Weather Radar Research Articles

Radar Echo Recognition of Gust Front Based on Deep Learning

Gust fronts (GFs) belong to the boundary layer convergence system. A strong GF can cause serious wind disasters, so its automatic monitoring and identification are very helpful but difficult in daily meteorological operations. By collecting convective weather processes in Hubei, Jiangsu, and other regions of China, 1422 GFs from 106 S-band new-generation weather radar (CINRAD/SA) volume scan data are labeled as positive samples by means of human–computer interaction, and the same number of negative samples are randomly tagged from no GF radar data. A deep learning dataset including 2844 labels with a positive and negative sample ratio of 1:1 is constructed, and 80%, 10%, and 10% of the dataset are separated as training, validation, and test sets, respectively. Then, the training dataset is expanded to 273,120 samples by data augmentation technology. Since the height of a GF is generally less than 1.5 km, three deep-learning-based models are trained for GF automatic recognition according to the distance from the radars. Three models (M1, M2, M3) are trained with the data at a 0.5° elevation angle from 65 to 180 km away from the radars, at 0.5° and 1.5° angles from 40 to 65 km, and at 0.5°, 1.5°, and 2.4° angles within 40 km, respectively. The precision, confusion matrix, and its derived indicators including receiver operating characteristic curve (ROC) and the area under ROC (AUC) are used to evaluate the three models by the test set. The results show that the identification precisions of the models are 97.66% (M1), 90% (M2), and 90.43% (M3), respectively. All the hit rates are over 89%, the false positive rates are less than 11%, and the critical success indexes (CSIs) surpass 82%. In addition, all the optimal critical points on the ROC curves are close to (0, 1), and the AUC values are above 0.93. These results suggest that the three models can effectively achieve the automatic discrimination of GFs. Finally, the models are demonstrated by three GF events detected with Qingpu, Nantong, and Cangzhou radars.

Measuring diameters and velocities of artificial raindrops with a neuromorphic event camera

Abstract. Hydrometers that measure size and velocity distributions of precipitation are needed for research and corrections of rainfall estimates from weather radars and microwave links. Existing optical disdrometers measure droplet size distributions, but underestimate small raindrops and are impractical for widespread always-on IoT deployment. We study the feasibility of measuring droplet size and velocity using a neuromorphic event camera. These dynamic vision sensors asynchronously output a sparse stream of pixel brightness changes. Droplets falling through the plane of focus of a steeply down-looking camera create events generated by the motion of the droplet across the field of view. Droplet size and speed are inferred from the hourglass-shaped stream of events. Using an improved hard disk arm actuator to reliably generate artificial raindrops with a range of small sizes, our experiments show maximum errors of 7 % (mean absolute percentage error) for droplet sizes from 0.3 to 2.5 mm and speeds from 1.3 to 8.0 m s−1. Measurements with the same setup from a commercial PARSIVEL disdrometer show similar results. Both devices slightly overestimate the small droplet volume with a volume overestimation of 25 % from the event camera measurements and 50 % from the PARSIVEL instrument. Each droplet requires processing of 5000 to 50 000 brightness change events, potentially enabling low-power always-on disdrometers that consume power proportional to the rainfall rate. Data and code are available at the paper website https://sites.google.com/view/dvs-disdrometer/home (Micev et al., 2023).

Diurnal Variation Characteristics of Raindrop Size Distribution Observed by a Parsivel2 Disdrometer in the Ili River Valley

The diurnal variation characteristics of raindrop size distribution (RSD) in the Ili River Valley are investigated in this study, using the RSD data from May to September during 2020-2021 collected by a Parsivel2 disdrometer in Zhaosu. Significant diurnal variations (02–07, 08–13, 14–19, and 20-01 local standard time (LST)) of precipitation and RSD in Zhaosu are revealed during the rainy seasons. Precipitation mainly occurs in the late afternoon and early evening. A higher concentration of small raindrops is observed in the morning, whereas more mid-size and large raindrops are observed in the afternoon. The RSD exhibits diurnal differences between different rainfall rate classes; the diurnal difference of RSD is more pronounced in the case of high rainfall rates. Stratiform precipitation can occur at any time of the day, yet convective precipitation mainly occurs during the late afternoon and early evening. The RSD of stratiform rainfall shows a similar distribution over the four time periods. For convective rainfall, the concentration of small raindrops is the highest (lowest) over 02–07 (14–19) LST, while the highest (lowest) concentration of medium and large drops is observed over 14–19 (02–07) LST. Convective rain in the Ili River Valley over 14–19 LST can be characterized as the continental convective cluster, while in the rest time of the day, it is neither in the maritime cluster nor in the continental cluster. The empirical relationships between the radar reflectivity factor and rainfall rate (Z-R) for stratiform and convective rain types are also derived. The purpose of this study is to advance our understanding of precipitation microphysics in arid mountainous region.

