presents clutter locations with higher spatial resolution.

Polarimetric phased-array radar (PPAR) creates biases in observed polarimetric parameters when the beam is pointed off broadside. Thus, a bias correction matrix needs to be applied for each beam direction. A bias correction matrix is developed for array elements consisting of either waveguide apertures or patches. Correction matrices are given for both the Alternate Transmission and Simultaneous Reception mode and the Simultaneous Transmission and Simultaneous Reception mode. The biases of polarimetric parameters measured with a PPAR without the application of a correction matrix are presented.

Considering that the statistics of the phase and the power of weather signals in the spectral domain are different from those statistics for echoes from stationary objects, a spectrum clutter identification (SCI) algorithm has been developed to detect ground clutter using single polarization radars, but SCI can be extended for dual-pol radars. SCI examines both the power and phase in the spectral domain and uses a simple Bayesian classifier to combine four discriminants: spectral power distribution, spectral phase fluctuations, spatial texture of echo power, and spatial texture of spectrum width to make decisions as to the presence of clutter that can corrupt meteorological measurements. This work is focused on detecting ground clutter mixed with weather signals, even if the clutter power to signal power ratio is low. The performance of the SCI algorithm is shown by applying it to radar data collected by University of Oklahoma-Polarimetric Radar for Innovation in Meteorology and Engineering.

Based on today's most widely used surface scattering model, the advanced integral equation model (AIEM), this study proposes a novel soil moisture inversion model that estimates bare surface soil moisture using double-incidence angle and dual-polarized L-band radar data. Compared with previous studies at L-/C-band, the proposed method provides the estimation of soil moisture without referring to the measured soil roughness and eliminates the requirement of an initial dry season condition. The root-mean-square error (rmse) of volumetric soil moisture varies from 0.8% to 3.2% at different incidence-angle combinations validated by simulated solving and from 4.0% to 7.9% by field measurements when the paired incidence angles are not both large. In case the paired angels are both large, not particularly suitable for soil moisture estimation, the rmse increases to 10.3%. Therefore, this method is applicable to bare surface soil moisture retrieval when at least one of the incidence angles is not large.

This paper introduces the Spectrum-Time Estimation and Processing (STEP) algorithm developed in the Atmospheric Radar Research Center (ARRC) at the University of Oklahoma (OU). The STEP processing framework integrates three novel algorithms recently developed in ARRC: spectrum clutter identification, bi-Gaussian clutter filtering, and multi-lag moment estimation. The three modules of STEP algorithm fulfill three functions: clutter identification, clutter filtering and noise reduction, respectively. The performance of STEP has been evaluated using simulated data as well as real data collected by the C-band polarimetric research radar OU-Polarimetric Radar for Innovations in Meteorology and Engineering. Results show that STEP algorithm can effectively improve quality of polarimetric weather data in the presence of ground clutter and noise.

This study presents a new identification and mitigation scheme of second trip contamination for pulsed Doppler polarimetric weather radars with the ability of random phase coding. This scheme can be easily implemented in a magnetron radar without any hardware changes. For relatively weak contamination, identification and mitigation are based on a multilag processing method, which uses multiple lags of both the auto- and cross-correlation functions to estimate radar moments. For relatively strong contamination, instantaneous phase variations of horizontal and vertical polarization channels are combined into a simple fuzzy-logic scheme to complete the identification. Data from the C-band OU-PRIME radar are used to demonstrate the effectiveness of the proposed scheme for identification and mitigation of second-trip echoes.

Cross-polar measurements made with an agile-beam phased-array weather radar are compared with theory. The intensity of cross-polar fields places conditions on the accuracy of meteorological measurements. Results reported herein support the hypothesis that polarimetric phased-array radar for weather observations can be designed to allow the use of the polarimetric data acquisition mode being implemented by the National Weather Service on upgraded WSR-88Ds which use parabolic reflector antennas.

Incorporating dual-polarized operation and microphysics-based processing is becoming a challenge to future scientific and commercial airborne weather radars. This paper introduces a Monte Carlo simulation-based approach to address the theoretical basis and uncertainties of hydrometeor scattering along with sensor platform properties. Detailed characterizations of mixed-phase aviation hydrometeor hazards (rain, snow, hail, and mixtures) and the impact of melting on polarimetric radar signature at X-band frequency are discussed. A “single resolution cell” Monte Carlo dual-polarization variable simulation technique is described and then applied in different radar scanning scenarios based on numeric weather prediction model output weather fields. The produced dual-polarization signatures of an X-band array radar for different scan scenarios are analyzed.

