Fourth Calibration and Monitoring Workshop Programme

Programme

The Programme is in GMT timezone.

Day 1 Programme
Day 1 Wednesday 8th November 2023
 09:00 - 09:30 Registration and Coffee (set up posters)
Introduction   
09:30 – 09:45

Introduction to the Workshop 

Met Office, Sandeep Sanghera, Timothy Darlington 

09:45 - 10:00

Welcome address 

Bruce Truscott

Met Office, Associate Director Technical Services

Monitoring and Processes  
10:00 – 10:25

Operational management of the Met Office Radar network

Sandeep Sanghera  

Met Office 

10:25 – 11:10 Coffee break (set up posters)
Monitoring and Processes continued   
11:10 – 11:35

Weather Radar Data Quality Monitoring using Operational Observations

Dirk Klugmann, Jordan Santillo, Robinson Wallace, Juha Salmivaara, Pekka Puhakka

Vaisala

11:35 – 12:00

Leonardo Data Centre: towards prediction maintenance

Hassan Al Sakka, Nipesh Dulal

Leonardo

12:00 – 12:25

On the use of single bright targets for monitoring the stability of dual-polarization measurables: operational scan program and stare mode observations  

Marco Gabella, M. Lainer, M. Sartori, D. Wolfensberger, M. Boscacci

Federal Office of Meteorology and Climatology MeteoSwiss

12:25 – 14:00  Lunch break (At the beginning of the break we will take a group photo)
Network Threats  
14:00 – 14:25

The Argentinian Meteorological Radar, real time RFI digital filter operational data quality impact analysis

Federico Renolfi, Roberto Costantini, Daniel Vela Diaz, Víctor Bravo INVAP S.E, Argentina

14:25 – 14:50

Monitoring and quantifying the influence of wind turbine clutter in weather radar data

Michael Frech

Deutscher Wetterdienst (DWD) 

14:50 - 15:05 Break
15:05 – 15:30

Detection of Wind Turbine Contamination with a Convolution Neural Network

Nawal Husnoo, Timothy Darlington, Sebastiàn Torres and David Warde

Met Office 

15:30 – 15:55

The New Canadian Weather Radar Network and the Impacts of Radio Frequency Interferences (RFI) on Radar Data Quality 

(online presentation)   

Qian Li, Alvin Au Duong, Hamid Nasr, Lubna Bitar and Sorin Pinzariu  

Meteorological Service of Canada, Environment and Climate Change Canada

15:55 – 17:30  Cream tea and poster session
17:30 End of day 1

 

Day 2 Programme
Day 2 Thursday 9th November 2023
Calibration  
09:30 – 09:55

Experiences with improving standard legacy calibration process for dual-pol radar receiver

Katherine Morris  

Met Office 

09:55 - 10:20

Long-term calibration of an X-band radar network in western Germany

Daniel Sanchez-Rivas, S. Trömel

Institute of Geosciences, Department of Meteorology, University of Bonn, Bonn, Germany 

10:20 – 11:10 Coffee break 
Electro/Mechanical Considerations  
11:10 – 11:35

Evaluation of the effects of different lightning protection rods on the data quality of C-Band weather radars

(online presentation)

Cornelius Hald, Maximilian Schaper, Annette Böhm, Michael Frech, Jan Petersen, Bertram Lange and Benjamin Rohrdantz

Deutscher Wetterdienst (DWD) 

11:35 – 12:00

Challenges of the Swedish weather radar network

Daniel Johnson, Günther Haase

(online presentation)

Swedish Meteorological and Hydrological Institute

12:00 – 12:25

Effectiveness of super-hydrophobic radome coating

Hiroka Aoki 

Japan Meteorological Agency

12:25 – 14:00  Lunch break 
New Technology  
14:00 – 14:25

Experience with target simulators for transmitted differential phase and absolute calibration measurements

(online presentation)

Marc Schneebeli, Philipp Schmid, Andreas Leuenberger  

Palindrome Remote Sensing GmbH

14:25 – 14:50

Validation of Vaisala SSPA X-band and C-band radar observations with Vaisala Forward Scatter Sensor FD70

Marjan Marbouti, Dirk Klugmann, Lasse Kauppinen, Pekka Puhakka, Jere Mäkinen

Vaisala

14:50 – 15:15

Maintenance experience with SSPA dual-polarization weather radar

Morihiro Sawada

Japan Meteorological Agency

15:15 – 16:00  Cream tea and poster session 
16:00 – 17:30 Met Office Radar tour 
17:30 – 18:00 Break
18:00  Dinner at the Met Office  
  End of day 2
Day 3 Programme
Day 3 Friday 10th November 2023
Calibration cont  
09:30 – 09:55

Real-time Monitoring and Calibration of Weather Radar Network using Multiple Techniques

(online presentation)

Valentin Louf

Bureau of Meteorology, Australia

09:55 – 10:30 Coffee break
Summary and discussion  
10:30 – 12:00  Panel discussion  
12:00  End of workshop 

Presentation Abstracts

Monitoring and Processes session 

Weather Radar Data Quality Monitoring using Operational Observations

Dirk Klugmann, Jordan Santillo, Robinson Wallace, Juha Salmivaara, Pekka Puhakka Vaisala Oyj

Vaisala

Monitoring the data quality of Weather Radar observations is crucial for several reasons. Firstly, it provides an insight into the quality of directly recorded observables and of derived products. This in turn can inform the subsequent use of these observables and products. As an example, the weight given to observations when assimilated for Numerical Weather Prediction models might depend on their quality. Furthermore, monitoring the data quality of Weather Radar observations might support an adaptive approach to Weather Radar maintenance. Rather than performing maintenance according to a fixed schedule, it might be possible to adapt maintenance activities to the state of individual Weather Radars. This might be supported by assessment methods utilising Artificial Intelligence / Machine Learning.

