Abstracts

Fog and blue content of artificial light mediate bird attraction during nocturnal migration

Carrie Ann Adams, Miguel F. Jimenez, Kyle G. Horton

Birds Tracking radar Applied radar aeroecology Spatial temporal movement processes Animal behaviour

Migratory birds are known to be attracted to artificial light, contributing to collisions with buildings that kill millions of birds each year. However, birds can show diverse behavioural responses when they encounter light during nocturnal migration, including avoidance. Identifying the light characteristics, weather conditions, and other factors that change birds’ behavioural responses is key to minimizing the disruption of their migrations as artificial light at night continues to increase and older lighting technologies are replaced with blue-rich LEDs.

We conducted an experiment in Texas to understand how atmospheric conditions and the spectral properties of artificial light mediate bird response to light emitting diodes (LEDs). We used a BirdScan MR1 radar to measure migration traffic rate and flight height under three different light treatments – a blue-rich LED, a neutral LED, and a no-blue LED – compared to a dark treatment. We found no change in migration traffic rate among the treatments, possibly because attraction or avoidance occurred at a scale smaller than the radar’s 1-km detection radius. When we analysed flight height using a quantile regression, we found an important interaction between aerosolised particles in the atmosphere (primarily fog and mist) and light treatments. On clear nights, the lowest-flying migrants flew higher when the blue-rich or neutral LED was turned on, indicating avoidance of the light. On foggy nights, the lowest-flying birds flew lower during all treatments than during the dark treatments, indicating light attraction.

On both foggy and clear nights, migratory birds may expend substantial energy if they change their flight height every time they encounter a bright light. Using LED luminaires that emit less blue can reduce the disruptive effects of avoiding artificial lights, but all LEDs are likely to attract birds on foggy nights. Reducing light pollution is crucial for protecting avian migrants from collisions.

WaterFowl Alert Network: Predicting waterfowl distributions within major flyways of the USA from models trained on weather surveillance radar data

Jeffrey Buler, Tristan Bond, Shane Feirer, Joseph Gendreau, Matthew Hardy, Jaclyn Smolinsky, Maurice Pitesky

Birds Weather radar Applied radar aeroecology Spatial temporal movement processes

Wild waterfowl pose a risk for transmission of avian influenza to commercial poultry, particularly during winter months in the United States. We developed the WaterFowl Alert Network web application; an automated early-warning system to inform poultry industry stakeholders when there is significant waterfowl activity near their facilities. The application produces “near-real-time” daily continuous surface maps of predicted waterfowl density at 250 x 250 m resolution within the Atlantic and Mississippi Flyways of the USA (31 of 50 states). The predictions are based on an ensemble of 2043 localized Boosted Regression Tree models within 489 overlapping modeling frames (400 x 400 km in size) trained on data from 79 weather surveillance radars of waterfowl densities at the ground as birds depart from diurnal roosting habitats across three winters (November 1 through March 31, 2019 - 2022). Predictors for the models include remotely sensed measures of soil moisture, precipitation, snow cover, temperature and land cover composition, as well as time and proximity to waterbodies. We will discuss novel ecological insights from the spatial and temporal variability in the direction and strength of waterfowl associations with environmental predictors within and between flyways. We will also discuss the features of the web application for visualizing data by users.

Active navigation and meteorological selectivity drive insect migration patterns through the Levant

Yuval Werber, Elior Adin, Jason Chapman, Don R. Reynolds, Nir Sapir

Insects Tracking radar Methodological advances Spatial temporal movement processes

Insect migration is crucial to many natural processes and human activities, yet large-scale patterns remain poorly understood. On the Mediterranean’s eastern shores lies a 70-km wide stretch of hospitable habitat between the sea and the Arabian Desert, which we term the Levantine Corridor, extending ~400 km south from Turkey to the edge of the Sahara. We deployed 7 biological radars over 8 years, recording 6.3 million individual large insects (>10 mg), revealing an important migration route at the nexus of three continents, with over 700 million large insects estimated to cross annually. Insects showed strong migratory directionality differing from downwind direction in spring and autumn, with mass migrations separated by periods of weaker movements. Migration intensity strongly depended on weather, with insects preferentially migrating in seasonally beneficial tailwinds and warmer temperatures. Surprisingly, a comparison with insect migration flows in Europe suggests that Levantine insect fluxes are substantially lower than those reported there, challenging the conjecture that the Levantine Corridor acts as a funnel for insect migration as reported for birds. This raises interesting ecological, behavioural and evolutionary questions regarding the drivers of this difference. Lower insect quantities in the Levant may point at substantial mortality of migratory insects en route, or broad-front migration involving long sea crossings over the Mediterranean, detouring the corridor. Alternatively, given insects’ limited sensory capabilities and reduced ability to maintain course in detrimental weather, funnelling may be counter productive, increasing risk and causing delays. The study reveals an unexplored insect migration route with implications for food-webs, pollination, disease transmission, pest outbreaks, and species invasions across West Asia, East Europe, and Northeast Africa.

Assessing migratory bat movements across seasons and years in Central Europe using vertical-looking radar and acoustic monitoring

Silvia Giuntini, Janine Aschwanden, Damiano Preatoni, Fabian Hertner, Birgen Haest, Baptiste Schmid

Bats Applied radar aeroecology Biodiversity monitoring

Bat migration is an ecologically important yet still poorly understood phenomenon. This is partly because tracking these migratory movements is difficult, given bats’ nocturnal behaviour and their frequent high-altitude flights. This study presents the first radar-based assessment of multi-year migratory bat phenology in Europe, using vertical-looking radar data collected on the Swabian Plateau in Germany between September 2019 and December 2022. Bat activity was consistently low in winter and rose gradually from March onwards, reaching a peak between July and September. Across all years, pre-maternity migration started between late February and mid-March, while post-maternity migration concluded between late October and mid-November. We estimated peak nightly abundances between 1159 and 2473 bats per km, with the highest peak recorded on 4 July 2022. Correlations between radar-derived nightly bat numbers and simultaneous acoustic recordings ranged from 0.47 to 0.70 during the pre-maternity season, and from 0.14 to 0.71 during post-maternity migration. Both monitoring methods showed peak bat activity in summer, with smaller increases in September and October. The radar, however, detected substantially more bats overall. These findings show how vertical-looking radars can quantify and describe seasonal variation in high-altitude bat movements. With strategic future radar deployments and the use of existing historical datasets, our understanding of migratory bat seasonality, routes, and intensity could improve considerably and help support effective conservation protocols.

Simulation-informed migration forecasting to improve and understand predictions of rare migration peaks

Bart Hoekstra, Emiel van Loon, Bart Kranstauber, Cecilia Nilsson, Judy Shamoun-Baranes

Birds Weather radar Methodological advances Applied radar aeroecology Spatial temporal movement processes

Dynamic aeroconservation relies on timely and accurate forecasts of migration peaks, for example to support curtailment decisions in wind energy. While the broader seasonal dynamics of bird migration can be predicted reasonably well with the use of weather radar data, accurately forecasting the timing and intensity of migration peaks remains challenging. A key reason is the rare nature of these events, which makes it difficult for statistical and machine learning approaches to learn the conditions under which migration peaks occur. One way to address this limitation is to guide machine learning models towards a more biologically realistic environmental search space, informed by theoretical movement models.

We present a hybrid migration forecasting framework that integrates outputs (trajectories and departure locations) from an individual-based model (IBM) with a machine learning (ML) model to predict time series of vertical profiles of bird migration, measured with weather radar. By using the wind-based simulated migratory movements of the IBM to sample environmental information, the ML model can learn from the biologically relevant environmental conditions in a dynamic way. Moreover, the framework enables a clear separation of environmental conditions experienced during stopover, en route towards the radar, and in the vicinity of the radar itself.

We show that the hybrid model performs comparably to existing models when using conventional performance metrics (RMSE and R²), but performs better when using metrics emphasizing correct prediction of peak events. The latter are more relevant for dynamic aeroconservation in the context of wind energy curtailment. In addition, by explicitly separating environmental conditions at the stopover sites, en route to and above the radar, the model allows us to disentangle where specific environmental drivers matter most.

We use methods from interpretable ML, primarily SHAP values, to provide insight into how environmental variables interact to shape rare migration peaks in our hybrid model. This is crucial for high-stakes applications like wind curtailment, where opening the “black box” helps build confidence and trust among a diverse range of stakeholders.

Directional patterns of nocturnal bird migration over the Netherlands: Insights from combining bird radars

Bart Kranstauber, Hans van Gasteren, Bart Hoekstra, Maja Bradarić, Johannes de Groeve, Jens van Erp, Judy Shamoun-Baranes

Birds Weather radar Tracking radar Spatial temporal movement processes

Research on population-wide bird migration has mostly focused on abundance and altitude patterns. However, insights into velocities and directional patterns, and how they vary spatiotemporally, can further improve conservation efforts and reduce aerial human-wildlife conflicts. Here we examine migration directions on a landscape scale using a dense network of twelve dedicated bird tracking radars and a weather radar in the Netherlands. We compare directional changes between radars located inland and at sea to uncover differences in migratory behaviors.

Using three years of nocturnal data across both migration seasons, we find that inland radars exhibit the highest similarity in directions, including a strong correspondence with the regional patterns observed by a nearby weather radar, despite the fact they record migration at different scales and altitudes. Radars offshore show more variation, the observed migration directions change more during the night, and seasonal migration directions are not directly opposing. During the night, there is general rotation of migration directions towards the east, land inwards, in both spring and fall. This rotation is most prominent offshore.

These results suggest that coastal regions, which can act as ecological barriers, experience more variable migration patterns with greater local variation in directions, while inland migration is more uniform. Understanding these differences is important for optimizing local conservation measures, highlighting where and when tailored conservation efforts are most desirable.

Is southward return migration viable? Resolving the mystery of autumnal insect migrations from Northeast China

Boya Gao, Sun Wei, Feng Hongqiang, Huang Jianrong, Gao Yuebo, Hu Gao, Jason Chapman

Insects Other radar Spatial temporal movement processes Biodiversity monitoring Animal behaviour

Whether migratory insects in Northeast China undertake a successful southward return migration in autumn is a long-standing ecological question. While trajectory analyses based on dominant wind fields have historically indicated a low probability of success, potentially trapping populations in a "Pied Piper" scenario, empirical evidence suggests a more complex reality. For instance, observations of high-altitude insect passage over islands in North China during late summer and early autumn hint at southward movements, though the origin of these migrants remains unconfirmed. This study presents a novel investigation to bridge this critical knowledge gap. To address our core question, we will utilize continuous monitoring data from vertical-looking entomological radars (VLRs) located in Northeast China from 2023 to 2025. We will first analyze this dataset to characterize the phenology, density, and flight behaviors (e.g., track direction, orientation) of autumnal insect migrants. Subsequently, we will integrate these findings with trajectory modeling and meteorological analysis to identify the specific synoptic weather systems that facilitate or constrain the formation of migration corridors. This integrated approach will not only resolve the existing controversy but is also expected to significantly advance our understanding of the drivers and adaptive strategies governing long-distance insect migration systems.

What goes up must come down: Using weather surveillance radar data to forecast migratory birds banded on the ground

Carolyn Burt, Sara Morris, Siddharth Chidambaram, Kyle Horton

Birds Weather radar Ecological forecasting

Ecological forecasting is a powerful framework to predict future states of ecological systems, including species distributions and supporting proactive conservation, management, and policy decisions that exist in a rapidly changing world. Forecasting efforts span a diversity of systems and organisms, yet many ecological phenomena, especially those involving highly mobile species, remain difficult to predict. Migration, for example, shapes population dynamics, species interactions, ecological services, and disease spread across terrestrial and aerial environments. Migratory pulses of birds observed in airspaces and stopover habitats are often studied separately, however the connection between these two phases of migration remains largely unquantified. Here, we test the hypothesis that weather surveillance radar (WSR) measures of nocturnal migration activity can predict the number of birds captured at a long-term bird banding station, thereby bridging aerial and terrestrial perspectives of migration. Using data from two WSR sites—KBOX (Boston, MA) and KGYX (Portland, ME)—and bird banding records from Appledore Island, Maine (1996–2023), we examined how radar-derived measures of migration aerial activity forecast terrestrial bird banding captures. Generalized additive models incorporating radar reflectivity, wind conditions, temperature, and visibility explained 54.3% of the variance in spring captures and 36.5% in fall captures. These findings underscore the potential for integrating aerial and terrestrial monitoring to enhance migration forecasts, with implications for conservation strategies such as targeted "lights out" initiatives and stopover habitat management. Our study provides a novel framework for linking large-scale atmospheric migration data with on-the-ground monitoring, offering a scalable pathway to forecasting the connection between broad atmospheric data to ground-level ecological processes that is applicable across migratory taxa.

Animal niches in the airspace

Cecilia Nilsson, Judy Shamoun-Baranes, Dara Satterfield, Sissel Sjöberg, Emily B. Cohen

Bats Birds Insects Weather radar Other radar Biodiversity monitoring Animal behaviour

The lowest part of the atmosphere is full of life, and we now recognize the airspace as habitat for an abundance of flying animals and as a crucial arena for a wide range of behaviors. Technological advances, including expanded use of radar, has increasingly shown how flying animals use the aerial habitat. This has enabled us to start to ask questions about the environmental patterns and ecological processes that shape habitat niches in the air, including energetics, biotic interactions, and risk due to growing anthropogenic conflicts. In our recent paper we identified some of the environmental conditions and biological interactions that influence where different groups of animals occur in the airspace throughout their life cycles. I will discuss our outline for an ecological framework for habitat use in the air. This can be used to advance understanding of how different properties of the airspace shape fundamental aerial habitat niches, and how biotic interactions influence the realized niches. Applying these fundamental and realized niches concepts can help characterize and investigate animal distributions in the airspace, including altitudinal space. This is crucial to understanding life in the air and its responses to future change.

