Polarization Radar Data Research Articles

Using SAR-data and optical data to delineate burned territories in Kalmykia, Chernye Zemli reserve

The study of fires and their consequences using remote sensing methods in the optical range is widespread. There are many studies devoted to forest fires, a little less to fires in arid areas. At the same time, even in arid regions, weather conditions do not always allow us to get a correct representation of the fire spread in a sufficiently high spatial resolution (the first tens of meters). Our study is related to the territory of Chernye Zemli reserve located in Kalmykia, Russian Federation. In this study, we are conducting an experiment on the joint use of multitemporal synthesis of vertical polarization radar data and the burn index (dNBR2) according to Landsat-8 satellite data. The use of the image segmentation approach gave good results in identifying the boundaries of the burned territories from radar data, and the use of a training sample from optical data index image made it possible to classify the selected superpixels according to the classes of “burned” or “not burned”.

2021年温州地区首场降雪的双偏振雷达观测分析

2021年温州地区首场降雪的双偏振雷达观测分析

Multiangle, Frequency, and Polarization Radar Measurement of Ice Sheets

Radio echo sounding of polar ice sheets provides important information on the ice bed topography and internal layers. These data have been used by scientists to create 3-D maps of polar ice sheets for climate modeling as well as to reconstruct the climate history that dates back to hundreds of thousands of years. In this article, we present the design and development of three surface-based multichannel radars in the VHF and UHF bands. We provide results from radar data multifrequency and polarization radar data collected over the Greenland ice sheet. All the three radars shared the same digital waveform generator and digitizer, and were installed in and operated on a tracked vehicle. The radars are operated with three different antenna arrays designed for operation over 170–230, 180–340, and 600–900 MHz. The results we obtained sounded more than 2.7 km thick ice with radars operating at frequencies as high as 850 MHz with more than 40 dB signal-to-noise ratio.

Assessing the Applicability of Random Forest, Stochastic Gradient Boosted Model, and Extreme Learning Machine Methods to the Quantitative Precipitation Estimation of the Radar Data: A Case Study to Gwangdeoksan Radar, South Korea, in 2018

Machine learning algorithms should be tested for use in quantitative precipitation estimation models of rain radar data in South Korea because such an application can provide a more accurate estimate of rainfall than the conventional ZR relationship-based model. The applicability of random forest, stochastic gradient boosted model, and extreme learning machine methods to quantitative precipitation estimation models was investigated using case studies with polarization radar data from Gwangdeoksan radar station. Various combinations of input variable sets were tested, and results showed that machine learning algorithms can be applied to build the quantitative precipitation estimation model of the polarization radar data in South Korea. The machine learning-based quantitative precipitation estimation models led to better performances than ZR relationship-based models, particularly for heavy rainfall events. The extreme learning machine is considered the best of the algorithms used based on evaluation criteria.

MistNet: Measuring historical bird migration in the US using archived weather radar data and convolutional neural networks

Abstract Large networks of weather radars are comprehensive instruments for studying bird migration. For example, the US WSR‐88D network covers the entire continental US and has archived data since the 1990s. The data can quantify both broad and fine‐scale bird movements to address a range of migration ecology questions. However, the problem of automatically discriminating precipitation from biology has significantly limited the ability to conduct biological analyses with historical radar data. We develop MistNet, a deep convolutional neural network to discriminate precipitation from biology in radar scans. Unlike prior machine learning approaches, MistNet makes fine‐scaled predictions and can collect biological information from radar scans that also contain precipitation. MistNet is based on neural networks for images, and includes several architecture components tailored to the unique characteristics of radar data. To avoid a massive human labelling effort, we train MistNet using abundant noisy labels obtained from dual polarization radar data. In historical and contemporary WSR‐88D data, MistNet identifies at least 95.9% of all biomass with a false discovery rate of 1.3%. Dual polarization training data and our radar‐specific architecture components are effective. By retaining biomass that co‐occurs with precipitation in a single radar scan, MistNet retains 15% more biomass than traditional whole‐scan approaches to screening. MistNet is fully automated and can be applied to datasets of millions of radar scans to produce fine‐grained predictions that enable a range of applications, from continent‐scale mapping to local analysis of airspace usage. Radar ornithology is advancing rapidly and leading to significant discoveries about continent‐scale patterns of bird movements. General‐purpose and empirically validated methods to quantify biological signals in radar data are essential to the future development of this field. MistNet can enable large‐scale, long‐term, and reproducible measurements of whole migration systems.