We are presenting an approach to assessing the data quality of Weather Radar observations based on operational observations of light to moderate precipitation below the Melting Layer Height (MLH). This approach uses observables from regular, operational Weather Radar observations to

  1. Select the maximum range of observations utilised for staying below the Melting Layer Height (MLH);
  2. Select and mask the subset of observations used for the assessment based on observations of radar reflectivity Z and differential reflectivity ZDR;
  3. Create statistics of the relevant observables, e.g. differential phase ΦDP or correlation coefficient ρHV;
  4. Fit theoretical curves to the observables for providing tangible metrics of the data quality.

The method so far has been implemented in Python code and tested with various Weather Radars operated by Vaisala and the FMI: two Magnetron C-band Weather Radars, a Solid-State Power Amplifier (SSPA) C-band Weather Radar, and an SSPA X-band Weather Radar. We will present the method in more details at the workshop and show results from the assessments made so far.

 

Leonardo Data Center: toward prediction maintenance

Hassan Al Sakka and Nipesh Dulal

Leonardo Germany GmbH – Neuss – Germany

In an era when industrial data is collected on a massive scale, an efficient approach to health monitoring and predictive maintenance of our machines is critical. In light of this, we at Leonardo Germany GmbH have established Datacenter, a specialized platform that gathers all data from more than 500 globally installed radar and Lidar systems. 

It acquires real-time electronic and machinery data of different radar and Lidar systems and their subsystems from internally developed software named RAVIS® and Rainbow®. This collection of data forms the base for the new approach; data-driven predictive maintenance. 

The data is gathered and systematically stored in a time series database within the Datacenter. Visualization tools have been developed to graphically display all radar subsystems, allowing us to make competent comparisons across these subsystems. Then, machine learning and deep learning methodologies are applied to the significant features distilled from this data. This process enables us to effectively detect anomalies, anticipate future system states, diagnose faults, and ultimately estimate the remaining useful life of our radar systems and their diverse subsystems.

Overall, the approach we employ represents an important advancement in redefining the radar maintenance industry, as we aim to enhance our predictive maintenance model by integrating cutting-edge machine learning algorithms with massive real time radar system data.

 

On the use of single bright targets for monitoring the stability of dual-polarization measurables: operational scan program and stare mode observations

Marco Gabella, M. Lainer, M. Sartori, D. Wolfensberger, M. Boscacci

Federal Office of Meteorology and Climatology MeteoSwiss

Previous studies have shown that the distinctive and stable echoes backscaterred by a Bright Scatterer (BS) can be used for monitoring dual-polarization measurables. In order to be “bright”, a large target with deterministic backscattering properties should be present at a near range and hit by the antenna beam axis. For the Lema C-band radar (1625 m altitude), the first identified BS is the tall metallic tower on Cimetta (1633 m altitude, 18 km range). The best stability and lowest dispersion have been observed during clear-sky days in winter months: typical daily standard deviation of the differential phase shift was less than 9° while that of ρhv, was, for instance, as low as 0.0008 on Jan. 3 and 6, 2019. In general, it was ranging between 0.0007 (Jan. 4 and 8, 2019) and 0.0050. On a weekly basis, it could be as small as 0.0014 (e.g., Jan. 3-9, despite some weak precipitation on Jan. 9). Such promising results in terms of stability have triggered the search of an additional BS. The very small dispersion of ρhv was a key factor in such search, which has eventually brought us to Monte Vada’: it is a Mountainous BS (MBS) 21 km west of Lema with even smaller dispersion and larger average values of ρhv. In summer 1999, after exchanging the Lema Calibration Unit (which includes the LNA, Noise Source and Down Converter Units), we observed a reproducible degradation (larger dispersion) of the polarimetric signature for both BS and MBS, a phenomenon we are investigating with the supplier. Meanwhile, we have started compiling a climatological summary of the polarimetric signatures. The lesson gained with the previous Calibration Unit is confirmed: the smallest (largest) standard deviation (average) values of ρhv are observed in winter, probably due to a reduced influence from propagation effects. In March 2019 and March 2020, MeteoSwiss was able to observe 3 large Wind Turbines by deploying a mobile X-band radar at ~8 km range. In 2020, in addition to PPI/RHI scans, a ~100-minute special sessions was added: fixed-pointing antenna towards the nacelle of the closest wind turbine (WT). Polarimetric radar signatures have been derived every 64 ms using 128 radar pulses transmitted every 0.5 ms (PRF = 2000 Hz). During 10-minute intervals (9375 samples) with zero rotor speed, no changes in blade pitch angle and nacelle orientation, minimum dispersion of the differential phase shift was observed. Changes between consecutive values of differential reflectivity and radar reflectivity factor were either 0 dBz or ±0.5 dBz, while ρhv was persistently equal to 1. Consequently, another lesson gained is that polarimetric signatures of a still WT are similar to those of a BS: a better stability has been observed thanks to a non-rotating antenna. We can confirm that “historical” polarimetric signatures of a BS can represent a benchmark for monitoring dual-polarization radar measurables. Finally, there is a recommendation related to ρhv quantization for the X-band radar: to use either two bytes or a Log-transformation (e.g., the operational one used at C-band). 