Seeing double: Concurrent dual-frequency observations of aerial migration with CSU-CHILL

Dylan Osterhaus, Patrick Kennedy, Kyle Horton

Birds Insects Weather radar Methodological advances Applied radar aeroecology

Nearly all radars used in aeroecology operate at a fixed wavelength or frequency band (e.g., X-, C-, or S-band). Radar frequency fundamentally influences how biological targets are detected and quantified, yet relatively few studies have explicitly examined how frequency shapes measurements of birds and insects in weather surveillance radar applications. To address this gap, we leveraged the CSU CHILL dual-frequency, dual-polarization radar (X-band and S-band) located in Greeley, Colorado to collect concurrent measurements of reflectivity, radial velocity, and polarimetric variables during spring migration in 2024. Data were collected on four nights (May 11, May 12, May 18, and May 19) using Plan Position Indicator (PPI) and Range Height Indicator (RHI) scans acquired every 3–5 minutes throughout the night. RHI scans were oriented both in the head-on direction of movement, as inferred from radial velocity, and 90° to the west. We observed pronounced frequency-dependent differences in reflectivity factor, highlighting the importance of converting biological quantities to reflectivity rather than comparing absolute values across radar systems. Finally, we compared flight-height estimates derived from PPI-based vertical profiles of reflectivity with those obtained from RHI scans, providing insight into how radar frequency and scan strategy influence estimates of aerial biomass distribution.

Large scale insect migration: the role of seasonal and latitudinal pressures

Elske Tielens, Birgen Haest, Silke Bauer

Insects Spatial temporal movement processes

Insect long distance dispersal exceeds the migration of any other taxonomic group in the world in both abundance and biomass, and yet general patterns in this seasonal insect movement are not well described. While small scale radars have long allowed the study of flight directionality, orientation, and drift, research into this behavior has mainly focused on single sites and a subset of key taxa. Here, we use a network of 21 specialized biological radars, distributed across Europe and spanning 17 degrees in latitude, to test a simple set of hypotheses around seasonal movement. In particular, we focus on understanding flight direction and common orientation as a way to assess flight strategies, and we evaluate how these vary across season and latitude. We find that insects migrate in seasonally preferred directions. Spring migration occurs in a broad range of directions, while fall migration is strongly southward. During summer, flight occurs in all directions. Next, we assess how insects make use of wind assistance during high altitude flight. During spring and summer, wind assistance carries insects in a broad range of directions, and they are not selective in their choice of winds. However, during fall migrants fly only on a subset of available days when winds occur in seasonally preferred directions. Lastly, we find that flight directionality and common orientation are strongly site specific. At southern latitudes, directional flight is weaker in both spring and in fall than at high latitudes. Similarly, common orientation is also weaker at lower latitudes than at northern latitudes. Together, these results suggest that the pressures to migrate differ across space and time.

Bird migration over the North Sea: Integrating radar and camera data to assess avian interactions with offshore wind farms

Erik Fritz, Sam Ordeman, Joep Breuer

Birds Tracking radar Applied radar aeroecology Animal behaviour

As offshore wind energy development intensifies along the Dutch coast, understanding the ecological implications for migratory birds is critical for the design of nature-inclusive wind farms. The BirdSafe project addresses this challenge by deploying a dual-sensor approach, combining macro-scale radar and micro-scale camera systems to monitor avian migration and behaviour in relation to offshore wind farms.

This research leverages continuous radar data, collected from multiple sensors positioned at different locations in (and adjacent to) the North Sea, to quantify migration intensity, timing, and directional patterns at a broad spatial scale. Complementary thermal camera systems, installed on 7 turbines within the Hollandse Kust Zuid wind farm, provide detailed observations of bird behaviour in the immediate vicinity of turbines, capturing micro-avoidance manoeuvres and potential collision events. Up to 18 cameras are installed per turbine, giving a detailed 3D view of bird-turbine interaction. Within BirdSafe the focus is on nocturnal autumn migration, but the data could also be used for other research.

The core objective is to analyse and synthesise these heterogeneous datasets to identify migration peaks, quantify bird traffic rates, and elucidate behavioural responses to turbine presence. A central research focus is whether radar and camera systems yield consistent narratives regarding migration dynamics and turbine interactions.

This multidisciplinary effort offers a unique opportunity to advance the ecological understanding of offshore wind impacts and to inform mitigation strategies for avian conservation. The outcomes are expected to contribute to the development of evidence-based guidelines for sustainable offshore wind farm design, balancing renewable energy goals with biodiversity protection. Furthermore, the extensive dataset has large potential for the validation of collision risk models in an offshore environment.

Integrating BirdFlow into BirdCast migration alerts: Towards species-specific risk

Ethan Plunkett, Yangkang Chen, Miguel Fuentes, Yuting Deng, Daniel Sheldon, Benjamin Van Doren, Adriaan Dokter

Birds Weather radar Methodological advances Applied radar aeroecology Spatial temporal movement processes

Weather radar networks have revolutionized our ability to observe continental-scale movements of migratory birds. We use the NEXRAD network to issue real-time alerts for high migration events via the BirdCast platform. Alerts can target different actionable events, such as 'lights-out' actions to reduce window collisions or warning pilots of bird strike risks. A key limitation of radar data is that it is species-agnostic, whereas risk levels often depend on species composition. In North America, migratory warblers in the Parulidae family are particularly sensitive to light attraction and collisions, while for pilots, larger species pose most concern.

Here we propose using BirdFlow models to adjust radar alert levels and improve risk assessment. BirdFlow is a probabilistic modeling framework that infers population-level movements from weekly species distribution maps produced by the participatory science project eBird. BirdFlow models previously required tuning with high-resolution individual tracking data, unavailable for most migratory species. We introduce a new model-tuning framework that removes this dependency and generalizes population-level movement inference to hundreds of migratory species. This framework extends BirdFlow by enabling tuning and validation with multiple complementary data sources, including GPS tracks, banding recoveries, and radio telemetry data from the Motus Wildlife Tracking System.

We evaluate the generalizability of this approach by: (1) assessing predictive performance compared to null models; (2) validating the biological plausibility of BirdFlow outputs by comparing route straightness, stopover frequency, and migration speed between simulated and empirical tracks; and (3) comparing models tuned on species-specific movement data to those tuned using hyperparameters transferred from other species. Results show that BirdFlow achieves biologically realistic performance even over prediction horizons of several months and thousands of kilometers.

Finally, we demonstrate how these validated models can enhance BirdCast alerts, illustrating differences in frequency and timing of alerts for light-sensitive species relevant to lights-out actions and large-bodied species relevant to aviation safety. Our study is accompanied by a first release of tuned BirdFlow models—providing a foundation for more accurate predictions in applications including conservation, disease surveillance, aviation, and public outreach.

Long-term trends in migratory landbird stopover distribution following 13 years of changes in the contiguous United States

Fengyi Guo, Jeffery Buler, Adriaan Dokter

Birds Weather radar Applied radar aeroecology Spatial temporal movement processes Biodiversity monitoring

The dramatic decline of migratory landbirds in North America has raised substantial conservation concerns. While migratory landbirds face multiple and interacting threats throughout their annual cycle, research and conservation efforts have historically been biased towards breeding and wintering grounds, leaving migration poorly understood. Recent advances in radar aeroecology now enable large-scale monitoring of habitat use during migration. We leveraged 13 years of standardized monitoring data from the Next Generation Weather Radar (NEXRAD) system (2013 – 2025) to map annual stopover distributions (in spring and autumn) and long-term population trends of migratory landbird across the contiguous United States. We detected considerable spatial heterogeneity in stopover density trends within and across radar domains, with widespread declines and particularly strong negative trends in the agricultural Midwest. Our next step is to integrate these spatial trend maps with histories of land-use change, climate change, and artificial light at night. We aim to quantify the relative contributions of multiple stressors during migration to population trends. This work will provide one of the first continental assessments of how threats encountered during migration shape long-term population trajectories, offering new insights into identifying the drivers of migratory bird population declines in North America.

Simulation hits reality: How simulated wildlife strikes correspond to collisions observed at an air base

Hans van Gasteren, Isabel Metz

Birds Tracking radar Applied radar aeroecology

Collisions between wildlife and aircraft pose an ongoing issue in aviation. Over the past years, an increasing number of airports have installed avian radar to increase situational awareness of wildlife movement in their vicinity. These tools can serve, for example to identify hotspots of current wildlife activity where deterring is required. They also can support retro perspective analyses, for example to evaluate the wildlife situation at the time of a reported strike, measure the number of runway crossings and near-misses or to assess the success of wildlife hazard management measures. Simulation studies including wildlife movement observed by the radar have been used to assess and quantify the risk of wildlife strikes for aircraft operations and to evaluate effectiveness of collision avoidance methods. In this presentation, we analyse, how metrics directly obtained from the radar, such as wildlife abundance runway crossings and near-misses are reflected in reported strikes as well as number of strikes as determined in simulations. Since wildlife aims at avoiding aircraft in reality, it is expected, that the real number of strikes is lower than in simulations, since the latter do not consider escape behaviour of wildlife. Furthermore, we assess how bird abundance around the airport area is connected with the risk of wildlife strikes to evaluate whether an increased number actually reflects a higher risk. Implications for modelling approaches of escape responses in simulations will be derived.

Detecting trends in flying insects from long-term cloud radar observations

Heike Kalesse-Los, Moritz Lochmann, Birgen Haest, Teresa Vogl, Maximilian Maahn

Insects Other radar Methodological advances

The abundance of flying insects is an important indicator of biodiversity, yet their biomass has declined over the past few decades. Still, insects are vital for food security and the functioning of ecosystems, which highlights the need for reliable monitoring methods. In the field of atmospheric science, flying insects are continuously observed by cloud and weather radars, albeit usually as unwanted clutter. Techniques that filter out insects from radar data can be used to extract the insect signal for further analysis. In this study, we analyse data from a 35 GHz Doppler cloud radar. The main radar data set comprises more than eight years of continuous observations in Lindenberg, Germany (2017–present).

The CloudnetPy target classification tool is used to differentiate insect echoes from hydrometeors. The insect radar spectra are then investigated further using the PEAKO and peakTree radar Doppler spectra peak-finding algorithms to derive the number concentration of flying insects at each range gate. This work will provide new insights into insect flight behaviour in different synoptic conditions and during the diurnal cycle, offering high temporal (approx. 5 s) and vertical (approx. 30 m) resolution. The length of the Lindenberg data set further provides an opportunity to assess long-term variability and to derive preliminary trends in airborne insect abundance from sustained radar observations. Adopting this approach to widely available operational meteorological radar data sets would substantially increase the quantity of high-quality observations available for entomological research. This could facilitate large-scale, long-term evaluations of insect populations in various regions.

Long-term phenology shifts and barrier-driven weather selective departure decisions in nocturnal bird migration over South Korea

Hwayeon Kang, Chang-Yong Choi

Birds Weather radar Spatial temporal movement processes Biodiversity monitoring

The Korean Peninsula serves as a critical bottleneck along the East Asian-Australasian Flyway, where migratory birds concentrate before and after crossing surrounding seas. Despite advances in radar aeroecology, quantitative studies on bird migration using weather radar remain limited in East Asia. South Korea operates a dense network of dual-polarization weather radars, allowing comparisons across sites with different ecological contexts while minimizing confounding factors such as latitude and climate. This study addresses two questions: (1) whether long-term shifts in migration phenology have occurred over the past two decades, and (2) whether birds departing to cross the sea exhibit stronger selectivity for favorable weather conditions compared to others.

We analyzed weather radar data from 10 stations across South Korea spanning 2007–2024, spring (Mar–May) and autumn (Sep–Nov) migration periods. Biological echoes and migration traffic rates (MTR) were calculated for the first hour after sunset, using the vol2bird algorithm. Migration timing was quantified as the dates when 5%, 50%, and 95% of cumulative seasonal migration traffic had passed. For the period of 2020–2024, we examined relationships between MTR and meteorological conditions, including temperature, wind-related variables, cloud cover, and short-term changes. Stations were categorized as sea-crossing (n=5) or non–sea-crossing (n=5) according to the autumn migration context.

Spring migration phenology showed advancing trends at 3 of 10 stations (7–10 days per decade, p<0.05), while autumn showed no significant trends at most stations, with only one site exhibiting a delayed pattern. The weather sensitivity of migration, especially with the wind variables, was stronger during autumn migration, when birds face an open-sea ecological barrier. In spring, temperature dominated migration intensity at all stations regardless of location, with minimal influence of wind conditions.

These findings suggest that birds facing ecological barriers apply stricter weather criteria before departure, particularly wind conditions critical to sea-crossing efficiency. Together with the observed long-term shifts in migration timing, our results highlight the value of East Asian weather radar networks for large-scale, long-term monitoring of avian migration dynamics across an understudied flyway.

Radar-controlled Shutdown on Demand to minimize collisions of migrating birds in a coastal wind farm

Jente Kraal, Koosje Lamers, Nienke Kwant-Heida, Jonne Kleyheeg-Hartman

Birds Tracking radar Applied radar aeroecology

Twice a year, millions of birds migrate through the Netherlands on their journey between wintering and breeding grounds. The rapid development of wind farms poses new dangers to migrant birds due to collision risk. Coastal areas are especially important for migrating birds. One such area is the ‘Tweede Maasvlakte’ in the Netherlands, a large industrial area on an artificial peninsula that stretches out into the North Sea. Birds following the coast on their migration will pass through this area. Moreover, because it is a peninsula, the Tweede Maasvlakte serves as an important departure and arrival location for birds migrating between mainland Europe and wintering areas in the United Kingdom. In 2022, a new wind farm (Windpark Maasvlakte 2) consisting of 22 wind turbines was built on the coastal protection structures bordering the Tweede Maasvlakte. To minimize the number of collisions among nocturnally migrating birds, a mitigation measure was implemented to curtail all wind turbines during peak migration events.

Peaks in migration activity at this site are caused by mass migration events of passerines, often at night. Therefore, curtailing the wind farm during these moments is likely to have the greatest impact on the number of collision victims. To detect such peaks in migration, a 3D bird radar Max® (Robin Radar Systems), was installed close to the wind farm. This radar allows us to detect and track birds in the area and thereby quantify migration intensity and identify migration characteristics, such as flight altitude and direction. These live measurements by the bird radar are used to automatically induce shutdown of the wind farm. Here, we present our analyses of the migration patterns at this site, we discuss the methodology used to determine the curtailment rules, and we present our current procedures for identifying mass migration and inducing curtailment. Moreover, we discuss practical considerations like differentiating mass migration from shorter bursts in movements of local birds. Finally, we discuss the effectiveness of this radar-controlled shutdown on demand mitigation.

The mechanisms of field flight behaviors in nocturnal insects in the Southern North China Plain, China

Jian Ma, Jianrong Huang, Guoping Li, Caihong Tian, Gensong Wang, Hongqiang Feng

Insects Applied radar aeroecology Animal behaviour

Approximately 9.3 trillion nocturnal insects undertake long-distance migration over the eastern plains of China annually, with a total biomass of 15,000 tons, exerting profound impacts on regional ecological balance and agricultural production. However, the mechanisms underlying their migratory behaviors (e.g., takeoff and landing timing, layered altitude) and their response patterns to meteorological conditions remain poorly understood.