Assessing sampling and retrieval errors of GPROF precipitation estimates over the Netherlands

Abstract. The Goddard Profiling algorithm (GPROF) converts radiometer observations from Global Precipitation Measurement (GPM) constellation satellites into precipitation estimates. Typically, high-quality ground-based estimates serve as reference to evaluate GPROF's performance. To provide a fair comparison, the ground-based estimates are often spatially aligned to GPROF. However, GPROF combines observations from various sensors and channels, each associated with a distinct footprint. Consequently, uncertainties related to the representativeness of the sampled areas are introduced in addition to the uncertainty when converting brightness temperatures into precipitation intensities. The exact contribution of resampling precipitation estimates, required to spatially and temporally align different resolutions when combining or comparing precipitation observations, to the overall uncertainty remains unknown. Here, we analyze the current performance of GPROF over the Netherlands during a 4-year period (2017–2020) while investigating the uncertainty related to sampling. The latter is done by simulating the reference precipitation as satellite footprints that vary in size, geometry, and applied weighting technique. Only GPROF estimates based on observations from the conical-scanning radiometers of the GPM constellation are used. The reference estimates are gauge-adjusted radar precipitation estimates from two ground-based weather radars from the Royal Netherlands Meteorological Institute (KNMI). Echo top heights (ETHs) retrieved from the same radars are used to classify the precipitation as shallow, medium, or deep. Spatial averaging methods (Gaussian weighting vs. arithmetic mean) minimally affect the magnitude of the precipitation estimates. Footprint size has a higher impact but cannot explain all discrepancies between the ground- and satellite-based estimates. Additionally, the discrepancies between GPROF and the reference are largest for low ETHs, while the relative bias between the different footprint sizes and implemented weighting methods increase with increasing ETHs. Lastly, our results do not show a clear difference between coastal and land simulations. We conclude that the uncertainty introduced by merging different channels and sensors cannot fully explain the discrepancies between satellite- and ground-based precipitation estimates. Hence, uncertainties related to the retrieval algorithm and environmental conditions are found to be more prominent than resampling uncertainties, in particular for shallow and light precipitation.

Improvements in deep learning-based precipitation nowcasting using major atmospheric factors with radar rain rate

Recently, deep learning-based precipitation nowcasting has been investigated and its usefulness has been recognized. However, existing approaches have treated precipitation nowcasting as a spatiotemporal sequence prediction problem and have mainly used only radar images. Radar images show the distribution of water or ice droplets, but are limited in providing information about the dynamic or thermodynamic processes involved in the development of the rainfall system. In this study, we introduced a deep neural network that can utilize atmospheric factors that play major roles in the development of rainfall systems over the Korean Peninsula, including divergence at 925 hPa and total column water vapor, in addition to radar images. We also proposed a loss function based on mean categorical scores (e.g., critical success index and false alarm ratio) to minimize the performance error. In a quantitative evaluation of 1- to 6-hour forecast results, our deep learning models outperformed the conventional approach based on radar echo extrapolation and produced higher equitable threat scores (ETSs) for moderate to severe rain (i.e., 5, 10, and 20 mm h−1 thresholds). By applying the proposed loss function and using the divergence at 925 hPa as an additional input, the deep-learning model not only obtained the highest ETS values, but also revealed its potential to predict new developments of some heavy rainfall systems. The relationship between the presented deep learning-based precipitation nowcasting and the distribution features of the additional atmospheric factors is discussed further in the case study.

Intelligent Reconstruction of Radar Composite Reflectivity Based on Satellite Observations and Deep Learning

Weather radar is a useful tool for monitoring and forecasting severe weather but has limited coverage due to beam blockage from mountainous terrain or other factors. To overcome this issue, an intelligent technology called “Echo Reconstruction UNet (ER-UNet)” is proposed in this study. It reconstructs radar composite reflectivity (CREF) using observations from Fengyun-4A geostationary satellites with broad coverage. In general, ER-UNet outperforms UNet in terms of root mean square error (RMSE), mean absolute error (MAE), structural similarity index (SSIM), probability of detection (POD), false alarm rate (FAR), critical success index (CSI), and Heidke skill score (HSS). Additionally, ER-UNet provides the better reconstruction of CREF compared to the UNet model in terms of the intensity, location, and details of radar echoes (particularly, strong echoes). ER-UNet can effectively reconstruct strong echoes and provide crucial decision-making information for early warning of severe weather.