 

Network threats: Wind turbines, interference etc. session

 

The Argentinian Meteorological Radar, real time RFI digital filter operational data quality impact analysis

Federico Renolfi, Roberto Costantini, Daniel Vela Diaz, Víctor Bravo

Affiliation: INVAP S.E, Argentina 

This work reports the latest advances made in Argentina in the fight against radio frequency interferences (RFI) that contaminates all C-band remote sensors of the national network of meteorological radars (SINARAME).

Specifically, it evaluates the improvement in data quality achieved with the new RFI digital filter, functionality recently incorporated into the system by means of the latest software update of the Argentinian Meteorological Radar (RMA) processor.

The filter is already operational on 9 of the 12 RMA radars that today shape the SINARAME network.

 The update of the processing software and the activation of the RFI filter in all currently operational RMA radars is carried out in parallel with the network expansion project, to which another 9 RMA radars will be incorporated in the coming months.

The results obtained so far are quite satisfactory. A significant improvement in the quality of the data was observed in all the cases analyzed. However, there is still room for improvement in filter performance. Notably, the filter parameters have not yet been individually fine-tuned for each site.

 

Monitoring and quantifying the influence of wind turbine clutter in weather radar data

Michael Frech

Deutscher Wetterdienst (DWD) 

Wind turbine clutter poses a significant problem to the quality of weather radar data. An algorithm to dynamically detect WTC in radar data has been implemented in the radar signal processor of the 17 DWD  weather radars about a year ago and is operational since then.  There is a strong political push to increase wind energy production.  As a consequence of this push the 5 - 15 km range from radars had to been opened for the development of  wind mills.  In order to monitor and quantify  the existing and future WTC situation in the vicinity of weather radars, a monitoring framework has been developed. The observed WTC flags, their timestamp, location and corresponding radar moments are stored in an InfluxDB data base. WTC data from the precipitation scan and the lowest 5 elevations of the volume scan are stored (0.5° - 5.5°).  For the analysis of this data base, a python package has been developed to  generate maps with WTC and the corresponding  radar moments (such as unfiltered depolarization ratio DR, Zh,v and RHOHV) and to allow for a statistical analysis. In this contribution we show results focusing on WTC with a  WTC severity larger than 0.5.  This severity refers to WTC that is present in 50% of the time during a chosen time interval. In order to identify trends, month long, non-overlapping time series are extracted from the database.

 

Detection of Wind Turbine Contamination with a Convolution Neural Network

Nawal Husnoo, Timothy Darlington, Sebastiàn Torres and David Warde

Met Office

The growing number of wind turbines, while supporting clean and renewable power, increasingly poses a threat to the data quality of the UK weather radar network. A surface rainfall estimate is usually more accurate if the measurement is made close to the ground, but if wind turbines are present, they increase the rainfall rate estimates over significant areas. Currently, the Met Office radar data processing pipeline uses a mask of known wind turbine locations to censor the lower elevation data and uses a higher elevation measurement in those regions. This then requires a larger and more uncertain correction for the vertical profile of reflectivity, leading to a less accurate surface rainfall rate measurement. It would be beer to detect the cases when the precipitation signal is sufficiently strong (in heavy rainfall) to overcome the clutter signal from the wind turbines, and dynamically use or censor the data based on the absence or presence of contamination. Unfortunately, wind turbines are notoriously difficult to detect dynamically, because the signal from moving blades frequently resembles precipitation signals. Unlike stationary ground clutter that has a single peak at zero Doppler velocity, wind turbine signals can have an additional multi-modal Doppler component, and thus resists the use of simple Doppler notch filters. 

In this work, we make use of the bootstrap spectral dualpol estimator and a convolutional neural network to classify between rainfall, stationary ground clutter and wind turbines. As in our previous work [1], we randomly but systematically combine tiles of pure rainfall signals with tiles dry wind turbine signals to produce synthetic tiles with a range of clutter to signal ratio. We then use the trained neural network to determine when the precipitation signal is strong enough to drown out the known wind turbines. We may also be able to use this to discover previously unknown wind turbines that currently contribute some contamination to the radar rainfall product.

[1] Husnoo, N., Darlington, T., Torres, S., & Warde, D. (2021). A Neural Network Quality-Control Scheme for Improved Quantave Precipitation Estimation Accuracy on the U.K. Weather Radar Network, Journal of Atmospheric and Oceanic Technology, 38(6), 1157-1172. Retrieved Feb 7, 2022, from hps://journals.ametsoc.org/view/journals/atot/38/6/JTECH-D-20-0120.1.xml

 

The New Canadian Weather Radar Network and the Impacts of Radio Frequency Interferences (RFI) on Radar Data Quality 

Qian Li, Alvin Au Duong, Hamid Nasr, Lubna Bitar and Sorin Pinzariu   

Meteorological Service of Canada, Environment and Climate Change Canada 

 

In early 2017, the Government of Canada announced the replacement of its weather radar network. At that time, Canada’s radar network consisted of 31 radars of varying types, including two radars operated in partnership with the Department of National Defence and one owned by McGill University. A contract was awarded to buy and install a uniform network of 33 new dual-pol S-band radars by March 2023. As part of this project, the operational radar network is being expanded by one radar in northern Alberta. 