Based on nighttime monitoring data from a vertical-looking entomological radar (VLR) in Xinxiang City, Henan Province, spanning 2022–2024, this study systematically extracted migratory events: a total of 402 migratory events involving large-sized insects (comprising 21,155 individuals) were identified, including 149 large-scale events and 54 events with significant collective orientation; 440 migratory events involving medium-sized insects (comprising 921,997 individuals) were recorded, among which 145 were large-scale events, all exhibiting distinct collective orientation.

This study innovatively used individual vertical velocity data from VLR to establish a behavioral classification framework for migratory insects' takeoff, cruising, and landing, thus clarifying the seasonal behavioral rhythms of insects of varying body sizes: takeoff peaks at one hour after sunset, while landing peaks at one hour before sunrise; the period from two hours post-sunset to two hours pre-sunrise corresponds to the main cruising phase. Additionally, we developed a 'Layer Quality' index to quantify insect layer concentrations.

Future work will integrate pressure-level meteorological data from the ERA5 dataset (e.g., wind speed, temperature, air pressure) to match the real-time meteorological conditions during individual insect flight, aiming to further dissect the correlation mechanisms between insects' selection of takeoff/landing timing, layered altitude, and meteorological factors by generalized additive model. This study quantifies migratory insects' flight behavioral indicators, enhancing understanding of nocturnal insect migration. It provides a theoretical basis for accurate trajectory prediction and early warning of migratory pests, with important practical implications for agricultural pest management and biodiversity conservation.

Integration of bird radar, visual observation and acoustic recordings to inform collision risk modelling of bird species in a remote site in the Pilbara, Western Australia

Jill Shephard, Karen J. Riley, Patricia A. Fleming

Birds Tracking radar Applied radar aeroecology

In the next 10-15 years proposed onshore wind turbine projects in Western Australia will construct and commission nearly 5500 turbines, 2.5 times the number that are currently in operation. This rapid upscaling has revealed significant knowledge gaps around flight aerodynamics and flight heights in Australian bird species. Typically, flight height data are estimated using unverified visual observations, are concentrated on large, obvious bird species, are reliant on skilled observers and can be dependent on logistical constraints such as road access (and therefore proximity to the bird observed). High-resolution GPS-tracking have also been used to model flight dynamics and height, but these data must be corrected to remove cumulative horizontal and vertical error. Bird radar offers an alternate approach. We have used two Robin Radar MAX ® x-band bird radar systems to map bird activity and flight heights at a proposed wind farm in the Pilbara, Western Australia. The area is remote with summer temperatures in the high 40 degrees Celsius. Eight seasonal observations surveys between 2023 and 2025 identified 123 species within the development area, of which 109 have flight height data, and 23 were in the proposed rotor swept area. Concurrent field-based visual validation of radar tracks has allowed the annotation of 690 flight tracks across 26 species with forward and backward track extension in post-processing capturing over 93,000 radar points and 1,698 minutes of flight data, enough to facilitate standard movement ecology analyses in 50% of species. Additionally, acoustic records have given the opportunity to assign presence to species ascribed as clutter by radar classification. These analyses are ongoing with future analysis concentrating on AI assisted species identification informed by labelled track data, realised and modelled flight aerodynamics, and flight modalities. These data will be used to populate collision risk models and represent one of very few studies utilising radar technology in Australia.

Physics-based simulation and inference of flapping-wing kinematics from FMCW micro-doppler radar

Johannes Nüesch, Felix Liechti, Hugo Aguettaz, Dominik Kleger

Bats Birds Insects Tracking radar Methodological advances Applied radar aeroecology Animal behaviour

Recent progress in X-band FMCW bird radars now yields micro-Doppler spectrograms that clearly resolve wingbeat dynamics of birds, bats and insects. However, automatically extracting biologically interpretable descriptors from these data remains difficult. We present a physics-based simulation and inference framework that connects articulated animal models directly to radar observables and recovers biologically meaningful flight descriptors from raw spectrograms without manual labels.

Our forward model describes wing and body kinematics, including stroke geometry, flapping dynamics and morphology, and propagates hundreds of moving scatterers through an intermediate-frequency radar model. Coherent summation at intermediate frequencies reproduces realistic micro-Doppler structure such as stroke harmonics, interference fringes, clutter and rain artefacts for a vertical-looking X-band FMCW system.

From large synthetic datasets sampled over morphology, kinematics and flight trajectories, we train a structured-output U-Net that maps spectrograms to interpretable quantities: per-body-part Doppler contributions, velocity traces along the body and wings, wing-stroke phase time series, and simple morphological proxies. The recovered parameters are fed back through the simulator to reconstruct radar signatures, allowing us to validate the accuracy of the inferred kinematics against real micro-Doppler field data.

We apply the framework to a vertical FMCW radar dataset capturing birds, bats and insects moving through a fixed airspace column. When run on these unlabeled field recordings, the model retrieves trait-based descriptors that expose systematic differences in flight behaviour across taxa and allow tracks to be organized into biologically meaningful groups. This demonstration shows that physically grounded interpretation can reveal structure that is otherwise inaccessible in raw micro-Doppler data, offering a practical basis for continuous, multi-taxon monitoring of aerial biodiversity. By linking radar physics to biological traits in a scalable, non-invasive way, the approach bridges the gap between radar physics and ecology for monitoring avian biodiversity and expanding the biological interpretability of radar observations.

Pathways to impact through long-term collaboration and serendipity

Judy Shamoun-Baranes, Hidde Leijnse, Maja Bradarić, Johannes De Groeve, Jens van Erp, Bart Hoekstra, Bart Kranstauber, Emiel van Loon, Berend Wijers, Hans van Gasteren

Birds Weather radar Tracking radar Applied radar aeroecology Biodiversity monitoring

Radar uniquely lends itself to quantifying life in the air. As such, harnessing radar’s potential facilitates generating new scientific insight and addressing challenges resulting from human-wildlife interactions. However, translating scientific insight into societal impact, in the form of conservation action or changes to human behaviour, can be challenging, particularly within academic settings. Due to the nature of scientific research, the investments needed to have societal impact may go far beyond the duration of a single project, the capacity of a single person, or organization. In the Netherlands, the University of Amsterdam, the Royal Netherlands Meteorological Institute and the Royal Netherlands Air and Space Force have been collaborating on radar aeroecology for over two decades, with diverse connections to industry and government catalyzed in part by the European Space Agency. Here we provide three examples of our pathways to impact: (1) the development of a bird warning system for military aviation safety, (2) mapping and modelling bird migration to reduce conflicts between avian conservation and wind energy development and operations and (3) public engagement in the Netherlands. We illustrate why the following building blocks have been essential to our work: vision, long-term collaboration, research infrastructure and co-design. Perhaps less tangible but highly relevant in shaping these pathways are passion, persistence and serendipity. Rather than presenting detailed research results we take you along our journey and share the lessons we have learned along the way.

Characterizing soaring-bird migration in Israel using radars

Korin Reznikov, Nir Sapir

Birds Weather radar Methodological advances Applied radar aeroecology

Israel is located at the heart of a globally important migration bottleneck for soaring migrants, with many raptors, storks, cranes, and pelicans crossing the country every year. Migration is a substantial part of the avian life cycle of different species and influences humanity in different ways, including the collision of soaring migrants with aircrafts and wind turbines. The study of nocturnal migration using radar technology advanced considerably during recent years, but that of soaring migrants is substantially understudied due to a lack of adequate tools for their detection and tracking. In 2023, a significant advancement occurred with the introduction of a deep learning model for the automatic detection of soaring bird flocks in weather radars. Building on this foundation, we developed two complementary methodological components: (1) a dedicated approach for estimating species-specific radar cross-section (RCS) values, and (2) an algorithmic framework that integrates both the model’s detections and the newly derived RCS estimates. Together, these tools enable the quantification of migrating soaring birds and the characterization of their spatial distribution and migration altitudes across central Israel. We have applied this framework to the autumn migration season, and the results, representing the first of their kind, estimate approximately twice the number of migrating soaring birds as previously documented by ground surveys in the region. To validate these estimates, we compared the counts obtained by the radar and those collected by ground observations by restricting the analysis to the same spatial and temporal ranges. The results show a strong correlation between the radar and survey counts, indicating that the higher overall counts from the radar are due to its much broader detection range compared with the limited detection range of ground surveys, rather than methodological bias. These findings also reveal new insights regarding the magnitude of diurnal bird flock migration over Israel. Our study establishes the groundwork for a deeper understanding of diurnal soaring bird flock migration, facilitating the application of radar aeroecology to conservation and aviation-safety challenges.

Seeing insects at scale: Weather radar insights from the UK & India

Mansi Mungee, Maryna Lukach, Chris Shortall, James R. Bell, Elizabeth J. Duncan, Freya I. Addison, Lee E. Brown, William E. Kunin, Christopher Hassall, Ryan R. Neely III

Insects Weather radar Spatial temporal movement processes Biodiversity monitoring

Insects dominate terrestrial and freshwater ecosystems in terms of diversity, abundance, and ecological function, yet their large-scale population dynamics remain poorly resolved due to fundamental limitations in conventional monitoring approaches. In recent years, weather surveillance radars (WSRs), originally designed for meteorology, have emerged as powerful tools for observing aerial insect abundance and movement across unprecedented spatial and temporal scales. My work brings together two complementary strands of work: an empirical analysis from the United Kingdom and a recent, short tropical case study from central India, to demonstrate how radar aeroecology can advance insect ecology in a rapidly changing world.

First, I present results from a recently published empirical study using the UK Met Office’s national weather radar network to quantify diurnal and nocturnal aerial arthropod abundance across the UK over an eight-year period. By extracting columnar vertical profiles from dual-polarisation radar scans, this work provides one of the most comprehensive spatial assessments of aerial insect abundance from the UK, revealing strong spatial heterogeneity, contrasting diurnal and nocturnal trends, and clear associations with land cover, climate variables, and artificial light at night.

Building on this empirical foundation, I then shift focus to the tropics, where insect diversity is highest but large-scale monitoring remains almost entirely absent. The second part of the talk synthesises current knowledge on aeroecology in India, situating insect movement and emergence within the distinctive atmospheric dynamics of the Indian monsoon system. I review the sparse history of radar-based biological research in India and highlight a striking paradox: India possesses one of the densest and fastest-expanding weather radar networks in the tropics, yet these data have scarcely been used for ecological research. To bridge this gap, I introduce a detailed case study from the dual-polarisation C-band ARCTI radar at Silkheda near Bhopal, central India, illustrating how Indian weather radars can detect diel patterns, vertical stratification, and seasonal variability in tropical aerial arthropod assemblages.

Mapping bird migration in space and time to support sustainable wind energy development in Norway

Øyvind Nyheim, Bart Hoekstra, Judy Shamoun-Baranes, Diego Pavón-Jordan, Roel May

Birds Weather radar Methodological advances Applied radar aeroecology Spatial temporal movement processes

Wind energy development is expanding rapidly worldwide, increasing the potential for conflicts with migratory birds through turbine collisions, barrier effects, and area avoidance. In Norway, limited knowledge of spatial and temporal migration patterns constrains our ability to assess and mitigate potential impacts during both wind farm planning and operation.

We use data from the Norwegian weather radar network to quantify the spatial and temporal patterns of bird migration across southern Norway. We quantified migration phenology by identifying periods of peak activity within nights and across seasons, and by estimating how concentrated migration was in time. For quantifying the spatial patterns, polar volume scans collected during peak migration periods were processed to remove non-bird signals and aggregated into spatially explicit maps (1 × 1 km) of average seasonal migration intensity within a 50 km radius of six radars. We then used the INLA framework to fit spatial GAMs of seasonal migration intensities across southern Norway to identify important migration corridors and understand how they are shaped by landscape features.

Across regions and seasons, migration intensity peaked approximately 1.5 hours after sunset, although the duration of peak migration activity varied between flyways. Approximately 50% of seasonal migration occurred within a continuous ~35-day period in both spring and autumn, although the exact timing of this period differed somewhat between flyways. Second, when looking at individual migration nights (not necessarily consecutive), migration was concentrated to fewer peak migration nights in autumn (15-17) than in spring (18-22), with little variation between flyways. The migration maps provide insights into the location of migration corridors and how these are shaped by the Norwegian landscape.

The obtained insights into spatial and temporal migration patterns across southern Norway can provide support for more ecologically benign wind energy development for migratory species. High-resolution spatially explicit migration intensity maps can provide a foundation for identifying areas of high migration activity that may should be avoided during siting or need additional mitigation measures, while temporal phenology estimates can support turbine curtailment strategies during periods of high collision risk.

Three-dimensional reconstruction of bird density using inverse weather radar modeling

Raphaël Nussbaumer, Baptiste Schmid, Thibault Desert, Adriaan Dokter

Birds Weather radar Methodological advances Spatial temporal movement processes

Weather surveillance radars provide unparalleled spatio-temporal coverage of bird movements in the airspace, yet recovering their fine-scale three-dimensional (3D) structure remains challenging. Beam geometry effects, incomplete sampling of the lowest altitudes, and range-dependent smoothing have led most existing approaches to rely on vertically averaged products, typically assuming a constant vertical profile of bird density within the radar volume. While effective for estimating total abundance, such approaches discard horizontal heterogeneity and limit spatially explicit reconstructions.

Here we present a flexible inverse-problem framework to reconstruct 3D bird density fields directly from raw weather radar measurements. The radar observation process is formulated as a linear forward model mapping a discretized Cartesian bird density field to radar observations in spherical coordinates, explicitly accounting for Earth curvature and the full three-dimensional Gaussian beam power distribution. The resulting ill-posed inverse problem is solved using regularization, allowing the joint incorporation of spatial smoothness, vertical profile constraints, physical discontinuities such as coastlines, and advection constraints that promote temporal coherence of bird density fields along estimated flow directions.

The framework naturally accommodates heterogeneous sampling scales, propagates information into undersampled regions, and enables the integration of observations from multiple radars into a single 3D reconstruction. Using synthetic simulations and case studies based on NEXRAD data, we show that the method recovers realistic 3D bird density structures and temporal evolution beyond the capabilities of traditional vertical-profile-based approaches. This inversion-based framework provides a general foundation for fine-scale, spatially explicit analyses of bird migration, stopover behavior, and interactions with environmental features.