Merging weather radar data and opportunistic rainfall sensor data to enhance rainfall estimates

Accurate rainfall estimates with high spatiotemporal resolution are vital for urban drainage applications, and if adjusted to ground observations, weather radar data has potential for these applications. The density of meteorological rain gauges for adjustment is, however, often sparse and non-uniformly distributed in space. Opportunistic rainfall sensors provide an increased density of ground observations but often with reduced or unknown accuracy for each individual station. This paper demonstrates the merging of data from weather radar, personal weather stations, and commercial microwave links into an integrated rainfall product. Merging opportunistic rainfall estimates is shown to improve the accuracy of the opportunistic rainfall observations through a quality control algorithm. In this study, we show that the accuracy of rainfall estimates is significantly improved by merging opportunistic rainfall data and weather radar data when compared to the accuracy of each rainfall product without merging. Nash-Sutcliffe efficiency (NSE) values up to 0.88 are obtained for daily accumulated merged rainfall products, while the NSE-values of the individual rainfall products range from −7.44 to 0.65, and similar tendencies are observed for root mean squared error (RMSE) values. For merging weather radar and opportunistic rainfall data, a novel approach, i.e., “moving median bias adjustment” is presented. Applying this approach, a high-performing rainfall product is derived independently from conventional high-quality rain gauges, which in this study are only used for independent validation. In addition, it is demonstrated that accurate rainfall estimates can be obtained by sub-daily merging, underscoring the importance of merging in nowcasting and near real-time applications.

An early warning approach for the rapid identification of extreme weather disasters based on phased array dual polarization radar cooperative network data.

In recent years, X-band phase array dual polarization weather radar technology has matured. The cooperative networking data from X-band phase array dual polarization weather radar have many advantages compared with traditional methods, namely, high spatial and temporal resolution (approximately 70 seconds in one scan, 30 m in radial distance resolution), wide coverages that can compensate for the observation blind spots, and data fusion technology that is used in the observation overlap area to ensure that the observed precipitation data have spatial continuity. Based on the above radar systems, this study proposes an improved hail and lightning weather disaster rapid identification and early warning algorithm. The improved thunderstorm identification, tracking, analysis, and nowcasting (TITAN) algorithm is used to quickly identify three-dimensional strong convective storm cells. Large sample observation experiment data are used to invert the localized hail index (HDR) to identify the hail position. The fuzzy logic method is used to comprehensively determine the probability of lightning occurrence. The comparative analysis experiment shows that, compared with the live observation data from the ground-based automatic station, the hail and lightning disaster weather warning algorithm developed by this study can increase warning times by approximately 7 minutes over the traditional algorithm, and its critical success index (CSI), false alarm ratio (FAR) and omission alarm ratio (OAR) scores are better than those of the traditional method. The average root mean square error (ARMSE) for identifying hail and lightning locations by this improved method is also significantly better than that of traditional methods. We show that our method can provide probabilistic predictions that improve hail and lightning identification, improve the precision of early warning and support operational utility at higher resolutions and with greater lead times that traditional methods struggle to achieve.

Evaluasi Spasial Estimasi Curah Hujan pada Radar Cuaca Menggunakan Metode Z-R Marshall-Palmer di Wilayah Jawa Barat

Rainfall is one of the weather parameters that affect various sectors. High rainfall intensity can trigger hydrometeorological disasters, so rainfall observation data is vital to monitor rainfall conditions in an area. An automatic rain gauge is an instrument that measures rainfall at an observation point, but the instrument has reasonably low coverage and has yet to reach the entire region. Weather radar is a remote sensing instrument capable of spatially estimating rainfall. Weather radar data can be used to estimate rainfall using the Marshall-Palmer Z-R method. The application of the method can be an alternative for areas that do not have rainfall observation equipment. However, the estimation needs to be evaluated to improve the accuracy of the estimation value. Based on the evaluation, the highest coefficient of determination was 0.92, and the lowest was 0.67. The lowest RMSE value was 2.40, the highest was 6.76, the highest ME value was 16.59, and the lowest was 5.93; the highest bias was 12.90, and the lowest was 5.30. The study results show that the weather radar can operate according to the specifications of the maximum observation distance of up to 220 KM, but the farther the observation distance to a point, the higher the performance of rainfall estimation accuracy.