 

With the first radar successfully replaced in 2017, the Canadian Weather Radar Replacement Program (CWRRP) has completed installation of the final weather radar replacement in the 7-year project! By August 2023, 33 new S-Band radar installations were completed across the country spanning from the west coast to eastern Newfoundland. The pandemic introduced significant challenges and complexities to the project, in particular for radars located near high population centres, where construction-related outages have drawn significant public attention, such as Toronto, Ottawa, Edmonton, Vancouver, and Halifax. Despite the COVID-19 challenges, the CWRRP team worked closely with the vendor and other enablers to successfully replace seven radars from 2020 to 2022 and finished the installation of the last 2 radars in 2023. The enhancements made to the radar network have translated into better severe weather forecasting, ultimately safeguarding the well-being of Canadians and their assets. This holds particular significance as we adjust to a climate characterized by increasingly frequent and severe storms. 

 

This talk will summarize the technology of the new radar systems, as well as the advantages and challenges we have experienced, with a focused discussion on the impacts of radio frequency interference primarily from mobile networks on radar data quality. 

 

Calibration: Experiences and approaches to reflectivity and dual pol. Calibration session

Experiences with improving standard legacy calibration process for dual-pol radar receiver

Katherine Morris

Met Office

The Met Office designs, builds and maintains our radar systems in-house, including our radar processing and control system. We have recently upgraded the servers across our operational network of 15 C-band radars and moved to a 64-bit operating system. This has necessitated the rework of some existing applications, including a graphical user interface (GUI) used for the standard legacy calibration of the dual-pol radar receivers during routine maintenance. In this presentation we will share our insights from improving our receiver calibration process, highlight some of the challenges we faced and discuss our future plans.

 

Historically, the standard legacy calibration was largely manual which was time consuming and prone to error. The calibration GUI now interfaces directly to the receiver which has partly automated the process and eliminated any manual data entry. We have written the calibration GUI in Python using a generic design with different elements separated out for easier software maintenance and implementation of new features. In the future, an interface to control the test signal generator could be implemented which would fully automate the calibration.

 

The new process has been reported to be 4 times faster and has significantly streamlined the process. We can also now perform the calibration at each site’s specific transmitter frequency for a more accurate calibration. With the time gained, by default we will repeat the calibration at a different point in the receiver chain, which will hopefully provide an early warning sign of issues in the waveguide.

 

Long-term calibration of an X-band radar network in western Germany.

D. Sanchez-Rivas, S. Trömel

Institute of Geosciences, Department of Meteorology, University of Bonn, Bonn, Germany

This study, embedded in the Digitales Geosystem -- Rheinisches Revier (DG-RR) project, aims to provide a high-resolved, quality-assessed radar-based Quantitative Precipitation Estimation (QPE) and nowcasts exploiting the local polarimetric X-band radar network located in western Germany. These products will serve as input for the multi-scale geosystem model environment ("GM-RR") for local flash-flood predictions. They will provide new insights into the dynamics of extreme events.

We present the approaches for calibrating and monitoring two polarimetric X-band (wavelength ~3 cm) radars in Bonn (BoXPol) and Jülich (JuXPol). Each X-band radar covers up to 150 km at 10 different elevation angles, monitoring the region with high spatial (~150 m) and temporal (~5 min) resolution.

We generate vertical profiles (VPs) from birdbath scans and (range defined) quasi-vertical profiles (QVPs/RD-QVPs) from high-elevation scans (~18O) for monitoring the calibration (offset biases) of the differential reflectivity(ZDR )and the differential propagation phase(ϕDP ),and observe the temporal evolution of precipitation events as well. A melting layer detection scheme enables us to differentiate between hydrometeors in the solid or liquid phase, and we monitor the calibration of horizontal reflectivity(ZH )comparing the results of the relative calibration offset (RCA) method with ZH derived from specific attenuation (A). Identified seasonal variations in the ZH calibration, however, still needs further investigations. Processing routines to generate precipitation composites with state-of-the-art hybrid polarimetric estimators are adapted and further developed for the local X-band radar network. Specifically, clutter echoes have been identified by applying artificial intelligence techniques, measurements of the cross-correlation coefficient (ρHV ) have been corrected for (range-dependent) biases, and ZH and ZDR have been corrected for (differential) rain attenuation using a ray-wise optimised approach. Finally, rain rate(R) retrievals R(ZH ),R(KDP ) and R(A)have been set up and are currently evaluated for operational implementation.

 

Electro/Mechanical Considerations: Pointing, component aging etc. session

Evaluation of the effects of different lightning protection rods on the data quality of C-Band weather radars

Cornelius Hald, Maximilian Schaper, Annette Böhm, Michael Frech, Jan Petersen, Bertram Lange and Benjamin Rohrdantz

Deutscher Wetterdienst (DWD)

Lightning protection is necessary for weather radars in exposed locations, but can have negative effects on data quality. The existing lightning protection of the DWD polarimetric C-Band weather radar network consists of four vertical poles with a maximum diameter of 10 cm. During radar operation, these rods cause significant reflections of the radar beam and beam blockage with negative impacts on radar products. Reflections cause a signal in places with no precipitation; beam blockage lowers the received signal strength, resulting in an underestimation of the actual rain rate. This led to the requirement of designing a new lightning protection concept. A new lightning protection must minimize the effect on data quality on the one hand, but also provide sufficient protection from lightning strikes according to the existing regulations. Three possible lighting protection concepts are described in this paper: two using vertical rods of different diameters and one with horizontally placed rods. They are qualified through a dedicated measurement campaign by analysing resulting antenna patterns and precipitation sum products. Antenna patterns are analysed with respect to the resulting side lobe levels compared to an antenna pattern without lightning protection and one with the rods currently in use for both horizontal and vertical polarization planes. Side lobe levels are slightly increased compared to no lightning protection and decreased compared to the current conditions but are within the accepted antenna specifications. This is found for all tested new lightning protection designs. Beam blockage is substantially reduced compared to the existing lightning protection as shown by the evaluation of QPE sums. These results and some structural considerations are a solid basis to recommend the installation of four rods with maximum 40 mm diameter for all 17 radar systems of the DWD weather radar network.