No effect of agri-environment schemes on radar-measured aerial insect abundance at landscape scale in England

Reuben O'Connell-Booth, Tommy Matthews, David Dufton, Julia Crook, Mansi Mungee, Ryan R. Neely III, William E Kunin, Christopher Hassall

Insects Weather radar Applied radar aeroecology Biodiversity monitoring

Agri-environment schemes (AES) commonly represent the largest financial investment in biodiversity conservation at a national or international level, but evidence for AES effectiveness remains equivocal. Here, we develop a novel and general method for assessing the impact of conservation interventions using Weather Surveillance Radar (WSR) to produce spatially explicit time series of aerial insect abundance over 1884 km^2 of agricultural land in England from 2015-2022. Using this dataset, we evaluated the landscape-scale causal effect of AES across 15 natural experiments involving paired AES intervention and control sites. We find no natural experiment which indicates a positive causal effect of AES, even at high levels of expenditure (up to £5122 km^2). When considering all 1884 km2 of agricultural land in England covered by the radars, we find a weak but significant negative correlation between AES and aerial insect abundance, corresponding to 0.29% fewer insects per £1,000 of AES expenditure. We find stronger, positive relationships between insect numbers of the percentage cover of woodland and semi-natural grassland. Our results provide the most robust evaluation of the benefits of AES and indicate that AES are not working to conserve aerial insects, as measured by WSR. We demonstrate the utility of landscape-scale conservation impact assessment using WSR-measured insect abundance paired with econometric impact assessment designs, a technique broadly applicable to problems in insect conservation science.

Analysis of aerofauna movement using Canadian S-band weather radar

Shannon Curley, Paul Knaga, Greg Mitchell, Adriaan Dokter

Birds Weather radar Applied radar aeroecology

We analyzed five years of data from five coastal weather surveillance radars to quantify the magnitude, timing, and altitude of landbird migration across offshore regions of Atlantic Canada. Offshore wind development is expanding rapidly in this region and may overlap with important migration corridors. By describing spatial and temporal offshore movements, our goal was to inform wind energy siting and identify periods and areas of elevated collision risk. We quantified terrestrial and offshore migration at two radars in Newfoundland and one in northern Nova Scotia; two additional radars in New Brunswick and central Nova Scotia were used to quantify terrestrial migration only. Migration intensity varied strongly by region. Offshore migration was very low in eastern Newfoundland (~1,500 birds/km/season in spring and ~3,300 in fall), suggesting this area represents the eastern edge of major migration pathways. In contrast, offshore movements across the Cabot Strait were substantial, averaging 130,879 birds/km/season at CASMM and 36,171 at CASMB. Similarly, migration across the Gulf of St. Lawrence was pronounced, with mean values exceeding 118,000 birds/km/season in both spring and fall at CASMM. Spring migration in Newfoundland and fall migration in northern Nova Scotia frequently showed two consecutive nocturnal peaks in migration intensity, one in the early night related to departing birds, and a second peak later in the night related to birds arriving after a sea crossing. These double peaks were equally pronounced in terrestrial and offshore data at CASMB, indicating that offshore routes are a common integral part of the migration system in this part of Atlantic Canada. We found limited evidence for strong seasonal differences in offshore migration magnitude, contrasting with terrestrial migration, which was on average 112% higher, and with prior studies along the U.S. Atlantic coast. In both spring and fall, 50% of offshore passage occurred within only a few nights, while 90% of passage was concentrated within a 2–3-month period. This temporal clustering highlights opportunities for forecast-based curtailment to reduce collision risk while minimizing operational downtime. These results provide actionable guidance for offshore wind planning and operation in Atlantic Canada.

Serendipitous detections from EarthCARE Cloud Profiling Radar provide first global climatologies of flying insects

Shannon Mason, Robin Hogan, Ryan Neely, Christopher Hassall, Tommy Matthews, Xu Shi, Jason Chapman

Insects Other radar Methodological advances

The Earth Cloud and Radiation Explorer (EarthCARE) satellite was launched in May 2024. Its Doppler cloud profiling radar (CPR), atmospheric lidar (ATLID) and multispectral imager (MSI) are designed for detecting the profile of atmospheric aerosols, clouds, and precipitation, and understanding their effects on the global radiation budget.

An unexpected feature of EarthCARE observations has been the widespread detection of bio-signals by CPR throughout the atmospheric boundary layer especially over land, where ATLID and MSI confirm that the detections occur in cloud-free volumes. Based on verification against ground based supersites and networks of weather radars and entomological radars, and using calculations of the estimated signals from birds and bats apparent to the spaceborne radar, we attribute the bulk of these detections to insects. This is consistent with their observed characteristics: deep plume features, sometimes capped by cumulus clouds, in convective boundary layers during the daytime, contrasting with more stratiform layers at night; a distinct preferential occurrence in warmer atmospheric temperatures; and strong seasonal cycles across that correspond to known insect distributions.

An insect detection variable has been added to the EarthCARE ATLID-CPR Target Classification (AC-TC) product. Profiling detections of insects in cloud-free volumes are provided with a vertical resolution of ~100m and an along-track scale of 1km, with a 25-day repeat cycle. The local equatorial crossing times of EarthCARE are 14:00 (descending) and 02:00 (ascending), providing some sampling of the diurnal cycle. Radar clutter obscures detections within 500m of the land surface. We show that some further insect signals can be extracted from the radar noise using dynamic along-track averaging, reducing the minimum detection threshold of CPR from -34 dBZ to -39 dBZ and revealing further detail in the spatial structure of insect detections.

We present the first global satellite climatologies of diurnal and nocturnal flying insects based on nearly 2 years of EarthCARE data. Additional information from EarthCARE products includes the height and temperature of insect detections, radar reflectivity, and meteorological information such as humidity and windspeed from the ECMWF analysis. We hope this novel and serendipitous dataset—and its potential for extension to a multi-decadal record from spaceborne cloud radars—will be of great use to the aeroecology community.

Annual intercontinental dynamics of passerine migration across the Strait of Gibraltar

Siméon Béasse, Alejandro Onrubia, José L. Ruiz, Birgen Haest

Birds Tracking radar Applied radar aeroecology Spatial temporal movement processes

Each spring and autumn, several billions of birds migrate between the European and African continent. Many of these fly along the East Atlantic Flyway and cross the Mediterranean Sea at the Strait of Gibraltar. This narrow sea crossing (14 km) is especially known to funnel migratory raptors and other large soaring birds like storks into a strongly concentrated corridor. For other bird groups like passerines, it is much less clear how much they are funneled towards the Strait because of their mostly nocturnal migratory movements, impeding visual observation. We put a vertical-looking radar on both the European and African shores of the Strait of Gibraltar from August 2024 to August 2025 to quantify the annual dynamics of migratory passerine movements and investigate how weather changes the spatiotemporal migratory pattern. We furthermore combine these measurements with standardized ground counts to estimate species-specific aerial migratory movements.

During spring, northward migratory traffic was similar at the African (236,000 passerines per km) and European (244,000 passerines per km) shore. In autumn, southward traffic was moderately higher on the African arrival shore (265,000 compared to 226,000 passerines per km). Up to 80% of passerine migration occurred at night.

Correlation between migratory traffic at both shores was higher in autumn (*R² = 0.56) than in spring (R² = 0.36). We detected strong, season-dependent wind effects on migration traffic. Migration intensity peaked under light and moderate directional favorable winds. In spring, winds towards the North and shifts in wind direction from West to East were associated with increased migratory activity. In autumn, winds towards the East increased southward passerines movements. These patterns suggest that passerines prefere to cross the Strait during times of calm or supportive winds.

Taxonomic resolution in dual-polarisation weather radar observations of biological scatterers: a simulation-grounded exploration

Tommy Matthews, Ryan Neely, Christopher Hassall

Birds Insects Weather radar Methodological advances

Weather surveillance radar is becoming firmly established as a tool for monitoring the complex large scale spatio-temporal dynamics of insect and bird populations, with multiple long-term entomological studies now in the literature. However, the questions of what we are really seeing and what we can possibly distinguish still loom large above these works. Two avenues to deepen our understanding of the biological radar observations we make are detailed “aerial-truth” sampling and simulating the radar data that would result from sampling different airborne species communities. To place simulation informed limits on the taxonomic resolution of biological weather radar observations, we use electromagnetic simulations to generate a library of scattering properties for different volant taxa. We ground our simulations in anatomically detailed models obtained from a variety of sources, including digitised natural history collections. From the library of scattering properties, we demonstrate that the dominant parameter is the ratio of length to wavelength (scattering parameter) and suggest that we do not see sufficient morphological variation within volant animals to see beyond the effect of the scattering parameter. Examining the dual-polarisation scattering properties in detail, we propose that, theoretically, there are four distinct biological categories; small, medium, and large insects, and birds. We show that the four categories have differing characteristics at S, C and X band, but remain, in principle, distinguishable. From this standpoint we explore the value of this level of taxonomic resolution, consider different behaviours and the relevance for taxonomic distinction, and investigate mixed communities of scatterers. We look to open-source our library of scattering results in the near future and happily invite requests for the addition of specific species (where a 3D model can be provided or found).

Quantifying bat flight heights

Trish Fleming, Angus Dempster, Kerryn Hawke, Karen Riley, Jesse Harper, Lazaro Roque-Albelo, Jill Shephard

Bats Tracking radar Applied radar aeroecology

Australia’s commitment to meeting emission targets has seen rapid growth in the renewable energy sector. Current wind farm proposals represent a projected 12-fold increase over current wind energy generation, with more, much larger turbines, planned. Understanding the potential impact of this growth of wind farms on threatened wildlife species requires species-specific collision risk data. However, reported turbine strike data is lacking for most Australian species, especially scarce, threatened species. Of 10 microbat species recorded at a proposed wind farm site in the Pilbara, Western Australia, the geographic ranges of half do not overlap with existing wind farms (and therefore there are no collision data available for these species), and there is flight height data available for only one species. We have deployed acoustic recorders around two Robin Radar MAX units, to collect species-specific flight data. Temporal and spatial joining of radar traces with acoustic records and weather data has been carried out to quantify (1) radar cross-sectional areas, (2) species’ flight height and speed, and (3) compare flight behaviour with environmental factors (wind speed, temperature and direction, air pressure, temperature). This work is building research and industry collaboration towards managing and mitigating the balance between green energy development and the conservation challenges that it represents. Understanding species-specific flight patterns will inform turbine collision mitigation choices.

What can and cannot be inferred from single- and multiple-point observations of ‘migration in progress’?

V. Alistair Drake

Birds Insects Other radar Methodological advances Spatial temporal movement processes

Since the late 1990s, small vertical-beam radars have played a leading observational role in aeroecological research by providing high-quality information about insects and birds as they pass directly overhead. These units examine only a tiny proportion of each migrant’s trajectory but accumulate data for a succession of different individuals: they observe ‘migration in progress’ at a single location. In contrast, researchers, conservationists, and pest managers are mainly interested in knowing where migrating populations originate from, where they are terminated, and what behaviours are employed to achieve these movements. This talk will examine what it is possible to reliably infer, from the extremely ‘local’ vertical-beam radar data, about the spatially extensive and intrinsically ‘non-local’ phenomenon of migration.

Observations of in-flight behaviour – the migrant’s heading direction – can be reliably related to the environment the migrant is directly experiencing – wind at the height of flight, landmarks, the direction of the sun or moon – as these are local phenomena measurable at the observation site. Vertical-beam radars also provide measures of migration intensity: fluxes and their integral over height, the migration traffic rate. Population transfers across the radar site can be estimated reliably and have ecological validity but as indicators of behaviour are biased by wind advection. Conversion to densities, by dividing by the migrants’ speeds, overcomes this issue but causal attributions relating differing densities to specific behaviours, such as a preference for movement in particular directions or when winds are favorable, are confounded by spatially varying and unknown source populations. Source and destination regions can be tentatively inferred by applying the radar observations to locations away from the radar site and incorporating established knowledge of flight initiation and termination behaviours. If winds are sufficiently variable, this allows populations of windborne migrants such as insects to be monitored over a wide area (a thousand square kilometres or more) at low (e.g. monthly) temporal resolution. Data from meteorological models can reduce uncertainty about winds along the estimated trajectories, and synoptic weather analyses can identify propagating wind changes before they reach the radar site, but unverifiable spatial extrapolation of the radar observations of heading direction and height of flight is still required.

Insect biomass plays an important role in seasonal migration intensity of passerines in Europe

Xiaodan Wang, Siméon Béasse, Judy Shamoun-Baranes, Vincent Delcourt, Birgen Haest

Birds Insects Spatial temporal movement processes

Seasonal variation in intensity of migratory bird movements has repeatedly been linked to local weather conditions, those earlier along the migratory route, or at the departure site. For many of these observed, sometimes strong, relationships between weather and avian migratory intensity, there is still unclear whether they act through direct or indirect pathways. The indirect effect on food availability is often put forward as the most likely pathway for effects of, for example, temperature. Separating the effects along the different pathways requires collecting data on both migratory bird abundance and their food. Collecting such combined predator-prey datasets is often very resource-intensive, and they as such remain rare. Here, we analyzed vertical-looking radar data from 18 European sites from 43° to 60° N and 1° W to 25° E over three consecutive years using Bayesian Structural Equation Models, to separate direct and indirect effects of weather (daily mean temperature, mean precipitation, and wind assistance at daily scale) and to quantify aerial insect abundance as a proxy for broader insect availability on the migration traffic of passerines. We found that insect biomass and precipitation were the strongest predictors of passerine migration in both spring and autumn. Migration traffic of passerines increased with higher insect biomass and lower precipitation during spring migration, whereas it decreased with lower insect biomass during autumn migration. In spring, insect biomass was most strongly associated with temperature, with higher temperatures corresponding to higher insect biomass. In autumn, insect biomass was positively associated with temperature and negatively associated with precipitation. Together, these findings suggest that weather conditions influence passerine migration intensity through pathways associated with food availability, while also being consistent with passerines selecting favorable environmental conditions for migration, but also highlight the interplay between food availability and environmental variability in shaping avian seasonal migrations.