Approaches to radar reflectivity bias correction to improve rainfall estimation in Korea

Abstract. Three methods for determining the reflectivity bias of single polarization radar using dual polarization radar reflectivity and disdrometer data (i.e., the equidistance line, overlapping area, and disdrometer methods) are proposed and evaluated for two low-pressure rainfall events that occurred over the Korean Peninsula on 25 August 2014 and 8 September 2012. Single polarization radar reflectivity was underestimated by more than 12 and 7 dB in the two rain events, respectively. All methods improved the accuracy of rainfall estimation, except for one case where drop size distributions were not observed, as the precipitation system did not pass through the disdrometer location. The use of these bias correction methods reduced the RMSE by as much as 50 %. Overall, the most accurate rainfall estimates were obtained using the overlapping area method to correct radar reflectivity.

이중편파레이더 강우추정 알고리즘의 국내 적용성 평가: 비슬산 강우 레이더를 위주로

이중편파레이더는 다양한 변수를 관측하여 종합적인 강우정보를 제공한다. 국내에서는 강우정보 획득을 위해 JPOLE 알고리즘과 CSU-HIDRO 알고리즘 등이 주로 이용되고 있다. 본 연구에서는 2014년 8월의 호우사상에 관한 비슬산 이중편파레이더 자료와 기상청 AWS 관측 자료를 이용하여 JPOLE 알고리즘과 CSU-HIDRO 알고리즘의 국내 적용성을 평가하였다. 이를 위해 총 11,372 쌍의 레이더-지상 강우강도 자료를 합성 알고리즘의 문턱값을 기준으로 적합 자료와 부적합 자료로 구분하였다. 다음으로 전체 자료와 적합 및 부적합 자료의 레이더 강우강도와 지상 강우강도를 비교하여 평가 지표를 산정하였다. 그 결과는 다음과 같다. (1) 두 알고리즘에서 비차등위상차를 이용하는 강우추정 관계식은 적합 자료의 결과가 비교적 우수하여 문턱값의 설정이 비교적 잘 되었다고 판단되었다. (2) 수평반사도 및 차등반사도를 이용하는 강우추정 관계식은 상대적으로 부정확한 강우추정 결과를 보였다. 또한 적합 자료만을 고려하는 경우에도 강우추정 결과는 개선되지 않았다. Dual polarization radar provides comprehensive information about rainfall by measuring multiple parameters. In Korea, for the rainfall estimation, JPOLE and CSU-HIDRO algorithms are generally used. This study evaluated the local applicability of JPOLE and CSU-HIDRO algorithms in Korea by using the observed rainfall data collected on the August, 2014 by the Biseulsan dual polarization radar data and KMA AWS. A total of 11,372 pairs of radar-ground rain rate data were classified according to thresholds of synthetic algorithms into suitable and unsuitable data. Then, evaluation criteria were derived by comparing radar rain rate and ground rain rate, respectively, for entire, suitable, and unsuitable data. The results are as follows: (1) The radar rain rate equation including specific differential phase was found better in the rainfall estimation than the other equations for both JPOLE and CSU-HIDRO algorithms. The thresholds were found to be adequately applied for both algorithms including specific differential phase. (2) The radar rain rate equation including horizontal reflectivity and differential reflectivity were found poor compared to the others. The result was not improved even when only the suitable data were applied.