Features of Atmospheric Dynamics During Tropical Cyclone AMPHAN Observed Using ST Radar and Doppler Weather Radar

Features of Atmospheric Dynamics During Tropical Cyclone AMPHAN Observed Using ST Radar and Doppler Weather Radar

On the Radial-Based Estimation of Colored Noise Power in Weather Radars

On the Radial-Based Estimation of Colored Noise Power in Weather Radars

Doppler Velocity Recovery and Dealiasing Algorithm for Multi-PRT Scans in Weather Radars

Doppler Velocity Recovery and Dealiasing Algorithm for Multi-PRT Scans in Weather Radars

An Animal Migration Forecast Model with Weather Radar and Meteorological Data

An Animal Migration Forecast Model with Weather Radar and Meteorological Data

Adopting a Neuro-Fuzzy Logic Method for Fall Armyworm Detection and Monitoring Using C-Band Polarimetric Doppler Weather Radar With Field Verification

Adopting a Neuro-Fuzzy Logic Method for Fall Armyworm Detection and Monitoring Using C-Band Polarimetric Doppler Weather Radar With Field Verification

Counter-Aliasing Is Better Than De-Aliasing: Application to Doppler Weather Radar With Aperiodic Pulse Train

Counter-Aliasing Is Better Than De-Aliasing: Application to Doppler Weather Radar With Aperiodic Pulse Train

Warn-on-Forecast System: From Vision to Reality

Abstract In 2009, advancements in NWP and computing power inspired a vision to advance hazardous weather warnings from a warn-on-detection to a warn-on-forecast paradigm. This vision would require not only the prediction of individual thunderstorms and their attributes but the likelihood of their occurrence in time and space. During the last decade, the warn-on-forecast research team at the NOAA National Severe Storms Laboratory met this challenge through the research and development of 1) an ensemble of high-resolution convection-allowing models; 2) ensemble- and variational-based assimilation of weather radar, satellite, and conventional observations; and 3) unique postprocessing and verification techniques, culminating in the experimental Warn-on-Forecast System (WoFS). Since 2017, we have directly engaged users in the testing, evaluation, and visualization of this system to ensure that WoFS guidance is usable and useful to operational forecasters at NOAA national centers and local offices responsible for forecasting severe weather, tornadoes, and flash floods across the watch-to-warning continuum. Although an experimental WoFS is now a reality, we close by discussing many of the exciting opportunities remaining, including folding this system into the Unified Forecast System, transitioning WoFS into NWS operations, and pursuing next-decade science goals for further advancing storm-scale prediction. Significance Statement The purpose of this research is to develop an experimental prediction system that forecasts the probability for severe weather hazards associated with individual thunderstorms up to 6 h in advance. This capability is important because some people and organizations, like those living in mobile homes, caring for patients in hospitals, or managing large outdoor events, require extended lead time to protect themselves and others from potential severe weather hazards. Our results demonstrate a prediction system that enables forecasters, for the first time, to message probabilistic hazard information associated with individual severe storms between the watch-to-warning time frame within the United States.

Phase and Amplitude Correction for Adaptive Beamforming of Phased Array Weather Radar

Phase and Amplitude Correction for Adaptive Beamforming of Phased Array Weather Radar

Improved Joint Phase–Attenuation Estimation With Adaptive and High-Resolution Empirical Coefficient Conditioning for Polarimetric Weather Radars

Improved Joint Phase–Attenuation Estimation With Adaptive and High-Resolution Empirical Coefficient Conditioning for Polarimetric Weather Radars

Phased Array Weather Radar Architectures for Doppler Estimation with Space Time Processing

Phased Array Weather Radar Architectures for Doppler Estimation with Space Time Processing

Аномальный грозовой процесс на территории Ставропольского края

The paper analyzes the dynamics of such anomalous weather phenomenon as a thunderstorm. The thunderstorm process that took place in the North Caucasus on June 27, 2023 is considered. The thunderstorm process under consideration differed significantly from the statistical average, both in duration and in lightning parameters. The analysis was carried out using the example of the Stavropol Krai, although anomalous thunderstorms on June 27, 2023 also occurred in other regions of the North Caucasus. The work used the lightning direction network of the Federal State Budgetary Institution VGI with spatially separated sensors LS8000 and LS7002. To determine the characteristics of thunderstorm activity, the work used long-term data from the specified lightning direction network. Studies of thunderstorm activity were carried out in conjunction with the characteristics of convective clouds determined by MRL-5 weather radars. The Federal State Budgetary Institution VGI has organized the uptake, processing, archiving and transmission to the ASPD network of Roshydromet of radar information from all MRL-5 weather radars of the anti-hail services of the North Caucasus. During the day, about 30,000 lightning strikes were recorded by VGI in this territory. Of these, 3,713 were ground discharges. Of the registered 3,713 ground-based lightning discharges in the Stavropol Krai, negative lightning amounted to 2,901, and the positive one – 812. The share of negative lightning discharges from the total number of ground-based lightning discharges in the specified territory was 78%. The largest number of negative lightning (more than 70%) had values up to 20 kA. More than 40% of positive lightning discharges had values from 21 to 40 kA. The average value of currents of negative polarity was -18.6 kA, and that of positive polarity was +53.2 kA.