Cross-calibration of weather radar and insect radars for quantifying aerial insect movements

Xu Shi, Boya Gao, Gao Hu, Jason Chapman

Insects Weather radar Methodological advances Spatial temporal movement processes

Weather radars are increasingly used worldwide to study the aerial movements and abundance of airborne insects. However, these applications often rely on broad assumptions about differences in self-powered airspeed and seasonality to distinguish insects from birds, and average radar cross-section (RCS) values across diverse insect taxa for quantification. Rigorous ground-truth validation are still lacking, making it difficult to quantify the accuracy, biases, and uncertainties associated with weather radar derived insect estimates. In particular, small deviations in assumed body size or species composition of airborne insects can substantially alter average RCS, and therefore strongly influence estimates of insect abundance and biomass. To address this gap, we will compare and validate weather radar observations with nearby entomological radars and aerial sampling data, collating data from three sites along a latitudinal gradient in East China. Our goals are to (1) use insect-radar and aerial sampling data to derive day- and night-time RCS distributions across taxa and size classes, (2) extract insect density at hourly and seasonal scales from weather radar and (3) evaluate the consistency of these estimates against insect radar measurements and aerial sampling. These results will be used to refine weather-radar-based estimates of insect abundance and biomass, and to develop guidance for future large-scale insect migration studies.

Millions of bats migrate along two continental flyways in Europe and the Middle East

Yuval Werber, Johannes De Groeve, David Troupin, Baptiste Schmid, Fabian Hertner, Judy Shamoun-Baranes, Arjan Boonman, Yossi Yovel, Nir Sapir, Birgen Haest

Bats Spatial temporal movement processes Migration

Animal migration is a fundamental driver of global ecological processes, linking distant ecosystems and delivering critical services across continents. Bats are known to engage in long-distance migratory journeys, but the scale and spatiotemporal dynamics of their movements are poorly documented. Here, we use a transnational network of 10 BirdScan MR1 radars spanning Europe and the Middle East to quantify seasonal movements of migratory bats across the two hypothesized flyways; the Western European Flyway and the Eastern Mediterranean Flyway.

Our measurements demonstrate massive seasonal migrations totalling over 3 million bats annually across both flyways in the Western Palearctic. Bat migration patterns largely follow flyways comparable to those of birds, but differ in timing, pace, and fine-scale directionality. Aerial densities of migratory bats varied with latitude, longitude, elevation, and Julian date, estimated by models capturing up to 48% of spring, and 37% of autumn variance in nightly densities along the Western European Flyway. Migratory movements peaked in spring and autumn in accordance with existing literature, though significant site-specific variability was observed. Within-night timing of migratory flights was largely similar in both flyways, with migration flights mainly occurring in the first quarter of the night, followed by a gradual decrease towards sunrise. In the Western European Flyway, bats predominantly migrated northeast in spring and southwest in autumn. In the Eastern Mediterranean Flyway, bats migrated north-northwest towards the reproductive grounds and south to wintering grounds.

The magnitude of these seasonal movements underscores an urgent need for systematic monitoring of bat migration to inform biodiversity conservation, maintain ecosystem stability, and mitigate human-wildlife conflicts, including zoonotic disease transmission and human-induced bat mortality.

Inverse estimation of vertical profiles in bioRad

Adriaan Dokter, Fengyi Guo, Raphaël Nussbaumer

Weather radar Methodological advances

Vertical profiles are widely used to quantify the density, speed, and direction of aerial taxa with weather radar. Most profiling methods only use data from ranges up to 25-35 kilometers, to prevent excessive vertical smoothing and loss of altitude resolution caused by the rapidly increasing beam width with distance from the radar. This short maximum range has limitations, such as a limited surveyed volume and a low upper altitude, especially for volume coverage patterns lacking high elevation sweeps.

Here we present an inverse estimation approach that fully accounts for the radar’s increasing beam width with range. Inverse estimates can resolve sub-resolution altitudinal structures in density and speed that are finer than the beam width. This feature allows us to incorporate data from longer ranges and thereby expand the surveyed area.

We show how the inverse vertical profile estimates compare to direct estimates for different hyperparameter settings that affect the profile’s smoothness and magnitude. Next, we highlight useful applications, such as quantifying high-altitude migration, sampling distant offshore movements by coastal radars, and increasing the altitude resolution of profiles. Finally, we show how vertical profiles can either be extracted for height bins defined relative to ground level or sea level. We discuss the adequateness of each when quantifying migration traffic and migratory stopover, especially in the context of regions with strong topography. All approaches discussed have been made available in the R-packages bioRad and vol2birdR.

How radar distance influences detectability and migration traffic rates: an experiment using a Detect Merlin True 3D radar at the Strait of Gibraltar

Gonzalo Munoz Arroyo, Javier Vidao, Alba Márquez-Rodríguez, Andrés De la Cruz

Birds Tracking radar Applied radar aeroecology Biodiversity monitoring

Bird radars are increasingly used to investigate avian migration, yet distance-dependent detectability remains a major source of bias in estimates of migration traffic rates (MTR). Calibrating radar performance across distances is therefore essential to identify the spatial range that best represents true migratory fluxes. We evaluated MTR consistency across multiple distance bands using a Merlin True 3D deployed at the Strait of Gibraltar, a key migratory bottleneck between Europe and Africa.

We analysed hourly radar data recorded at four concentric distance bands (1, 2, 3, and 4 km from the radar). Detectability was quantified as the proportion of detected echoes at each distance relative to the nightly maximum across all distances. Differences among distances were assessed using Friedman tests with paired post-hoc comparisons. Distance-dependent patterns were further modelled using a generalized additive mixed model fitted to logit-transformed detectability proportions, including a smooth term for distance and a random intercept for night. Consistency among distance bands was evaluated using Spearman rank correlations of hourly MTR values.

Radar detectability declined sharply and non-linearly with distance. Median detectability decreased from 1.00 at 1 km to 0.41 at 2 km, 0.17 at 3 km and 0.07 at 4 km, with all pairwise differences highly significant. The additive mixed model confirmed a strong non-linear effect of distance, explaining over 80% of the deviance. Consistent with this pattern, MTR correlations were strong between adjacent distance bands (Spearman’s ρ = 0.82 between 1 and 2 km; ρ = 0.83 between 2 and 3 km) but weakened markedly between the closest and farthest ranges (ρ = 0.26 between 1 and 4 km).

Our results demonstrate that radar-based estimates of migratory intensity are highly sensitive to observation range. We identify the 2 km distance band as an optimal compromise between detectability and ecological representativeness, minimizing local terrestrial movements captured at shorter ranges while avoiding severe detection loss at greater distances. This approach provides a practical framework for distance calibration in aeroecological radar studies and improves the robustness of migration flux estimates at coastal migration corridors.

Novel harmonic radar tools for tracking flying insects

Anastasia Lavrenko, Andrei Mogilnikov, Greg Storz, Graeme Woodward, Stephen Pawson

Insects Tracking radar Other radar Animal behaviour

The rapid decline of insect populations has intensified the need for robust tools capable of resolving fine-scale movement and habitat use across a wide range of flying and terrestrial species. However, existing tracking technologies remain severely constrained by tag mass, power requirements, and limited suitability for small invertebrates. Harmonic radar offers a unique solution by enabling the use of ultra-lightweight, battery-free tags, yet its broader adoption has been limited by the availability of flexible, field-deployable systems and validated performance data under realistic ecological conditions.

In this contribution, we present recent advances in the development of harmonic radar tools for tracking small insects, building on our earlier prototype systems and substantially extending their capabilities and experimental validation. We present two complementary platforms operating in the X-band/Ka-band regime:

(i) a stationary scanning harmonic radar designed for continuous area-based monitoring, and

(ii) a portable handheld harmonic radar intended for periodic re-localisation of tagged individuals in complex natural environments.

We report new results on system architecture, signal processing, and tag design, alongside expanded field experiments covering both aerial and ground-based use cases. Measurements demonstrate reliable detection and localisation at ranges exceeding 100 m under favourable conditions, with sub-metre localisation accuracy at short to medium distances. For terrestrial and low-flying species, we quantify the impact of vegetation, ground proximity, and tag orientation on detectability, highlighting key constraints and practical deployment considerations. We further discuss design trade-offs between operating frequency, tag miniaturisation, and environmental robustness.

By integrating stationary and mobile system design approaches, we aim at delivering a flexible toolbox for insect movement studies, supporting applications ranging from pollinator foraging analysis to the monitoring of cryptic or invasive invertebrates. The presented results outline both the current capabilities and remaining challenges, and point toward future developments including improved tag characteristics, adaptive sensing strategies, and scalable monitoring deployments for aeroecological research.

Effects of low-light levels on nocturnal migratory birds in flight

Simon Hirschhofer, Peter Ranacher, Robert Weibel, Barbara Helm, Davor Ćiković, Sanja Barišić, Louie Taylor, Maja Bjelić Laušić, Baptiste Schmid

Birds Other radar Animal behaviour

Low-levels of artificial light at night can influence the migratory behaviour of birds, but their effects remain poorly studied. We investigated the effect of artificial lighting on the flight behaviour of migratory birds along the Croatian Adriatic coast. We deployed two BirdScan MR1 ornithological radars: one in the illuminated city centre of Rovinj and the other at the edge of a nearby relatively dark ornithological reserve. Over two spring seasons, we monitored the migration intensity, flight altitude and airspeed of birds crossing the Adriatic Sea and those migrating along the coast.

At the illuminated site, birds reduced airspeed and lowered flight altitudes compared to the dark site, and these differences were intensified under cloudy conditions. Additionally, at the illuminated site, migratory traffic rate increased by 20% compared to the dark area.

These behavioural responses were documented in response to relatively low-intensity night-time illumination and thus suggest that even modest coastal lighting could impair orientation and elevate collision risk with illuminated infrastructure. This emphasises the need for targeted mitigation in light-sensitive migration corridors.

Towards a quantitative criterion for Italian alpine bird migration corridors

Clara Tattoni, Silvia Giuntini, Francesca Sanguinetti, Damiano G. Preatoni

Birds Applied radar aeroecology

Italian Alpine migratory corridors currently recognized by law as key routes for bird migration have been identified mainly based on historical information and qualitative assessments. However, the lack of a quantitative criterion makes it difficult to objectively verify their actual function as migration corridors. In this study, we evaluated whether the intensity and direction of migratory bird traffic can serve as operational parameters to distinguish recognized corridors from other sites with similar geographical features without legal protection.

Firstly, we ranked alpine passes as suitable for migration based on their location, elevation, orientaton and other geograpical caharacteristics. Then, we analyzed vertical-looking radar data collected in 2024–2025 at 33 Italian Alpine sites, including legally recognized corridors, corridors proposed for verification, and non-corridor sites. Parallel bioacoustic surveys were used to describe vocal activity and complement the interpretation of movement patterns. The *Migration Traffic Rate* (MTR), derived from radar data, was used as a quantitative indicator of movement intensity, and flight direction was considered to separate coherent migratory traffic from local movements and refine site classification.

Comparisons of MTR distributions and flight directions among site categories revealed systematic differences in crossing patterns, suggesting that these parameters may reflect corridor function. Based on these results, we identified a preliminary quantitative threshold potentially useful to distinguish sites serving as functional migratory corridors.

This approach provides an empirical basis for defining Italian Alpine migratory corridors operationally, supporting more consistent classifications and guiding the planning and management of migration monitoring activities.

Using corresponding radar and ringing data to explore migratory decisions of songbirds

Tsafra Evra, Daniel Bloche, Inbal Schekler, Yuval Werber, Noam Weiss, David Troupin, Nir Sapir

Birds Tracking radar Animal behaviour

Each year, billions of birds undertake seasonal migration across the globe, of which the vast majority are songbirds. They alternate between aerial endurance flights and stopovers, repeatedly deciding when and where to land, as well as when to depart. These migratory decisions are crucial for the completion of the journey and for the individual’s fitness. Understanding the landing decision considerations is particularly important to recognise the functions of the stopover that are required by the bird, which are crucial for the conservation of declining migratory songbirds. Despite this importance, landing decisions remain poorly studied, largely due to technical challenges in detecting the intrinsic condition of birds immediately after landing. To address this knowledge gap, we combined BirdScan MR1 radar data with standardised bird ringing data from one spring and autumn season, 2016, in Eilat, a globally important stopover site at a migratory bottleneck along the Eastern African-Eurasian Flyway. Using the vertical movements detected in the radar, we quantified the number of landing songbirds. We specifically asked: 1) is the number of landing songbirds better explains the number of migratory songbirds ringed on the following day than migration intensity aloft; and 2) whether combining radar and ringing data can promote our understanding of the landing decision. We found strong correlations between all radar-derived measures and standardised ringing data, with the number of landing songbirds providing the best explanatory power for the number of migratory songbirds ringed on the following day. Using the combined data, we showed that landing probability was higher during spring, despite generally higher migration intensity during autumn. Tailwind was the most important extrinsic condition, promoting passage and reducing the number of songbirds landing. During migration waves with high migration intensity, ringed songbirds had lower standardised body mass in spring but not in autumn. Overall, our results demonstrate the importance of using landing bird numbers and show a complex, likely density-dependent effect on migrant fuel stores. Integrating radar and ringing data provides a powerful approach for studying migratory decisions and addressing critical knowledge gaps in the stopover ecology of songbirds.

Predicting bird tracks with the help of nearby antecedent tracks

Emiel van Loon, Graciela DeCuba, Judy Shamoun-Baranes, Isabel Metz, Hans van Gasteren

Birds Tracking radar Methodological advances

Bird tracking radars are used at airfields to manage wildlife or adjust operations such that collision risks are minimized. To be useful air traffic control, bird tracks need to be forecasted with a lead time of a few minutes. A basic approach for bird track forecasting is to use the preceding track properties (geometry and speed) and extrapolate from there. This study investigates if the forecast accuracy and prediction horizon up to two minutes improve if properties from nearby and antecedent tracks are used in addition. The challenges are a) to define a suitable neighborhood (which nearby tracks might be informative and should be considered), b) to adequately measure similarity between the focal track and its neighbors and c) to combine the forecast ensemble to a best estimate with an uncertainty range.

We design a workflow which deals with these three steps, using insights from the latest literature on track similarity measurement and prediction. It is subsequently applied to a sample of bird tracks from a tracking bird radar (Robin Radar 3D MAX) covering a year-round period.

Preliminary assessment of the Detect Merlin True 3D radar to monitor bird migration across the Strait of Gibraltar.

