Study on Radar Echo-Filling in an Occlusion Area by a Deep Learning Algorithm

Radar beam blockage is an important error source that affects the quality of weather radar data. An echo-filling network (EFnet) is proposed based on a deep learning algorithm to correct the echo intensity under the occlusion area in the Nanjing S-band new-generation weather radar (CINRAD/SA). The training dataset is constructed by the labels, which are the echo intensity at the 0.5° elevation in the unblocked area, and by the input features, which are the intensity in the cube including multiple elevations and gates corresponding to the location of bottom labels. Two loss functions are applied to compile the network: one is the common mean square error (MSE), and the other is a self-defined loss function that increases the weight of strong echoes. Considering that the radar beam broadens with distance and height, the 0.5° elevation scan is divided into six range bands every 25 km to train different models. The models are evaluated by three indicators: explained variance (EVar), mean absolute error (MAE), and correlation coefficient (CC). Two cases are demonstrated to compare the effect of the echo-filling model by different loss functions. The results suggest that EFnet can effectively correct the echo reflectivity and improve the data quality in the occlusion area, and there are better results for strong echoes when the self-defined loss function is used.

The speed of movement of piece space objects along the Earth's surface and its determination using incoherent scatter radar

Methodological features of registration and separation of coherent radar reflections from space objects and elements of “space debris” operating in orbit are considered. Registration occurs against the background of signals that are scattering of the probe radio wave on particles of the ionospheric plasma. Methods of how to obtain information about the components of the velocity vector of these objects in near-earth space with the help of specialized ground-based radar facilities are analyzed. Their disadvantage is the unreliable control of weak reflections from the elements of “space debris” if they have a small (up to centimeters) scattering cross section. The authors proposed to use the existing high-energy radar installations. Using the signals after the analog-to-digital conversion generated in quadrature, it is proposed to calculate the phase characteristics of the coherent reflection. The radial velocity of the objects along the radar beam is calculated by isolating the Doppler phase difference and statistically averaging these values ​​in the time of reflection. Similarly, by analyzing the time spent in the radar beam, the velocity component associated with the horizontal movement along the Earth’s surface is calculated. Real examples are given, when in one of the observation sessions on the reflection of a signal from a space object, the phase shift in each of its periods is calculated, and then, using the formula, proposed by the authors, the vertical component of the velocity of this object is calculated. Analyzing the observation time of this object in the beam of the transmitter antenna, an example of the calculation and the component of its horizontal velocity is shown. The block diagram of the radar used to calculate the specified parameters of the movement of space objects is presented. The developed approach is an effective solution of many practical problems in those industries that ensure the operation of spacecraft, ensuring the safety of space stations, optimal placement of objects in orbit, etc. Keywords: Incoherent scatter radar, space objects, coherent reflection, signal phase characteristics, radial and horizontal speed

Radar Quality Index for a Mosaic of Radar Reflectivity over Chao Phraya River Basin, Thailand

The weather radar is one of the tools that can provide spatio-temporal information for nowcast which is useful for hydro-meteorological disasters warning and mitigation system. The ground-based weather radar can provide spatial and temporal information to monitor severe storm over the risky area. However, the usage of multiple radars can provide more effective information over large study area where single radar beam may be blocked by surrounding terrain Even though, the investigation of the sever storm physical characteristics needs the information from multiple radars, the mosaicked radar product has not been available for Thai researcher yet. In this study, algorithm of mosaicked radar reflectivity has been developed by using data from ground-based radar of Thai Meteorological Department over the Chao Phraya river basin in the middle of Thailand. The Python script associated with OpenCV and Wradlib libraries were used in our investigations of the mosaicking processes. The radar quality index (RQI) field has been developed by implementing an equation of a quality radar index to identify the reliability of each mosaicked radar reflectivity pixels. First, the percentage of beam blockage is computed to understand the radar beam propagation obstructed by surrounding topography in order to clarify the limitations of the observed beam on producing radar reflectivity maps. Second, the elevation of beam propagation associated with distance field has been computed. Then, these three parameters and the obtained percentage of beam blockage are utilized as the parameters in the equation of RQI. Finally, the detected radar flare, non-precipitating radar area, has been included to the RQI field. Then, the RQI field has been applied to the extracted radar reflectivity to evaluate the quality of mosaicked radar reflectivity to inform end user in any application fields over the Chao Phraya river basin.

Swell and Wind Wave Inversion Using a Single Very High Frequency (VHF) Radar

A hybrid, empirical radar wave inversion technique that treats swell and wind waves separately is presented and evaluated using a single 48-MHz radar unit and in situ wave measurements. This hybrid approach greatly reduces errors in radar wave inversion during swell seas. Our analysis suggests that, prior to the inversion, the second-order spectrum should be normalized using Barrick’s weighting function because this process removes harmonic and corner reflection peaks from the inversion and improves the results. In addition, the resulting calibration constants for the wind wave component are not wave-frequency dependent and are similar in magnitude to those found in previous studies using different operating-frequency radars. This result suggests radar frequency independence, although additional experimental verification is required. The swell component of the model presented here ignores the effect of swell’s propagation direction on the radar signal. Although this approach has several limitations and may only be useful near the coast (where swell propagates close to perpendicular to the coastline), the resulting wave inversion is accurate even when swell is propagating close to perpendicular to the radar beam direction. RMS differences relative to in situ wave height measurements range from 0.16 to 0.25 m as the radar beam angle increases from 22° to 56°.

Using High-Resolution Simulations to Quantify Errors in Radar Estimates of Tornado Intensity

Observation experiments are performed on a set of high-resolution large-eddy simulations of translating tornado-like vortices. Near-surface Doppler wind measurements are taken by emulating a mobile radar positioned from 1 to 10 km south of each vortex track and conducting single-level scans every 2 s. The departure of each observed gust (wind measurement averaged over two successive scans) from the corresponding true maximum 3-s gust at 10 m AGL (“S10–3s”) is partitioned into error sources associated with resolution volume size, wind direction relative to the radar beam, beam elevation, and temporal sampling. The distributions of each error type are diagrammed as functions of range, observed wind speed, and predicted deviation between the wind direction and the radar beam. The results indicate that the deviation between the wind direction and the radar beam is the predominant source of error in these rapid scan scenarios, although range is also a substantial factor. The median total error is ~10% for small deviation at close range, but it approximately doubles if the range is increased from 1 to 10 km; a more pronounced increase in both the median value and the variance of the total error is seen as the deviation becomes large. Because of this, the underestimate of the global maximum S10–3s approaches 30–40 m s−1 at a longer range, although the global maximum of the time-averaged observed wind speed gives a reasonable approximation of the time-mean maximum S10–3s in many cases. Because of simplifying assumptions and the limited number of cases examined, these results are intended as a baseline for further research.