 

Challenges of the Swedish weather radar network

Johnson, Daniel, et al.

Swedish Meteorological and Hydrological Institute

The Swedish weather radar network (Swerad) currently consists of twelve C-band dual polarimetric Doppler weather radars operated by the Swedish Meteorological and Hydrological Institute (SMHI). Between 2014 and 2021 the network was upgraded to dual polarimetric capacity.

Throughout the modernisation the calibration and adjustment routines evolved and in February 2020 the entire network was consistently calibrated and configured. Since November 2020 all radars performed in simultaneous transmit and receive (STAR) mode. It would turn out that the transmitters were running the magnetrons out of specification. The magnetrons eventually started to fail prematurely and forced us to degrade to H mode only and a lower duty cycle. Since June 2023 all radars are running H mode only. The medium-term goal is to slightly increase the duty cycle to allow radars to operate in STAR mode again, and the proposed long-term solution is to replace the transmitters.

About six months after the modernisation of radar Åtvidaberg (SEATV) in 2019, the intensity data started to frequently vary over time. Several endeavours later, the origin was identified as elevation play and its gearbox was successfully replaced in May2023.

 

Effectiveness of super-hydrophobic radome coating

Hiroka Aoki

Japan Meteorological Agency

The Japan Meteorological Agency (JMA) applies a super-hydrophobic radome coating that produces a water contact angle of at least 150° to dual-polarization weather radars updated since spring 2023, mainly to prevent desensitization and positive biases of differential reflectivity (ZDR) due to attenuation of wet radome surfaces (Frech, 2009). Data from radars updated before 2022 show occasional ZDR biases of 1 – 2 dB due to heavy rain falling directly above radars, and these have a non-negligible effect on hydrometeor classification. Conversely, radars with a super-hydrophobic radome coating have shown several short periods of heavy rain with intensity of around 10 mm/10 min but no similar ZDR bias.

Super-hydrophobic coatings previously developed by Japanese manufacturers have had service lives of only a few years, and have not been adopted for radome coatings due to cost-benefit consideration. JMA’s new radome coating was developed in Japan, and has a longer service life. This presentation covers problems with conventional radome coating and the performance of super-hydrophobic radome coating.

 

New Technology: SSPA, remote calibration signal sources, polarisation basis control, AI, etc. session

 

Validation of Vaisala SSPA X-band and C-band radar observations with Vaisala Forward Scatter Sensor FD70

Marjan Marbouti, Dirk Klugmann, Lasse Kauppinen, Pekka Puhakka, Jere Mäkinen

Vaisala

Precipitation is one of the essential variables in rainfall-runoff modeling. For hydrological purposes, the most commonly used data sources of precipitation are rain gauges and weather radars. However, the rain gauge networks are often not able to detect rainfall due to their limited sampling capability. In this study, the Vaisala Forward Scatter Sensor FD70 has been used instead of traditional rain gages to validate weather radar measurements. A comparison between radar reflectivity derived from observations of the Vaisala Forward Scatter Sensor FD70 and radar reflectivity measurements of two Vaisala weather radars (Vaisala X-band SSPA WRS400 prototype and Vaisala C-band SSPA WRS300 prototype) located in Finland is presented. The Vaisala Forward Scatter Sensor FD70 is located at Vantaa airport. The WRS400 radar and the WRS300 radar are located at a distance of 6.64 km and 13.61 km with headings 42° and 358°, respectively.

Five rainy days during 2021 and 2022 have been chosen for this study. Every radar is configured with a list of tasks with different elevations angles. As FD70 sensor is located on ground, it is most sensible to compare the lowest elevation radar beams to FD70 data. To assure that radar beam purely detects rain, Vaisala sounding data has been used in this study. Elevation of the lowest radar beam and melting layer were calculated. The result proved that the lowest radar beam above the FD70 was well below the melting layer for all five rain events, and purely rain events are detected with the lowest elevation angle. Measurements of droplet size distribution by FD70, converted to radar reflectivity within the instrument, were compared with radar reflectivity measurements of both the WRS400 and the WRS300. 

The comparison between the three datasets shows a high correlation between all three sensors. This will be shown in detail at the workshop.

 

Maintenance experience with SSPA dual-polarization weather radar

Morihiro Sawada

Japan Meteorological Agency

As of July 2023, the Japan Meteorological Agency (JMA) operated 20 C-band solid-state power amplifier (SSPA) dual-polarization radars, which undergo regular JMA/manufacturer inspections. Twelve of the units are part of the nationwide observation network, and involve the use of one short and three long pulses for each polarization depending on the observation range. For comprehensive and effective monitoring/calibration of these pulses, JMA has tested 1) the use of a peak-power sensor capable of automatic pulse analysis for improved efficiency in analysis of pulse width and transmitted power, and 2) calibration and examination to determine the accuracy of ranging and the bias of polarimetric parameters for each pulse width. Bird-bath scans, which are a major factor in the calibration and monitoring of polarimetric parameters, can only be applied to short pulses. Hence, low-elevation scans for weak precipitation are used in monitoring of long pulses. A JMA comparison of bird-bath and low-elevation scans showed similarly trending changes in polarimetric parameters associated with equipment maintenance, albeit with slight differences in absolute values. This presentation highlights experience with maintenance specific to SSPA dual-polarization weather radar.