Gonzalo Munoz Arroyo, Javier Vidao, Birgen Haest, Siméon Béasse, Alba Márquez-Rodríguez, Andrés De la Cruz

Birds Tracking radar Methodological advances Applied radar aeroecology Biodiversity monitoring

The Strait of Gibraltar is a key migration bottleneck in the Western Palearctic, yet nocturnal bird migration dynamics and their meteorological drivers remain difficult to investigate. Radar-based monitoring provides powerful tools to describe migration phenology and weather-driven variability, although migration estimates depend on radar system characteristics. We assessed nocturnal bird migration across the Strait during the post-breeding period (Aug–Nov 2024 and 2025) using a horizontal fixed-beam Detect Merlin True 3D radar deployed on the northern shore and compared its output with a simultaneously operating vertical-looking BirdScan MR1 radar at the same site.

Nocturnal migration intensity was quantified using nightly migration traffic rates derived from both radar systems, and seasonal patterns were described using generalized additive models. Merlin consistently produced lower migration rate estimates than BirdScan, representing on average about 20% of BirdScan values (geometric mean ratio = 0.20, 95% CI: 0.15–0.28). Despite this discrepancy in absolute numbers, temporal dynamics were strongly correlated between systems (Spearman’s ρ ≈ 0.72). Seasonal phenology showed a consistent bimodal structure across years, with a first migration wave in late September to early October and a dominant wave from mid-October to early November. Agreement was highest during the early and late phases of the migration season and decreased during the seasonal peak.

Meteorological effects on migration intensity were broadly consistent between radar systems. Migration activity increased under favourable wind conditions, particularly westerly and northerly components, and decreased under easterly and southerly winds. Seasonal and meteorological effects on radar discrepancies were weak, explaining only a limited proportion of the deviance.

Overall, while quantitative estimates from the two radar systems show substantial discrepancies, both radars reliably capture relative temporal patterns, phenology and weather-driven variability. Combining the spatial information provided by horizontal radar with the quantitative accuracy of vertical-looking systems offers a promising framework to improve estimates of nocturnal migration across the Strait and other major migration corridors.

Predicting bird trajectories using avian radar data and deep learning models to improve wildlife strike prevention

Graciela de Cuba, Isabel Metz, Emiel van Loon, Hans van Gasteren, Judy Shamoun-Baranes

Birds Tracking radar Applied radar aeroecology

In aviation, collisions between wildlife and aircraft pose a constant safety and economic challenge. Currently, airports implement wildlife management strategies to minimize the number of animals on the airport. However, these strategies are not sufficient to minimize wildlife strikes outside of airport boundaries. Near real-time warnings of potential wildlife strikes could be used to alter take-off, landing and holding patterns of aircraft. This requires knowledge of both current and near-future wildlife positions. The near-future wildlife positions can be obtained by predicting the wildlife trajectory based on their current positions. Within trajectory prediction research, substantial progress has been made in modeling and predicting movements by individual agents with Long Short-Term Memory (LSTM) models. These models are able to learn from the history of the trajectory and detect complex patterns in the data to predict future positions. In this research we apply and adapt an existing LSTM model for pedestrian trajectory prediction to 3D avian radar data capturing bird movement. In addition to the trajectory shape, covariates such as time of day, season and bird heading will be included in the model. The importance of the covariates and accuracy of the predictions will be assessed. The predictions of the model could be used in the future to provide warnings to airspace users when the probability of wildlife strikes is predicted to be high.

Estimating relative collision risk for migratory birds with wind turbines along the St. Lawrence River, Quebec, Canada

Greg Mitchell, Ana Diaz

Birds Weather radar Applied radar aeroecology

Canada’s wind energy capacity increased by 35% between 2019 and 2024 with 341 active wind energy projects consisting of more than 6700 turbines. Further, the number of wind turbines is projected to increase by more than 1000 by 2030. Unfortunately, many of the new wind turbines have blades extending 200 m into the aerosphere above ground level, posing a collision risk for migratory birds engaged in low altitude flights and there is currently no information on relative collision risk among locations within Canada. One region where significant investment in terrestrial wind energy turbines is expected to occur is along the St. Lawrence River in Quebec, Canada. As a first step towards quantifying relative collision risk for nocturnal migrants with wind energy infrastructure along the St. Lawrence River, we estimated migration traffic rates for spring and fall migration for 3-5 years of weather radar data from seven s-band weather radars in Canada (one from Ontario, five from Quebec, and one from New Brunswick) and three radars in the USA (one in New York, one in Vermont, and one in Maine). Three of the radars in Quebec are situated directly along the St. Lawrence River. We compare seasonal migration traffic rates and direction of flight within the 100-200 m altitudinal bin above ground level across radars to determine if collision risk is higher along the St. Lawrence River as opposed to regions not directly adjacent the river. In doing so, we also assess whether the St. Lawrence River functions as a migratory corridor in Canada. Last, we assess the weather conditions resulting in low altitudinal flights. The results of this study will be used by policy makers in Canada to help inform placement of wind turbines and mitigate collision risk.

Progress of China's Vertical Looking insect Radar Network

Jianrong Huang, Boya Gao, Jian Ma, Qi Chen, Haixia Lei, Zhi Zhang, Tao Zhong, Wei Sun, Jian Liu, Christopher Hassall, Ryan Neely III, Jason Chapman, Hongqiang Feng

Insects Tracking radar Spatial temporal movement processes Biodiversity monitoring

Each year, billions of insects undertake long-distance migratory journeys spanning thousands of kilometers, exerting a profound influence on both natural and agricultural ecosystems. To monitor these movements, Vertical-Looking Radar (VLR) has been validated as the most ideal tool for continuous and automated aerial surveillance, as it can precisely retrieve key biological parameters of individual targets, including body mass, wingbeat frequency, flight altitude, speed, and body orientation. Driven by recent advancements in industrial technology, high-resolution VLRs have been successfully developed in China. Between 2021 and 2025, a collaborative insect radar network was established by a consortium of research institutions and agricultural extension departments. As of December 2025, more than 20 VLR units have been deployed along a north-south transect across major agricultural provinces, including Hainan, Hunan, Henan, Jiangsu, Anhui, Zhejiang, Beijing, Liaoning, Jilin, and Heilongjiang. The raw data generated by the VLR network are stored locally at each radar site, while data quality assessment and the retrieval of biological parameters are conducted using specialized software developed by Hongqiang Feng. Current research progress facilitated by this network primarily focuses on two areas. First, preliminary analyses of the collected historical data have been performed to compare biomass flux and migratory patterns across different regions. This process has validated the reliability of the network’s data and provided quantitative descriptions of regional migratory characteristics, offering foundational datasets for studying the interactions between migratory pests and their environment. Second, the Henan Academy of Agricultural Sciences has developed a localized platform for the real-time analysis and visualization of radar data. This platform is already providing predictive intelligence on migratory pests, such as the beet armyworm (Spodoptera exigua), to agricultural extension departments, thereby offering critical technical support for the integrated management and control of migratory agricultural pests.

Integrated monitoring of avian behaviour around wind turbines: Insights from the Zwarte Wiek and ZWEMT projects

Joep Breuer

Birds Tracking radar Applied radar aeroecology Animal behaviour

The expansion of wind energy necessitates a comprehensive understanding of its ecological impacts, particularly concerning avian collisions and behavioural responses. The Zwarte Wiek (Black Blade) project, and its technological extension ZWEMT, addressed this challenge by deploying a suite of advanced sensors at the Eemshaven wind farm in the Netherlands to monitor bird activity and collision risk around turbines with and without a black-painted blade.

A multi-sensor approach was implemented, combining a Robin Radar MAX® system, thermal and visual cameras, vibration sensors (TNO’s WT-Bird® system), and both ground-based and turbine-mounted microphones. The radar provided continuous, three-dimensional tracking of bird flight paths, offering insights into flight height, speed, and straightness within a 200-metre radius of the turbines. Camera systems, including a novel stereo thermal setup, enabled detailed behavioural observations near the rotor disk, such as avoidance manoeuvres, perching, and interactions with turbine structures. Microphones were evaluated as a cost-effective alternative for collision detection, complementing the vibration-based WT-Bird® system.

Results indicate that the presence of a black blade did not significantly affect flight height, speed, or straightness of flight, as measured by radar and corroborated by camera data. While the WT-Bird® system registered only a limited number of collisions—primarily due to installation constraints and sensor placement—thermal cameras and microphones demonstrated potential for enhancing detection and behavioural analysis. The integration of these sensor modalities provided a more holistic view of avian activity, revealing both the limitations and opportunities of current monitoring technologies.

The findings underscore the value of complementing radar data with other sources, such as cameras and acoustic sensors, for robust avian monitoring at wind energy sites. This integrated approach not only advances the understanding of bird-turbine interactions but also informs the development of mitigation strategies and sensor technologies for future onshore and offshore wind farms.

Regional synchrony and directional coherence in offshore nocturnal bird migration

Maja Bradarić, Emiel van Loon, Johannes De Groeve, Bart Kranstauber, Bart Hoekstra, Judy Shamoun-Baranes

Birds Tracking radar Applied radar aeroecology Spatial temporal movement processes Using unique offshore tracking radar network

Offshore bird migration is often assumed to be more spatially and temporally heterogeneous than migration over land, as seas are commonly perceived as ecological barriers that may prompt varying migratory responses among different species. Alternatively, it could be argued that the absence of spatial features offshore may lead to homogeneous migration patterns at the regional scale. Until recently, offshore bird migration data collection was limited to sporadic deployments of different radar systems, therefore preventing regional comparisons of offshore bird migration patterns. The recent coordinated deployments of identical radars at sea overcome this limitation, allowing for region-wide assessment of synchrony in bird migration patterns. Here, we provide a multi-year regional overview of low-altitude nocturnal bird migration phenology across the Dutch North Sea. We further investigate spatiotemporal variation in primary migration direction and airspeed. In both cases, we focus on identifying and comparing seasonal and within-night patterns across six offshore radar locations. Our findings show that, although complex, the regional patterns are rather homogeneous, particularly at the seasonal scale during spring migration. We discuss the implications of our findings for current and future offshore wind-turbine curtailment strategies.

Radial velocity-induced biases in aeroecological products derived from C-band weather radar data

Nadja Weisshaupt, Jarmo Koistinen

Birds Insects Weather radar Methodological advances

Weather radar algorithms for meteorological purposes have been designed for interventions at different stages in the signal processing chain and postprocessing of volume scans. They were typically tested in precipitative conditions and optimized accordingly. However, when using weather radar data for biological purposes, these algorithms may not work equally efficiently or, commonly, they can even falsify outputs.

A radar measurement particularly prone to processing and other bias is radial Doppler velocity. It is one of the key variables behind a wide range of meteorological services to many stakeholders. A large number of algorithms has been therefore developed to address relevant biases in velocity data, such as ground clutter and aliasing.

While insects exhibit similar characteristics in radial velocities as precipitation, birds’ velocity features clearly differ because of birds’ self-powered flight. Radial velocity data is essential for estimating migration fluxes as migration traffic rates (MTR). Using undedicated algorithms in an ornithological setting requires thus careful inspection of their behaviour. Also, using data for bird analysis that have already been processed by meteorological algorithms, bears a significant risk for bias in birds’ radial velocities or also other measurements.

In this presentation we discuss the performance and impact of selected meteorological algorithms on bird data products derived from radial velocities. We also show impacts of different pulse repetition frequencies (PRF) and mixtures of scatterers on data content and quality.

Findings show that algorithms not finetuned for aeroecological purposes as well as a diverse pool of scatterers can lead to serious biases in radial velocities. In most cases, biases lead to underestimation of birds’ flight speeds and deviations in flight directions derived from single- and dual-PRF radial velocities.

It is important to understand the dependencies of certain scan parameters and their effect on measurements to avoid misinterpretation of aeroecological results. Many quantitative differences observed between weather radars are often ascribed to spatial variations in bird fluxes. However, radial velocity biases can be easily misinterpreted as biological phenomenon or missed entirely, if data is not scrutinized appropriately.

Biological classification of vertical-looking radar data using convolutional neural networks

Paraskevi Nousi, Elske Tielens, Michele Volpi, Silke Bauer, Birgen Haest

Bats Birds Insects Other radar Biodiversity monitoring

Small scale radar such as Birdscan MR1 identifies individual echoes in the sky and provides detailed information on the features of the animal target. Taxonomic identification in such radar images remains a challenging topic. In this work, we use MR1 radar data to differentiate between bats, and several bird and insect groups using machine learning. The data consists of echoes, i.e., the returned timeseries of reflectivity from the MR1 radar, for individually detected animals. Current approaches involve standard signal processing of the time series, e.g., working with the frequency content, to establish features such as the wing flap frequency. We instead use neural networks to model the raw timeseries itself and use learned features to then subsequently cluster the animal samples and look for subgroups. The developed neural networks use learned convolutional filters and operate on both the timeseries as well as the frequency series derived from it. We further incorporate other signal processing methods within the network to try and imitate the processing done during handcrafted feature extraction, without fully handcrafting features. As a first experiment, we show that the developed methods can reach the classification performance of the handcrafted features and surpass it in certain cases.

Simulating bird flight through offshore wind farms: An agent-based modelling approach

Sam Ordeman, Erik Fritz, Joep Breuer

Birds Tracking radar Methodological advances Animal behaviour

The rapid expansion of offshore wind farms along the Dutch coastline necessitates a robust understanding of their ecological impact on migratory birds. The BirdSafe project addresses this imperative by developing a simulation framework to model avian flight behaviour and collision risk, thereby supporting the design of nature-inclusive wind farms that minimise harm to wildlife.

This research adapts an agent-based model, originally conceived for traffic simulations, to emulate the flight trajectories of migratory birds traversing offshore wind farm arrays. The model is designed to capture a spectrum of behavioural responses, ranging from macro-avoidance of entire wind farms to micro-scale collision risk at the level of individual turbines. Crucially, calibration and tuning of the model are performed using radar data collected in the North Sea, enabling the behavioural parameters to be grounded in empirical migration patterns and observed collision probabilities.

A key focus is the exploration of how varying wind farm layouts influence predicted collision rates, providing insights into the optimisation of turbine placement for ecological mitigation. The model’s flexibility allows for the integration of empirical data and the testing of alternative scenarios, enhancing its utility as a tool for ecological impact assessment.

This multidisciplinary initiative fosters collaboration across technical and ecological domains. The outcomes are expected to inform evidence-based guidelines for the sustainable development of offshore wind energy, balancing the imperatives of renewable energy generation with the conservation of migratory bird populations.