C-Band Polarization Radar Data Preprocessing and Its Application to Rainfall Estimation

C-Band Polarization Radar Data Preprocessing and Its Application to Rainfall Estimation

이중 편파 레이더 자료를 이용한 강수 추정 알고리즘의 오차 분석

강수량의 정확한 추정은 자연 재해의 예측 및 방지를 위해 필수적이다. 최근 강우레이더는 높은 시공간 해상도로 정량적 강수 추정(quantitative precipitation estimation; QPE)을 수행하는 기술로 주목받고 있다. 특히, 레이더의 강수 추정 성능은 다양한 규모에서 여러 수문 기상 변수를 관측하는 이중 편파 기술의 도입에 의해 향상되었다. 이중 편파 레이더는 다양한 매개변수를 측정하기 때문에, 기존의 반사도-강수 관계만을 활용하는 단일 편파 레이더에 비해 강수 추정 정확도 향상이 가능하다. 본 연구의 목적은 이중 편파 레이더에 대한 대표적인 강수 추정 방법인 JPOLE(Ryzhkov et al., 2005a)와 CSU(Cifelli et al., 2011) 알고리즘의 국내 강우레이더 자료에 대한 적용성 및 불확실성을 평가하는 것이다. 이를 위해 S밴드 이중 편파 레이더인 비슬산 강우레이더의 2012년 호우사상을 대상으로 두 가지 알고리즘을 적용하여 강수를 각각 추정하고, 일별 및 시간별 강수의 시공간 분포 정확성에 대해 지상관측 강수와 비교하였다. 또한 관측지점 별 오차를 분석하여 레이더로부터의 거리 및 관측지점 고도와 레이더 정확도와의 상관성을 분석하였다. 분석 결과, JPOLE 및 CSU 알고리즘 모두 표준화된 평균절대오차(nMAE)를 기준으로 일별 강수에 대해 0.23, 시간별 강수 기준으로 0.35 이내의 비교적 정확한 강수량 추정 결과를 산출하였다. 모든 정확도 평가 지표 및 기준에 대해서 JPOLE 알고리즘이 CSU 알고리즘에 비해 우수한 정확도를 보였으나, 그 차이는 평균제곱근오차(RMSE)를 기준으로 50 mm/day 이상 일별 강수에 대해 5.5 mm(약 17%차이), 10 mm/hr 이상 시간별 강수에 대해 0.6 mm(약 8%차이)였다. 관측소 위치 별 오차 분석 결과, 불확실성이 높은 지점의 위치는 일정한 경향이 없었으나, 지상관측지점이 레이더로부터 거리 60 km, 해발고도 고도 100 m 이내의 조건에서 정확도 높은 관측지점이 존재하였다. Accurate estimation of precipitation is essential to predict and prevent natural disasters (e.g. flood, landslide, and heavy snowfall). Recently, weather radars have become a popular tool for quantitative precipitation estimation (QPE) with high spatio-temporal resolution. Especially, in the last decade, QPE performance has been improved by introduction of polarimetric technology to observe multiple hydrometeorological variables at various scales. By being able to measure variables such as the differential reflectivity, specific differential phase, and cross-correlation coefficient, the reliability of estimation has significantly improved as compared to the reflectivity-based method (Ryzhkov et al., 2005a). However, QPEs using dual polarization radar data are still subject to uncertainties resulting from rainfall conversion relationships, combination methods of different parameters, and sampling errors. In this study, we assessed the uncertainty and applicability of conventional QPE algorithms, such as JPOLE (Ryzhkov et al., 2005a) and CSU (Cifelli et al., 2011) algorithms, using the dual polarization radar at Mt. Biseul in the south-eastern part of the Korean Peninsula. Analysis results illustrated that the JPOLE algorithm outperformed the CSU algorithm slightly for daily and hourly rainfall analysis using various metrics. The higher accuracy was found in the stations located within less than 60 km from radar and 100 m in the elevation.