This paper discusses ways to avoid and/or mitigate biases in polarimetric variables inherent to agile-beam planar phased-array radars. Two bias-avoiding schemes produce unbiased estimates of the polarimetric backscattering covariance matrix which are then combined into bias-free polarimetric variables. One concerns full polarimetric measurements and calls for adjusting the amplitudes and phases of the array elements so that the transmitted field equals that generated by a mechanically steered polarimetric weather radar antenna; this is followed by an additional adjustment of the received fields. The second scheme is also applicable to full polarimetric measurements but involves adjustments only of the received fields. Crucial to both schemes is decoupling of the Doppler effects from the terms of the covariance matrix. It is a significant part of the bias issue that had not been previously addressed. A scheme to reduce bias applicable to nondepolarizing media (i.e., diagonal backscattering matrix) is also addressed; it calls for multiplication of the fields received by each dipole as opposed to a combination of multiplication and addition required for full correction. The schemes are applied to the alternate transmission and simultaneous reception polarimetric mode and the simultaneous transmission and simultaneous reception mode.

Phased array radars (PARs) are attractive in weather surveillance primarily because of their capability to electronically steer. When combined with the recently developed beam multiplexing (BMX) technique, these radars can obtain very rapid update scans that are useful in monitoring severe weather. A consequence is that the small number of contiguous samples of the time series obtained can be a challenge for temporal/spectral filters used for clutter mitigation. As a result, the accurate extraction of weather signals can become the limiting performance barrier for PARs that employ BMX in clutter-dominated scattering fields. By exploiting the spatial correlation of the auxiliary channel signals, the effect of clutter contamination can be reduced in these conditions. In this paper, three spatial filtering techniques that used low-gain auxiliary receive channels are presented. The effect of clutter mitigation was studied using numerical simulations of a tornadic environment for changes in signal-to-noise ratio, clutter-to-signal ratio, number of time series samples, varying clutter spectral widths, and maximum weight constraints. Since such data are not currently available from a horizontally pointed phased array weather radar, experimental validation was applied to an existing data set from the turbulent eddy profiler, which is a vertically pointed PAR. Although preliminary, the results show promise for clutter mitigation with extremely short nonuniform sampling.

It is becoming widely accepted that radar polarimetry provides accurate and informative weather measurements, while phased-array technology can shorten data updating time. In this paper, a theory of phased array radar (PAR) polarimetry is developed to establish the relation between electric fields at the antenna of the PAR and the fields in a resolution volume filled with hydrometeors. It is shown that polarimetric measurements with an electronically steered beam can cause measurement biases that are comparable to or even larger than the intrinsic polarimetric characteristics of hydrometeors. However, these biases are correctable if the transmitted electric fields are known. A correction to the measured scattering matrix that removes biases in meteorological variables is derived. The challenges and opportunities for weather sensing with a polarimetric PAR are discussed.

It is becoming widely accepted that radar polarimetry provides accurate and informative weather measurements, while phased array technology can shorten data updating time. In this paper, a theory of phased array radar polarimetry is developed, and the relationship between the wave field at the radar antenna coordinate and that at the hydrometeors is established, along with the correction matrix to the scattering matrix. The challenges and opportunities for the weather sensing are discussed.

Characterization of rain microphysics requires information on raindrop size distributions (DSDs). DSD measurements and retrievals, however, contain errors. In this paper, data from side-by-side disdrometer comparisons are presented to provide information that is not possible from single disdrometer measurements alone, allowing error effects to be quantified and a rain DSD model for radar retrieval to be improved. We also propose methods to mitigate sampling errors by filtering with sorting and averaging and by fitting to a Gamma DSD model with truncation and extrapolation. The Constrained- Gamma DSD model has thus been refined for the Southern Great Plains region of the United States.

Presents a range interferometry technique to determine radial wind. The wind velocity is obtained by the ratio of range cross-correlation function at positive and negative lags. The feasibility of the method is studied through error analysis. The standard error of the estimated radial wind is derived and its sensitivity to range resolution, sample time and turbulence is analyzed.