Cross the sea or follow the coast? Results from 3 years of data along the Gulf of Lion, French Mediterranean Sea

Vincent Delcourt, Hélène Schopper, Baptiste Schmid, Cyprien Daide

Birds Applied radar aeroecology Spatial temporal movement processes Animal behaviour

The Gulf of Lion is a major migratory corridor for terrestrial and marine birds in the western Mediterranean Sea, but it is also an area with significant offshore wind energy development. To better understand the potential effects of these offshore wind farms on birds, the French Ministry launched the Migralion programme (2021–2025), with three complementary work packages for data collection: telemetry (WP3), boat surveys (WP4) and coastal radar monitoring (WP5). We present here the final results of work package 5, using a multi-method approach combining coastal bird radars, visual monitoring, acoustic recordings and bird ringing. During three years (2022–2024), we used two BirdScan MR1 radars at eight locations along the coast of the Gulf of Lion and collected 40,000 hours of data. Analyses focused on migration phenology, intensity, flight directions, flight altitudes and the identification of taxonomic groups, with particular attention to offshore movements and flight altitudes overlapping potential offshore wind farm areas. Results show that migration in this part of France occurs between mid-February and May 20, and between July 20 and November 30, with strong inter-daily variability and a marked concentration of movements over a limited number of nights. We found that, on average, 70% of migratory movements occur at night, and passerines account for more than 80% of recorded flows. Flight directions reveal strong spatial contrasts between the western and eastern parts of the Gulf, with a significant proportion of movements departing towards or arriving from offshore waters, particularly in the east and within the Camargue area. We estimate that the total migratory flows crossing the Gulf of Lion below 1,000 m above sea level are between 45 and 90 million birds in spring, and between 140 and 210 million birds in autumn. As a substantial proportion of birds migrate at low altitude (50–300 m), these results highlight the importance of taking into account terrestrial migratory birds in offshore wind farm development.

A local and regional spatial analysis of bird stopover distributions in the Great Lakes basin

Virginia Halterman, Fengyi Guo, Emily Cohen, Jared Wolfe, Jeffrey Buler

Birds Weather radar Applied radar aeroecology Animal behaviour

Forests in the North American Great Lakes basin provide essential stopover habitats for migratory birds. Understanding how migratory bird stopover densities relate to habitat availability is necessary to plan management for future stopover patterns under ongoing and projected land cover change across the region. The composition and geographic context of forested landscapes are known to influence stopover distributions in the region. However, the contributions of fine-scale forest structure and landscape configuration remain poorly resolved. Therefore, we assessed the influence of fine-scale landscape characteristics on stopover distributions by pairing stopover habitat use measures from 11 weather surveillance radars (WSR-88D) with forest characteristics measures from aerial lidar throughout the Great Lakes region. We used five years of radar data (2015-2019) during pre-breeding (April-June) and post-breeding (August-October) migration and fit boosted regression trees to model bimonthly average bird stopover densities among years. After removing nights with precipitation, non-bird movement, and radar anomalies, we had over 3,756 radar nights of bird-dominated exodus for our analysis. To get forest cover and configuration measures, we used a combination of USGS forest height and land cover data. For fine-scale characteristics we used USGS 3DEP aerial lidar data. Configuration measures include patch area, patch shape, and edge density. Preliminary results during the post-breeding season indicate higher stopover density correlated with greater forest patch area, but not with patch circularity. Forest height was not important in explaining bird stopover density. However, we expect to see increased stopover densities in forests with greater canopy height variation and structural diversity with the finer-scale lidar data as it is incorporated. Forest type is also important, with the amount of deciduous forest, and coniferous forest at the more northern latitudes, positively correlated with stopover densities. Our results will help inform forest management for migratory bird stopover at both the local and landscape levels across the U.S. Great Lakes basin.

Crossings and altitudes used by migratory birds around the Baltic Sea

Yohan Sassi, Fausto Dal Monte, Cecilia Nilsson

Birds Weather radar Applied radar aeroecology Spatial temporal movement processes Animal behaviour

The Baltic Sea, in northern Europe, is part of the East Atlantic flyway, one of the 8 major bird migratory flyways. This area is regularly crossed by millions of migratory birds and has more than 600 offshore wind turbines already operating and several new offshore wind energy facilities projected. This will likely lead to an increase in the bird-human activities conflict in the coming years. A well thought spatial planning thus appears to be a crucial tool to prevent the implementation of wind turbines in areas intensively used by birds and increasing collision risks. We leveraged data collected by weather radars around the Baltic Sea, over 4 years, to map migration flows and highlight areas where conflict can arise. Specifically, we used vertical profile time series quantifying bird migratory behaviours to create movement networks between radar stations to identify where and at which altitude nocturnal migratory birds cross the Baltic Sea. We then produced maps showing migration flows across radar stations around the Baltic, highlighting the proportion of birds flying at altitudes overlapping rotors swept zones, thus providing the wind energy industry with information on sensitive areas that should be avoided.

Novel estimates of nocturnal bird migration traffic at the continental scale using participatory science data

Yuting Deng, Ethan Plunkett, David L. Slager, Miguel Fuentes, Yangkang Chen, Benjamin Van Doren, Adriaan Dokter, Daniel Sheldon

Birds Weather radar Biodiversity monitoring

Quantifying movement patterns of migratory birds throughout their annual cycles can guide effective conservation and help mitigate risks. Integrating migratory trajectories models across species and with other data sources could advance migration ecology and conservation applications. However, trajectories models are often not well suited for integration with location-based monitoring technologies such as weather radar. Here, we provided a scalable, species-level measurement of nocturnal bird migration traffic and assessed its utility for integrating with radar monitoring. We introduce BirdFlow migration traffic rate (BMTR), a location-based metric of migratory passage for individual species computed from BirdFlow models. BMTR quantifies the weekly proportion of a species’ population passing over a transect, can be calculated across a species’ entire range, and can be converted to absolute numbers using population estimates. We applied this framework to compare with average weekly nocturnal migration traffic detected by 152 NEXRAD weather surveillance radars. We also investigated how BMTR can be used to disaggregate radar-derived migration traffic into species-level estimates. BirdFlow and radar-derived migration traffic showed strong agreement (r = 0.784), particularly along the Mississippi and Atlantic Flyways. Models incorporating demographic adjustments performed best. Annual nocturnal migration traffic estimated by BirdFlow and radar was closely aligned, with BirdFlow estimates averaging 33.3\% higher. BMTR effectively captured flyways, high-traffic routes, and seasonal migration dynamics. It also enabled disaggregation of radar traffic into species-level contributions, revealing dominant species at specific locations. BMTR provides a new methodological bridge between trajectory-based BirdFlow movement models and location-based monitoring approaches. It offers a scalable, species-specific tool for migration ecology and conservation, supports monitoring in areas without radar coverage, and holds promise for public engagement through visualization of near real-time species-level migration.

Using the new getRad R package to process open weather radar data in an operational pipeline

Bart Kranstauber, Berend Wijers, Pieter Huybrechts, Bart Hoekstra, Koen Greuell, Judy Shamoun-Baranes, Peter Desmet

Weather radar Methodological advances

The recent advancement of radar aeroecology has been driven in part by unrestricted access to unfiltered weather radar data. In Europe, however, access to such data remains fragmented, as many national meteorological institutes prioritize distributing products that are processed for meteorological applications. At the same time, several meteorological institutes are releasing radar data through open data initiatives, but heterogeneous access protocols still hinder transnational studies.

To facilitate working with open weather radar data, we developed the open source R package `getRad`. The package provides unified programmatic access to radar data for aeroecological applications, allowing users to acquire, standardize and load polar volume data from ten different country specific repositories directly into R. In addition, derived biological information in the form of vertical profiles can be retrieved from the Aloft data portal. Here, we introduce the `getRad` package and invite community contributions.

Furthermore, to work towards a continent-scale, high quality dataset, we implemented an operational pipeline to continuously compute vertical profiles for European radars. This pipeline uses `getRad` within Notebook-as-a-Virtual Research Environment (NaaVRE) developed by LifeWatch ERIC and the University of Amsterdam. We present results from this pipeline and compare them with vertical profiles generated with data processed at BALTRAD through the European collaboration of meteorological institutes. Together these steps highlight the potential of an open infrastructure for radar aeroecology.

Quantifying intercontinental migratory insect flows across the Strait of Gibraltar

Siméon Béasse, Alejandro Onrubia, José L. Ruiz, Birgen Haest

Insects Tracking radar Other radar Spatial temporal movement processes Biodiversity monitoring Animal behaviour

Giant masses of insects migrate through the skies whenever weather conditions permit. Many of these movements are migratory and happen at high altitudes in the sky, making them invisible to the naked eye. Radar remains one of the few tools to quantify these, otherwise largely unnoticed, aerial movements. For migratory birds, the Mediterranean Sea forms a migratory barrier and the Strait of Gibraltar a major bottleneck through which seasonal movements between the African and European continent are funneled. To which extent intercontinental migration across the Mediterranean Sea also occurs in insects, and whether these follow similar spatial and temporal patterns, remains poorly understood.

We put a vertical-looking Birdscan MR1 radar on both the European and African shores of the Strait of Gibraltar from August 2024 to August 2025 to quantify the annual dynamics of migratory insect movements and investigate how weather changes them. We derived daily (24 hour) migratory insect traffic from the overall measured insect flight activity by subtracting the number of insects moving south from those moving north. Migration had a clear overall tendency towards the north in spring and towards the south in autumn at both continental shores. During summer, there was a lot of insect flight activity, but this translated to little migration because of the omni-directionality of movements throughout the daily cycle. In spring, 5.2 million insects migrated northwards from the African shore per horizontal km of airspace and around 2.7 million insect per km arrived due North at the European shore. During autumn, 3.8 million insects per km migrated due South from the European shore, and 1.2 million per km arrived at the African shore. Daily numbers of migratory insects per km were strongly correlated during both seasons but more so in autumn (Pearson correlation = 0.61 compared to 0.46 in spring). More insects migrated during the night than during the day, with the nocturnal proportion varying from 54 to 73%, depending on the season and location. These findings not only indicate the existence of massive migratory movements between both continents across the Mediterranean Sea but also suggest likely high mortality rates during the crossing.

Testing hypotheses regarding the composition of bird migration in the Levant

David Troupin, Nir Sapir

Birds Other radar Spatial temporal movement processes

We analyzed nocturnal migration data collected by radars (Birdscan MR1) at 10 sites throughout Israel over a period of 8 years (2015-2023, n=2,908 days of observation, 2.66 million identified birds). We tested two hypotheses regarding the composition of migrating bird groups (waders versus passerines): (1) higher proportions of wader species would be observed in proximity to the coast of the Mediterranean Sea, and (2) higher proportions of wader species would be observed at locations with larger areas of water bodies (e.g., reservoirs and fish ponds) in their proximity. Overall, passerines comprised 62% of identifications and waders 38%, with variation by season (autumn: 40% waders; spring: 33% waders) and location (range: 30-55% waders). Generalized linear mixed models on Box-Cox transformed ratios revealed that geographic and seasonal factors explained 14.0% of variance (R² marginal), with site-specific characteristics adding 4.3% (R² conditional=18.3%). Contrary to hypothesis one, coastal proximity showed no significant effect (season × distance to Mediterranean coast: β=-0.033, p=0.281). The strongest predictor was latitude (β=-0.866, p=0.036), with northern sites showing 30-34% waders versus 42-55% at southern sites. Regarding hypothesis two, the variable of total area of water bodies within a 5 km radius showed complex patterns: larger proximate water body areas generally favored waders (β=0.201, p=0.038), but in the spring we found a significant negative interaction (β=-0.107, p<0.001), indicating that in the spring migrants prefer smaller, more concentrated wetlands as stopover sites. Altitude effects reversed seasonally (β=0.099, p=0.001), with higher proportions of waders in higher altitudes in spring but not autumn. These findings demonstrate that nocturnal wader migration is structured primarily by north-south climatic gradients rather than coastal proximity.

Large-scale synchrony of the autumn bird migration in Baden-Württemberg, Germany

Herbert Stark, Felix Liechti, Marc I. Förschler

Migratory birds are increasingly affected by expanding human infrastructure such as wind turbines and artificial light, which create additional risks along migration routes. Understanding spatial and temporal migration patterns is essential for developing mitigation measures. In Baden-Württemberg, migration was recorded simultaneously for the first time at two sites (Swabian Alb and Black Forest National Park) during autumn 2025 (August 31–November 30) using identical radar systems. Seasonal and diurnal patterns were very similar at both stations. Daytime migration intensity on the Swabian Alb was twice as high as in the Black Forest (183 vs. 93 birds·h⁻¹·km⁻¹), whereas nighttime intensity differed by only 5% (610 vs. 557). The lower daytime values in the Black Forest are likely due to very low-altitude migration (<50 m) in ridge areas, which was not adequately detected by radar. Migration intensities were strongly correlated both during the day (r = 0.65) and at night (r = 0.86). Daily changes showed stronger correlation at night (ρ = 0.83) than during the day (ρ = 0.28), indicating that short-term fluctuations occur synchronously only at night. The high synchronisation over 120 km enabled extrapolation of migration volume for the region, suggesting that at least 200 million birds crossed Baden-Württemberg between August 31 and November 30, 2025. These findings indicate that even a limited number of monitoring sites can accurately predict migration patterns and support targeted conflict-reduction measures.

An operational system for real-time bird migration monitoring and forecasting for military aviation

Mats Veldhuizen, Bastiaan Anker, Jouke Jacobi, Ely Deckers, Belén Torrente, Bart Kranstauber, Hans van Gasteren, Judy Shamoun-Baranes, Hidde Leijnse

Birds Weather radar Applied radar aeroecology Spatial temporal movement processes

It is vital to monitor and forecast bird migration for flight safety. Contrary to civil aviation, military aircraft operate at heights where bird migration frequently occurs, greatly increasing the risk of colliding with birds en-route (bird strikes). Because bird strikes affect operations and damages are costly, the Royal Netherlands Air Force uses FlySafe: a bird migration monitoring and forecasting system. FlySafe is operated by the Royal Netherlands Meteorological Institute (KNMI) and uses weather radar data from the Netherlands, Belgium, and Germany to map bird migration. A forecasting model (developed by the University of Amsterdam and trained using weather radar data) is used to provide bird migration forecasts.