Poster Abstracts

Monitoring and Processes:

Weather Radar Data Quality Monitoring using Operational Observations

Dirk Klugmann, Jordan Santillo, Robinson Wallace, Juha Salmivaara, Pekka Puhakka

Vaisala Oyj, Vantaa, Finland

Monitoring the data quality of Weather Radar observations is crucial for several reasons. Firstly, it provides an insight into the quality of directly recorded observables and of derived products. This in turn can inform the subsequent use of these observables and products. As an example, the weight given to observations when assimilated for Numerical Weather Prediction models might depend on their quality. Furthermore, monitoring the data quality of Weather Radar observations might support an adaptive approach to Weather Radar maintenance. Rather than performing maintenance according to a fixed schedule, it might be possible to adapt maintenance activities to the state of individual Weather Radars. This might be supported by assessment methods utilising Artificial Intelligence / Machine Learning.

We are presenting an approach to assessing the data quality of Weather Radar observations based on operational observations of light to moderate precipitation below the Melting Layer Height (MLH). This approach uses observables from regular, operational Weather Radar observations to

  1. Select the maximum range of observations utilised for staying below the Melting Layer Height (MLH);
  2. Select and mask the subset of observations used for the assessment based on observations of radar reflectivity Z and differential reflectivity ZDR;
  3. Create statistics of the relevant observables, e.g. differential phase ΦDP or correlation coefficient ρHV;
  4. Fit theoretical curves to the observables for providing tangible metrics of the data quality.

The method so far has been implemented in Python code and tested with various Weather Radars operated by Vaisala and the FMI: two Magnetron C-band Weather Radars, a Solid-State Power Amplifier (SSPA) C-band Weather Radar, and an SSPA X-band Weather Radar. We will present the method in more details at the workshop and show results from the assessments made so far.

 

Monitoring the French Radar Network with PIEUVRE

Jean Millet

Météo-France

In order to improve the control of raw radar data and final products quality, several tools have been developed at Météo-France over the years. The PIEUVRE project aims to bring them all together. The software has three main objectives :

  • monitoring the french radar network and radar data quality : solar control monitoring, polarimetric variables, interference.
  • producing long-term time series of statistical scores and quality indicators of final products for monthly and annual reports
  • producing and visualizing the additional data necessary for real-time data processing : adjustment factors, ground clutter identification maps, wind turbines detection and beam blockage maps.

A web-based visualization allows the users to access all data at any time, and a database provides storage for long-term time series study. Under continuous development, the software will include administration tools and more parameters to monitor.

 

Long and short term monitoring of hardware components: performance degradation observed on magnetrons and TR-Limiters on the Swiss Weather Radar Network. 

Maurizio Sartori, M. Gabella, M. Boscacci

MeteoSwiss

The radar hardware (HW) monitoring tools used by MeteoSwiss work at two different levels. The first level is a quasi-real-time visualization of the errors detected by the radar system, which in the Swiss radars occurs via Profibus protocol. A query to the Radar Control Processor occurs every three seconds, resulting in a status report, which is GUI based. A report occurs only for status parameters that meet the fault of warning condition.

At the same time, we save all status parameters defined by the radar supplier: 20 values (one per sweep) every 5 minutes (volume time), for each status parameter (around 340 in total). A GUI based application allows retrieval and display for all data since radar installation.  In addition, we collect data from ancillary systems every five minutes: HVAC, UPS, weather station, etc.  Considering the amount of data generated by a radar for each volume, the impact of this additional data is negligible.

In 2022 we exchanged two magnetrons for the first time: in April 2022 at Lema, in September 2022 at Plaine Morte. The first was running since 2011, the second one since 2015, both with a Pulse Width of 500ns, PRFs between 600Hz and 1500Hz, power around 400kW. The magnetron Lema showed signs of degradation starting since fall 2021; the measured TX-power became unstable. The norm is a deviation in power of a few tenths of a dB due to PRF, while between 2022 and 2023 this value has increased to several dB. The pulse-to-pulse correction, applied to the IQ data for each transmitted pulse, is able to compensate for these deviations; a degradation is therefore not necessarily obvious from the beginning.

The magnetron Plaine Morte started showing a similar behavior in 2023. In addition to the status parameters information, we measured the basic RF characteristics on site: a) pulse envelope and its stability using an oscilloscope in persistence mode, b) the spectral characteristics over a wide frequency spectrum (several hundreds of MHz). The pulse jitter and the presence of spurious signals in the frequency spectrum, in addition to the increased instability of the transmitted power, forced us to replace the magnetron.

The magnetron is a good example for the usefulness of long-term status parameter analysis. In the case of TR-Limiters, the degradation pattern is less predictable. In 2016, we observed an increase of the insertion loss of 0.2 dB over a period of six months (radar Weissfluhgipfel). On the other hand, in 2023 (radar Albis) we observed first an increase of the insertion loss of 15dB (!) from one volume to the next, then we observed everything going back to normal two days later, from one volume to the next. In this case, short-term status parameters become relevant, not only for the operators, but also as a feedback for the radar supplier.