Comparison and integration of Landsat optical and PALSAR quad polarization radar: a case study in Bangladesh

With the recent increase in the availability of quad polarization radar data, the need to assess the utility of these datasets for land cover/use classification is important. The relative classification accuracies of four land covers/uses in Bangladesh using spaceborne quad polarization radar from the Japanese ALOS PALSAR system and optical Landsat Thematic Mapper (TM) data were evaluated. In addition, the utility of radar texture and sensor fusion were analyzed. Supervised signature extraction and classification (maximum likelihood) were used to classify different land covers/uses followed by an accuracy assessment. The original four-band radar had an overall accuracy of 91%. Variance texture was the most useful of the four measures examined, but did not improve overall or individual class accuracies over the original radar. Landsat provided a higher overall classification accuracy (94%) as compared to radar. The merger of Landsat with the original radar increased overall accuracy to 99%, which indicates the advantages of sensor integration.

Correction of Reflectivity in the Presence of Partial Beam Blockage over a Mountainous Region Using X-Band Dual Polarization Radar

AbstractTwo approaches to correcting the partial beam blockage of radar reflectivity in mountainous areas were evaluated using X-band dual polarization radar data from the Hakone mountain region, Kanto, Japan. The comparatively simple digital elevation model (DEM) method calculates the power loss in the received signal based on the geometrical relationship between radar beams and a DEM. The second approach, the modified DEM method, attempts to account for unknown power losses related to ground clutter, hardware calibration errors, etc. Comparison between ground data and reflectivity data corrected by both methods suggests that the DEM method alone was insufficient to correct beam blockage problems but that the modified DEM data were in generally good agreement with the ground data. The authors also estimated 10-min rainfall amounts using reflectivity corrected by the modified DEM method and compared these with data from a network of rain gauges in the mountainous region. In general, the results show good agreement between radar estimates and rain gauge measurements. On the basis of their results, the authors conclude that the modified DEM method offers a suitable solution to the problem of beam blockage in mountainous regions, provided that the beam blockage rate is less than 80%.

Comparison of S-Band Radar Attenuation Prediction With Beacon Measurements

A comparative analysis of the rain attenuation evaluated from beacon measurements and from a single polarization S-band Radar is performed. The beacon measurements are obtained for two slant satellite paths with different elevation angles in the Ku- and Ka-band. The single polarization S-band Radar reflectivity data is used to predict the attenuation along the satellite propagation paths. This is done by first converting the reflectivity into rain rate using the Z-R relations suitable for the tropical region and, afterwards, by evaluating the slant path attenuation through the integration of the specific rain attenuation derived from the point rainfall rate. An empirical calibration factor for the Radar reflectivity is provided. A comparison of the rain fade suffered from the two satellite paths is presented both on event basis and in terms of cumulative distribution functions. The empirical calibration factor and a single Z-R relation suitable for the tropical region are used for rain attenuation evaluation from single polarization Radar data.

Optimization of threshold ranges for rapid flood inundation mapping by evaluating backscatter profiles of high incidence angle SAR images

Rapid satellite-based flood inundation mapping and delivery of flood inundation maps during a flood event can provide crucial information for planners and decision makers to prioritize relief and rescue operations. The present study is undertaken to optimize the threshold ranges for the classification of flood water in Synthetic Aperture Radar (SAR) images (of 20° to 49° incidence angles) for quick flood inundation mapping and response during flood disasters. This is done through assessing the signature of flood water in Horizontal transmit and Horizontal received (HH), Horizontal transmit and Vertical received (HV), Vertical transmit and Horizontal received (VH), and Vertical transmit and Vertical received (VV) polarization radar data. The mean backscattering signature profiles of various water bodies were analyzed to discriminate flood water from other water bodies. The study shows that there is better demarcation of land-water surface in HH polarization. VV polarization has the potential to identify partially submerged features, which can be useful in flood damage assessments. The backscatter of flood water in HV and VH is the same and both HV and VH polarizations are adequate for the mapping of flood water. At near range to far range, −8 to −12 dB, −15 to −24 dB, and −6 to −15 dB can be used as optimum ranges for the classification of flood water in HH, HV, and VV polarizations. These optimum threshold ranges can be applied to the automation of flood mapping using SAR images in near-real time, where much time was often spent on finding the thresholds in order to produce flood inundation maps in a short time from the onset of flood disasters and deliver such maps to the concerned agencies.