The authors study the sampling effect on radar measurements of inhomogeneous media and the resultant rain estimation. A two-level drop size distribution (DSD) model is proposed, in which DSD parameters are assumed to be variable for representing the sampling effects. The dependence of statistical moments on the variation of DSD parameters are calculated and applied to radar-based rain estimation.

Polarimetric radar measurements are sensitive to the size, shape and orientation of raindrops and provide information about drop size distribution (DSD), canting angle distribution and rain rate. The authors propose and demonstrate a method for retrieving DSD parameters for calculating rain rate and the characteristic particle size. The DSD is assumed to be a gamma distribution and the governing parameters are retrieved from radar measurements: reflectivity (Z/sub HH/), differential reflectivity (ZDR), and a constrained relation between the shape (CL) and slope (/spl Lambda/) parameters derived from video disdrometer observations. The estimated rain rate is compared with that obtained from more traditional methods and the calculated characteristic size is compared with the measured values. The calculated K/sub DP/ based on the retrieved Gamma DSD is also compared with measurements. The proposed method shows improvement over the existing models and techniques because it can retrieve all three parameters of the gamma distribution. For maintaining the continuity of earlier published results, raindrop shape is assumed to be equilibrium.

Dual-wavelength (K/sub /spl alpha//- and X-band) radar measurements have shown promise in estimating the amount of liquid water in a cloud. By taking advantage of the attenuation by liquid water of the K/sub /spl alpha//-band signal as compared to X-band, the range-differentiated difference in reflectivity can be used to estimate the spatial distribution of cloud liquid water. One limitation is that the method is based on the assumption that all particles in the radar beams act as Rayleigh scatterers, that is, their diameters are small compared to the radar wavelengths. In natural clouds in wintertime conditions, this often may not be the case. This paper presents simulations of the response of these two wavelengths to conditions measured in several geographic locations. The simulations are used to build simplified relations between radar reflectivity and total mass and size distribution functions of liquid droplets and ice particles. Using these relations, it may be possible to estimate the sizes of the droplets, as well as total mass contents and size distributions of ice particles that may also be present in the sampled volume. Results of radar-based retrieval methods applied to measurements in a winter stratiform cloud are discussed, and compared with a previous result. A technique is described for detecting regions of non-Rayleigh scattering and for subsequently estimating liquid water content (LWC). Additional examples of dual-wavelength measurements in regions containing cloud droplets, small ice particles, and larger snowflakes are discussed.

The authors study a correlation imaging method for the detection of targets embedded in clutter environment. The result is based on Monte Carlo simulations. In the simulations, the targets are embedded in a medium consisting of randomly distributed small scatterers and the scattered fields are calculated. The scattered fields are then used for image processing. The incident and scattered directions are chosen to avoid the "memory effect" of clutter scattering. It has been shown that the correlation function of scattered fields due to clutter is small if the memory effect is avoided. The correlation imaging is to calculate the correlation function with focusing on desired locations, Therefore, it can suppress clutter and have finer resolution. The angular correlation (ACF) imaging is used for circular synthetic aperture radar (SAR) and frequency angular correlation (FACF) imaging is used for linear SAR. Numerical results show improvement over the conventional field imaging.

The discrimination of the scattered wave from an object buried in shallow ground from that of the rough surface is a difficult task with present ground penetrating radar (GPR) systems. Recently, a new approach for this classical problem has been proposed and its effectiveness has been verified. This new method is based on the angular correlation function (ACF) of the scattered wave observed at two or more different incident and scattered angle combinations. It has been shown that the angular memory signatures of rough surfaces are substantially different from those of typical man-made targets and by choosing the appropriate incident and scattered angles, the surface scattering can be minimized whereas the scattering from the target is almost unchanged. The authors present detailed numerical studies of the ACF of the scattered wave from rough surfaces with and without a buried object. To obtain the ACF, the three averaging methods: realization, frequency and angular averaging, are tested numerically. It is shown that a single random rough surface of moderate extent can exhibit memory effect by using frequency averaging. Frequency averaging with a wide bandwidth is also effective for suppressing fluctuation in ACF and is most useful for practical applications. Numerical simulations indicate that even when the ratio of scattered intensities with and without the buried object is close to unity, the corresponding ratio of ACF magnitude can be more than 10 dB. Thus, using the ACF is superior to using the radar cross section (RCS) in the detection of buried objects.