 

Enhancing Radar Network Reliability with Grafana Monitoring

Maximilian Schaper, Jan Petersen, Annette Böhm, Bertram Lange, Michael Frech, Cornelius Hald, Benjamin Rohrdantz and David Michaelis

Deutscher Wetterdienste (DWD), Germany

Radar systems play a pivotal role in meteorological data collection, necessitating the optimization of data quality and minimization of system downtime. This imperative has led to the integration of comprehensive monitoring practices within the radar network at the German Weather Service (Deutscher Wetterdienst, DWD), ensuring the consistent availability of critical meteorological information and radar status information. This contribution revolves around the utilization of Grafana, an open-source visualization and analytics software, as a tool to centrally monitor all 17 weather radars of the DWD radar network at the same time.

The core objective of this work is to leverage Grafana's capabilities to process and visualize extensive radar data sets efficiently. By representing data as informative time series, the system allows for the early detection of potentially critical events. Beyond hardware monitoring, Grafana offers a versatile means to assess data integrity. The visualization of parameters such as ZDR offset and the comparative analysis of auxiliary sensors contribute significantly to maintaining high data quality. The value of these functionalities extends to both the scientific personnel and field service engineers, empowering them with actionable insights for informed decision-making.

Additionally, the poster will showcase DWD's latest innovations in monitoring through Grafana, highlighting the implementation of comprehensive quality scores across the radar network. These scores provide a consolidated evaluation of system performance and data reliability, offering a valuable tool for informed decision-making and operational enhancement.

 

Network Threats:

Statistical Analysis of Extent and Occurrence of Wind Turbine Clutter 

Pauli Anttonen, Jenna Ritvanen, Mikko Kurri, and Annakaisa von Lerber

Finnish Meteorological Institute

Wind turbines can have significant impacts on weather radars through three primary mechanisms: i) beam blocking, ii) strong reflections causing increased clutter levels, and iii) interference leading to inaccuracies in Doppler velocity measurements. Beam blocking and clutter can both hinder the radar's ability to detect severe weather events such as convective precipitation or hail. One of the current challenges is radar-based automatic warning system of Cumulusnimbus for aviation, which falsely triggers alarms due to undetected residual wind turbine clutter. Wind turbines are detectable by weather radar from distances exceeding 100 km. However, the magnitude of their impacts and the clutter generated depend on factors like turbine distance, their number and density, orographic conditions, and meteorological conditions.

This study focuses on presenting the statistical analysis of wind turbine clutter observed throughout the year 2022 using C-band weather radar data in Finland. We examine wind parks located within a radius of 100 - 200 km from five operational weather radars. The analysis considers the three lowest elevation angles and assumes a 4/3 propagation model for radio wave propagation. Wind parks are identified using aviation maps for tall obstacles. 

The study defines the occurrence and magnitude of clutter within each identified wind park, comparing the impact zone's extent to the actual wind park area. Different clutter identification thresholds are evaluated to determine an optimal value, with a final choice of 8 dBZ. The analysis provides both yearly and monthly statistics. Additionally, the study investigates the occurrence and magnitude of multipath signatures, often referred to as 'tails,' using a -6 dBZ threshold. We have also investigated Finnish composite product and gathered the statistics of the occurrence wind turbine clutter.  

Preliminary findings indicate that, on average, the impact zone's extent is 2.5 - 3 times larger than the actual wind park area. Moreover, in the azimuthal direction, the impact zone is approximately 1 - 3 km wider than the park's physical spatial extent. Additionally, we have some preliminary results of the filtering of wind turbine clutter from radar measurements at measurement volume level based on combined thresholds of different dual-pol and data quality radar variables.

 

The Argentinian Meteorological Radar, real time RFI digital filter operational data quality impact analysis

Federico Renolfi, Roberto Costantini, Daniel Vela Diaz, Víctor Bravo

Affiliation: INVAP S.E, Argentina 

This work reports the latest advances made in Argentina in the fight against radio frequency interferences (RFI) that contaminates all C-band remote sensors of the national network of meteorological radars (SINARAME).

Specifically, it evaluates the improvement in data quality achieved with the new RFI digital filter, functionality recently incorporated into the system by means of the latest software update of the Argentinian Meteorological Radar (RMA) processor.

The filter is already operational on 9 of the 12 RMA radars that today shape the SINARAME network.

 The update of the processing software and the activation of the RFI filter in all currently operational RMA radars is carried out in parallel with the network expansion project, to which another 9 RMA radars will be incorporated in the coming months.

The results obtained so far are quite satisfactory. A significant improvement in the quality of the data was observed in all the cases analyzed. However, there is still room for improvement in filter performance. Notably, the filter parameters have not yet been individually fine-tuned for each site.

 

Unwanted Effects in Operational Data from the SHMU Weather Radar Network

Luboslav Okon, Marian Jurasek, Ladislav Meri, Jan Kanak, Jozef Ulicny, Mikulas Lendak

Slovak Hydrometeorological Institute 

Since 2015-16, SHMU has been operating four identical Leonardo METEOR 735 CDP10 C-band dualpolar radars. The data from these radars are crucial for the detection and nowcasting of hazardous weather phenomena and are used in many fields, including civil protection, hydrology, aviation, etc. 