Class Separability Analysis and Classifier Comparison using Quad-polarization Radar Imagery

The potential of quad polarization radar data for the target discrimination has been analyzed. Quad polarization data of the RADARSAT-2 fine resolution mode has been utilized. Class separability analysis has been carried out on different polarization combinations using Transformed Divergence (TD) method and it is observed that HH-HV/VH-VV polarization combination gives better class separability when compared to other polarization combinations. Classification has been carried out on the optimized polarization combination using Maximum likelihood (MLC) and Support Vector Machine (SVM) classifiers. It is observed that SVM classification gives better classification accuracy compared to MLC. Overall classification accuracy is 93.03% for SVM and 88.78% for MLC. Class separability and classification accuracy comparison results are presented.

Comparison and integration of optical and quadpolarization radar imagery for land cover/use delineation

With the recent increase in the availability of multiple polarization radar data, the need to assess the relative usefulness of these data for land cover/use classification and other applications is important and is the premise of this study. The primary methodology employed is spectral signature extraction and use of Transformed Divergence (TD) separability measures to evaluate the relative utility of multiple radar bands and polarizations (Shuttle Imaging Radar-C), optical imagery (Landsat Thematic Mapper) and texture derived information for a study site in Bangladesh. Variance texture measures at a 7x7 window size are used in an evaluation of improving the overall separability among classes. SIR-C L-band provided slightly better results than C-band but no single polarization provided consistently higher TD values. Texture greatly improved separability for all SIR-C bands but did not assist TM separability. This study also indicated the value of combining TM and SIR-C and particularly the TM original data and SIR-C texture.

Regolith properties in the south polar region of the Moon from 70-cm radar polarimetry

The south polar region of the Moon contains areas permanently shadowed from solar illumination, which may provide cold traps for volatiles such as water ice. Previous radar studies have emphasized the search for diagnostic polarization signatures of thick ice in areas close to the pole, but near-surface regolith properties and regional geology are also important to upcoming orbital studies of the shadowed terrain. To study regional regolith variations, we collected 70-cm wavelength, 450-m resolution, dual-circular polarization radar data for latitudes 60–90° S using the Arecibo and Greenbank telescopes. The circular polarization ratio, μ c , is sensitive to differences in rock abundance at the surface and up to tens of m below the surface, depending upon the regolith loss tangent. We observe significant variations in μ c , attributed to changes in the surface and subsurface rock population, across the south polar highlands. Concentric haloes of low polarization ratio surrounding Hausen, Moretus, and other young craters represent rock-poor ejecta layers. Values of μ c up to ∼1 occur in the floors and near-rim deposits of Eratosthenian and Copernican craters, consistent with abundant rocky ejecta and/or fractured impact melt. Enhanced μ c values also correspond to areas mapped as Orientale-derived, plains-forming material [Wilhelms, D.E., Howard, K.A., Wilshire, H.G., 1979. USGS Map I-1162], and similar polarization properties characterize the permanently shadowed floors of craters Faustini and Shoemaker. Small areas of very high (>1.5) circular polarization ratio occur on shadowed and seasonally sunlit terrain, and appear to be associated with small craters. We suggest that regolith in low-lying areas near the south pole is characterized by a significant impact melt component from Orientale, which provides a source for excavation of the block-rich ejecta around small craters observed in this and earlier radar studies. The lower portion of the interior wall of Shackleton crater, permanently shadowed from the sun but visible from Earth, is not significantly different in 70-cm scattering properties from diurnally/seasonally sunlit areas of craters with similar morphology.

A model for retrieval of dual linear polarization radar fields from model simulation outputs

An algorithm for retrieving polarimetric variables from numerical model fields is developed. By using this technique, radar reflectivity at horizontal polarization, differential reflectivity, specific differential phase shift and correlation coefficients between the horizontal and vertical polarization signals at zero lag can be derived from rain, snow and hail contents of numerical model outputs. Effects of environmental temperature and the melting process on polarimetric variables are considered in the algorithm. The algorithm is applied to the Advanced Regional Prediction System (ARPS) model simulation results for a hail storm. The spatial distributions of the derived parameters are reasonable when compared with observational knowledge. This work provides a forward model for assimilation of dual linear polarization radar data into a mesoscale model.

New airborne imaging radar observations of sand Dunes: Kelso Dunes, California

High resolution multiple wavelength and polarization radar data of Kelso Dunes were acquired during the Mojave Field Experiment. They offer the opportunity to develop criteria for the selection of optimal wavelengths and polarizations for imaging of dune areas. The backscatter from unvegetated dune surfaces at all wavelenghts (C-, L-, and P-band) with direct (HH, VV) polarization is largely controlled by dune topography and incidence angle. These data (especially P-band) provide the best images of dune topography. The best discrimination between active and inactive dunes is obtained by cross-polarized C-band data (CVH, CHV) as a result of multiple scattering from vegetation and small-scale topography (e.g., shadow dunes).

A field study of small raindrop oscillations

Measurements were made at the ground in Hawaiian showers to detect oscillations of small raindrops as predicted by recent laboratory experiments. Oscillation frequencies and amplitudes were determined from photographs using backscattered light of the primary rainbow. A strobe provided highlights so that the fall speeds and corresponding sizes of each drop could be calculated. The drop‐size distributions were measured independently with a collocated disdrometer. The observations of 0.3 mm to 2.0 mm diameter raindrops clearly showed the onset of oscillations near 1 mm, at the same size where eddy shedding begins. Alternative causes of drop oscillations from drop collisions and wind shear were eliminated using estimates based on drop‐size distributions and winds. Our inferred oscillation sizes, modes and amplitudes were consistent with several previous measurements where small raindrops were found to be generally more spherical than the non‐oscillating shapes predicted by theory. This field study firmly establishes eddy shedding as the major cause of natural shape variations for small raindrops in precipitation—and provides further evidence that significant corrections are needed in estimating the scattering of microwaves in light to heavy rain showers (e.g., for interpreting polarization radar data and evaluating rain outages in communication links).

Utilization of a mobile laboratory for storm electricity measurements

Since 1978, we have been developing and using a mobile laboratory to make observations of storms. With our mobile lab, we record electric field, slow and fast electric field changes from lightning, optical transients, return stroke velocity, temperature, pressure, dew point, wind speed and direction, lab position, and video images of clouds and lightning. The mobile lab idea has been key in obtaining new information on several phenomena, e.g., field changes to compare with high‐altitude airplane measurements, electrical aspects of tornadic storms, and positive ground flashes. Mobile lab data were critical in an evaluation of the two ground strike location networks that provide scientific and operational data in the United States. We have documented that storms often are more dominated by intracloud (IC) lightning than previously reported. Observations on two tornadic storms show the IC flashes averaged ≥80% and 95% of the total lightning, respectively. Mobile lab and polarization radar data of storms producing wet microbursts in Alabama suggest that high percentages of IC flashes (95%) occur when a larger percentage of the total storm mass is convected above the freezing level. With the recent addition of mobile balloon launch and tracking, we also make soundings of the electric field and meteorological parameters within storms.