Despite state-of-the-art technology and advanced signal processing, the operational data from these radars are affected by several problems that result in either false echo detection or total or partial loss of useful information. These problems can have a significant impact on the quantitative application of the data as well as on the automatic detection of various weather phenomena. Unwanted effects in the data presented in this poster include RLAN interference, wind farm clutter or second-trip echoes. A special section is dedicated to the effects of TR limiter aging on volumetric scan and bird-bath data.

 

Calibration:

On the use of a UAV-mounted corner reflector as a calibration target for an X-band polarimetric weather radar 

D. Wolfensberger, M. Lainer, S. Monhart, R. Gugerli, Z. Schauwecker

MeteoSwiss

Radar calibration is traditionally performed with reference targets such as corner reflectors or calibration transponders, which are either mounted on a tall building or pole, or suspended from a tethered balloon. The first approach makes it difficult to avoid interferences from ground-clutter, whereas the second does not ensure a stable position of the target.  A novel approach that has been proposed in the literature is to suspend a calibration target directly from an unmanned aerial vehicle (UAV) or drone, by using a rope. Unfortunately, this is currently forbidden in Europe without a special authorization that requires a considerable amount of time and energy to obtain. To address these challenges, we propose an approach that makes use of a corner reflector directly mounted on a drone. In this case, no special requirement is required as long as the payload specifications stay within the limits given by the UAV manufacturer.

By using a drone, it is possible to monitor at any time the precise location of the radar target, thanks to the accurate drone IMU and GPS. The RCS of the drone is much smaller than that of the corner reflector, which makes its impact on the radar signature small . Overall, this approach has the advantages of mobility and flexibility, allowing for rapid deployment to diverse locations and varied altitudes. Thanks to the controlled drone velocity, it is also possible to evaluate the accuracy and precision of Doppler observables.

The proposed setup employs a lightweight yet robust corner reflector, engineered to ensure a consistent and accurate RCS. This corner reflector is mounted on a DJI Matrice 300 commercial drone. This work first details the choice of materials, geometric considerations, and mounting mechanisms employed to optimize the corner reflector's performance for radar calibration. Furthermore, this study presents the calibration procedures and data processing methodologies adopted to derive calibration coefficients for the weather radar system. We also discuss the challenges associated with the drone's flight dynamics and their impact on calibration accuracy.

Preliminary results from a field experiment on an X-band weather radar demonstrate the feasibility and efficacy of the corner reflector-mounted drone as a calibration target. Finally, the stability of the target signature is thoroughly assessed and conclusions regarding potential future uses are made.

 

Electro/Mechanical Considerations:

Integration of four X-band radars into the existing C-band network

Nikolaos Antonoglou, Manuel Werner, Ulrich Blahak, and Kathleen Helmert

Deutscher Wetterdienst, Forschung und Entwicklung, Germany

The combination of C-band and X-band radar networks has become increasingly popular in recent years, as it offers significant benefits for a range of applications. C-band and X-band radars use different frequencies, with the prior operating at around 5.6 GHz and the latter operating at around 9.3 GHz, leading to wavelengths of 5.3 and 3.2 cm, respectively. This means that the two types of radar have different characteristics and are suited to different applications.

When combined, the two radar types complement each other's strengths and weaknesses. C-band radars can cover a larger area and provide a general overview of the environment being monitored, while X-band radars can provide detailed information about specific targets within that area. By combining these two types of radar, it is possible to obtain a more accurate and complete picture of the environment being monitored, with both high resolution and extended range. On the other hand, the combination of different-frequency radars in one processing chain brings several challenges. The backscattering profiles are different in each system, resulting in distinct observation patterns. Moreover, each frequency is subject to different attenuation rates, according to which, the appropriate corrections need to be applied.

The German Weather Service (Deutscher Wetterdienst – DWD) operates a network of 17 C-band radars and in the following months will start to install four additional X-band systems in the urban areas of Karlsruhe, Nürnberg, Halle, and Bremen. The goal is to extend the coverage of the network and improve the early detection of thunderstorms that could potentially cause flash floods. As an initial step, we aim at quantifying the specific attenuation for all types of precipitation and the two frequencies. This process is fundamental for the realistic correction of the observations, particularly for higher-frequency signals that are more heavily impacted by attenuation.

 

Upgrading the DWD weather radar with a new STALO and an internal test signal generator

Jan Petersen, Bertram Lange, Benjamin Rohrdantz, Maximilian Schaper, Michael Frech,

DWD Hamburg and Observatory Hohenpeißenberg

In this poster we describe the steps  that led to  the cost-efficient replacement of the STALO and the internal test signal generator in the DWD weather radar by  a commercially available TSG.  After a  tender, the Anapico APSIN6G. The replacement started November 2022 and was completed in August 2023.   The replacement became necessary because of the lack of spare parts for the STALO and  technical problem of the internal TSG reaching end of life time (the introduction of the DWD radar systems started in 2010 and was completed in 2015). Prior introduction into the radar network, comprehensive tests were carried out. The TSG was tested in a climate chamber in order to measure the output power stability and the phase noise as a function of temperature.  The setup with the two new TSGs was then incorporated into the research  radar Hohenpeißenberg in order to demonstrate the full functionality by checking the radar coherence using clutter target scans. The internal TSG plays a crucial role to monitor calibration of the receive chain including the TR-limiter. Both the new STALO  and the ITSG are now  identical devices. So in case the STALO fails,  it can be easily replaced by the ITSG improving the operational reliability.