FlySafe has been operational for almost 15 years. A recent modernization includes the latest bird migration extraction algorithms and forecast models, and modern visualization techniques. In particular, we use KNMI’s GeoWeb/Adaguc stack for interactive visualizations of animated maps of vertically-integrated bird densities and height profiles of bird densities at radar locations. GeoWeb is a web-based integrated system that is used for monitoring the atmosphere, both in an operational and a scientific context. It is easily extendable with other geospatial (including meteorological) data, and it is therefore also well-suited for exploring bird migration data. All bird migration data (both profiles and maps) are available through the, publicly available, KNMI Data Platform, allowing for further use of these data.

We will demonstrate the new FlySafe service with migration case studies, the emphasis will be on the improvements that were realized in the recent upgrade. We demonstrate the added value of the interactive visualization of bird migration data and the possibilities to extend to other countries.

Detecting songbird flocks with a tracking radar

Jacco Leemans, Elisa Bravo Rebolledo, Koen Kuiper, Abel Gyimesi

Birds Tracking radar Applied radar aeroecology

In the Dutch North Sea, offshore wind farms are shut down when major nocturnal bird migration events are expected, to reduce collision fatalities. These mass migration events are primarily predicted by a model trained on data of offshore tracking radars. Most nocturnally migrating birds are passerines and other terrestrial birds, which are known to often flock together during migration. The radars can classify tracks as a flock. However, we have little information on how well these radars are able to detect songbird flocks. Thus far, we still consider these tracks as one bird as we have no more information on the actual group size. In this study, we investigated how songbird flocks of different group sizes are tracked by the radar, and whether the group size of flocks can be linked to radar track characteristics. For this, we analyzed data collected at a coastal test site of these radars, during the autumn migration of 2025. The radar concerned a horizontal Furuno magnetron-based S-band radar of Robin Radar, identical to the ones operational offshore. We show that flocks of songbirds often result in a cluster of radar tracks that may be a tenfold lower in number than the size of the flock. These tracks were not always classified by the radar as a flock. We found that groups up to several tens of birds may be detected as a single radar track that is not classified as a flock. On the other hand, tracks classified as a flock may only concern a single bird. We discuss whether these findings could contribute to an improvement of the migration prediction model.

Operational use of avian radar to identify, predict, and respond to flocking waterfowl hazards

Jeff Follett, David Bradbeer

Lesser snow geese winter on the Fraser River estuary in southwestern British Columbia and are a hazard to aviation operations at Vancouver International Airport. Though the geese did not consistently use habitats at the airport, they actively transited between habitats outside the airfield. This includes movements between tidal sedge marshes, upland agricultural fields, and suburban amenity turf grass fields and lawns. The hazard associated with these daily movements was most acute when flocks transited the departure runways. Understanding the routes taken by the geese and the timing of the flights allowed airport managers to proactively deploy resources to manage these specific wildlife hazards. It also supported the planning and implementation of novel operational procedures as further mitigation.

We used an avian radar system to characterize patterns of lesser snow goose movements. Analysis of flight patterns were initially conducted by studying hour-long blocks of radar data to visually identify radar tracks that were associated with goose movements. We also conducted observations of geese in the field to create a validated database of radar tracks. From these data we identified target characteristics associated with geese and created geospatially defined zones within the avian radar system that would automatically alert if goose flocks persisted within the zones. This provided on-duty wildlife management personnel with real-time awareness of goose movements. Using the analysis of flight pattern methodology, we identified a new trend of goose flight behavior in spring 2022. With the available information, we were able to strategize mitigations to the new flight behavior, which included a daily scheduled watch to detect the geese; real-time radar alert zones to improve situational awareness; and the coordination of dynamic departure management with ATC, airport safety officers and wildlife management personnel.

This work demonstrates that avian radar systems can be used to 1) identify new patterns of flight behavior; 2) visually communicate the hazard to airport stakeholders; and 3) provide real-time situation awareness to on-duty wildlife personnel. Another finding is that novel departure management procedures can reduce the likelihood of striking flocking waterbirds.

Biological potential of KaRVIR; Ka-band Rapid-scan Volumetric Imaging Radars

Jeffrey Kelly, David Bodine, Boonleng Cheong, Jorge Salazar Cerreno

Other radar Methodological advances

Motivated by the need to better understand the role of clouds and boundary layer processes in Earth system science and extreme weather, the Advanced Radar Research Center at the University of Oklahoma is developing millimeter-wavelength phased array radar with rapid volumetric scanning, fine spatial resolution narrow beamwidths (<100 m), and excellent sensitivity. The project will develop and field two Ka-band, polarimetric, mobile phased-array radars (PAR) called the Ka-band Rapid-scanning Volume Imaging Radars (KaRVIRs). Each KaRVIR system will have a 0.3 degree beamwidth and provide volume scans in 20 s or less, enabling an order of magnitude reduction in resolution volume size compared to existing PARs. While the project is primarily motivated by the need to understand cloud physics, there is also growing awareness of the key role of the atmosphere as a habitat for biodiversity. Most animals (mostly insects) are small (<1 cm3), fly, and cannot be tracked in nature with traditional biological methods. Existing radars have limited spatial and temporal resolution which obscure our understanding of key aerial biological phenomena. The proposed KaRVIR instruments will greatly advance our ability to assess behaviors of very small animals while they are in flight. These observations are expected to provide unprecedented insights into the relationships between dynamic atmospheric conditions and animal behaviors. Examples of areas where KaRVIR will improve our understanding include a number of studies of pest insects that have significant detrimental effects on agriculture and human health. Another application will be to explore spatial and temporal dynamics in insect biomass. Understanding fundamental aerial flight dynamics of airborne insects is vital to modeling their abundance and long-range dispersal and therefore mitigating or adapting to these biomass changes.

Species-level bat identification from radar using micro-doppler and acoustic ground-truthing

Kseniia Kravchenko, Dominik Kleger, Johannes Nüesch, Lucile Morcelet, Amee Assad, Kevin Barre, Jannis Gottwald, Alexandre Millon, Christian E. Vincenot

Bats Tracking radar Methodological advances Applied radar aeroecology Spatial temporal movement processes Biodiversity monitoring Animal behaviour

Bats are highly mobile mammals whose small size, nocturnal habits, and often high-altitude movements make them difficult to study using conventional field methods. While high-resolution tracking approaches such as GPS telemetry can provide detailed movement data, their application is typically limited to a small number of individuals, short deployment periods, and a restricted subset of species, and may still involve substantial spatial uncertainty. As a result, obtaining species-specific information across the full vertical and temporal range of bat activity remains a major challenge.

Radar offers a unique potential for monitoring aerial fauna over large spatial scales, but distinguishing bat species from radar signals alone has not yet been achieved. This work introduces a framework that combines micro-Doppler radar signatures with acoustic ground-truthing to advance species-level bat identification.

The approach integrates broad-range radar monitoring with co-located acoustic detectors capable of providing reliable species labels within their effective range. By synchronising detections across the two systems, acoustic identifications can be associated with corresponding radar returns. These matched events enable exploration of how morphological and behavioural differences among bat species—reflected in micro-Doppler modulation, wingbeat patterns, and movement characteristics—may allow classification from radar features alone. In this way, the method aims to extend species inference beyond the acoustic detection radius, offering the possibility of species-aware interpretation across full altitudinal and seasonal gradients, including high-altitude migratory movements, while accounting for variability in effective altitude coverage driven by meteorological conditions.

The framework also outlines a workflow for analysing coupled acoustic–radar datasets using both supervised and unsupervised learning approaches. By linking species identity to radar-derived kinematic and micro-Doppler information, this methodology opens new perspectives for large-scale monitoring of bat ecology and migration dynamics. Beyond ecological research, the approach holds potential for future applications such as assessing species-specific interactions with wind energy infrastructure and improving models of aerial wildlife movement.

Compact entomological lidars for comparative seasonal surveillance of free flying insects over ground

Katrine Gerassimovitch Eskildsen, Leonard De Causmaecker, Timotej Zuntar, David Dreyer, Meng Li, Mikkel Brydegaard

Insects Biodiversity monitoring Lidar

While entomological radar shows great potential for surveying migratory insect biomasses and fluxes on continental scale, radar opportunities for species specificity and thus biodiversity assessment of foraging insects close over ground are limited. Malaise trapping campaigns and subsequent genetic barcoding suffer from trap design species biases, slow weekly sample rates and inability to accurately resolve abundance of present species. Photonic sensing of insects could complement existing techniques, with unbiased view of insect activity, differentiation of hundreds of classes, and provide clues on the species-abundance-distribution. Further, accurate timestamping allows estimation of environmental preference with potential for climate predictions.

We present a compact eye-safe weatherproof entomological lidar for non-invasive monitoring of foraging insects over ground. The system is based on off-the-shelf components and 3D printed parts, simple to reproduce and align, with a footprint of 60x30x12 cm and power consumption of 50-100 W. The aim is to deploy several systems in field for comparative seasonal monitoring of diversity of free flying insects. The instrument uses a time division multiplexed near-infrared beam and the Scheimpflug configuration to achieve ranging, focal depth and high sample rate. The design range is 5-100 m. Wavelength bands are alternated at 8kHz, while the instrument captures multispectral backscattering from insect bodies and oscillatory cross sections and harmonics from wingbeats. These echoes allow to differentiate species according to their melanization, wing surface roughness and membrane thicknesses. Observations are automatically preprocessed and transferred remotely. This allows for real-time, continuous and long-term monitoring of insect activity. Unsupervised hierarchical clustering is used to estimate signal diversity and thus species richness.

We present the first signals and results from testing *in situ*. We revise system performance, noise levels, detection ranges and insect sizing. We evaluate probe volume, daily observation numbers and specificity in terms of number of discernible signals.

We touch upon an outlook and opportunities of a distributed lidar network for insect diversity monitoring and discuss challenges for implementation and upscaling.

Integrating localized flight call detections and radar-derived trajectories to study nocturnal avian migration

Mason Maron, Benjamin Van Doren

Birds Tracking radar Methodological advances Biodiversity monitoring

Approximately 3.5 billion birds migrate across the United States annually, with the vast majority traveling nocturnally. In spite of this, we still lack a fundamental understanding of how nocturnal migration is structured, organized, and impacted by characteristics of the landscape below. Furthermore, research techniques, such as nocturnal flight call monitoring and weather radar, are increasingly popular among the scientific community, but are not fully validated for passage rate estimation. To expand our knowledge and toolset in this field, we established a ground-based acoustic planar array, consisting of four microphones placed equidistant in cardinal directions, coupled with a FaunaScan MR2 X-band radar in the center. We deployed this setup in an agricultural region of Illinois in the US. We simultaneously collected data on migratory bird trajectory, shape, size, spatial behavior, vocalization, and taxonomic identity, assigning localized flight call detections and their corresponding taxonomic identities to radar-tracked flight paths. Using this output, we provide early findings of the social, spatial, and behavioral structure of nocturnal bird migration. Additionally, we identify the strengths, drawbacks, and accuracy of acoustic monitoring and radar as accessible tools for migration research and management. Our findings and those that continue to develop from this work provide valuable insight into the nature of nocturnal bird migration and further refine existing methodologies used to study it.

Radar detection of avian responses to sonic booms

Maarten Reyniers

Birds Weather radar

Sonic booms generated by supersonic aircraft produce abrupt acoustic disturbances that may trigger collective responses in wildlife. On 16 December 2025 at 13:20 UTC, an F-16 aircraft crossed the sound barrier near Leuven, Belgium. Weather radar data from the Helchteren radar (VMM) were examined to assess whether this event produced a detectable biological signal, analogous to radar-observed bird departures during fireworks. The radar imagery shows a brief, spatially coherent enhancement of biological echoes consistent with a sudden take-off of birds coincident with the sonic boom. To our knowledge, such a radar-detected avian response to a sonic boom has not been previously reported. This preliminary observation suggests that sonic booms may be detectable in operational radar data via their biological impacts, and motivates a broader search for similar signatures in other radar archives.

Insect identification in weather radar data

Paraskevi Nousi, Birgen Haest, Michele Volpi, Elske Tielens, Silke Bauer

Bats Birds Insects Weather radar Biodiversity monitoring

To expand the use of weather radar data for entomological applications, improved differentiation between classes of bioscatterers is necessary. Within the BioAIRad project, we develop machine learning algorithms to classify bird and insect movements weather surveillance radar data, and we explore the flexibility of such algorithms for transfer between S-band and C-band. Our approach is incremental, starting with classifying voxels from polar volumes of radar data into eight common categories of bioscatterers. We trained several ML models using existing annotated data from the US NEXRAD network, supplementing with new annotations from radars based in the Netherlands to evaluate models’ performance on this data, which is completely unseen during training. Our results indicate that we can differentiate between vertebrates and arthropods as well as precipitation using different ML models, though performance degrades when using a more fine-grained level of classification labels, i.e., when trying to differentiate between different kinds of insects or birds. Furthermore, the predictions are spatially non-cohesive and noisy, as each voxel is seen individually by the models, separated from its neighborhood, both during training and testing. Finally, when deploying the model on radars outside of its training data, we observe out-of-distribution phenomena that require additional attention to fix, such as low performance on precipitation detection. In a next step, we use the existing, sparse, voxel-level annotations to train 2D convolutional neural networks on entire volumes of data to better capture spatial correlations in the data. In the final step, 3D neural networks will be developed to capture temporal information, in an effort to better differentiate between species based on the patterns of their movements.

Identifying potential ecological traps for migratory insects along the West Coast of North America

Peter Coggan, Santiago Ramirez

Insects Weather radar Spatial temporal movement processes

Ecological traps likely create unique dynamics in community structuring. These traps occur when animals settle in areas that will reduce their fitness. Aerially migrating insects may be particularly at risk because their movement heading is controlled by the wind. If the wind is blowing in a non-advantageous heading, insects risk either traveling in potentially dangerous directions or must wait until the wind shifts. Early work on migratory insects posited the Pied Piper Effect (PPE) where communities of insects would get blown north and settle in areas that would later become fatal due to the dispersing community’s inability to weather harsh winters. Subsequent work has shown limited evidence for the PPE for many aerial dispersing communities. However, shifts in wind patterns due to climate change are leading to phenological mismatches between migrants and their needed wind heading. This results in species getting trapped north of their overwintering range potentially causing a PPE. The West Coast of North America offers a unique space-for-time substitution to test how aerial insect migrants adapt to non-advantageous wind patterns because many areas experience seasonally fixed wind patterns due to local geography. The goal of this project is to test how aerial insect abundance, heading, and phenology are structured at locations where wind has static seasonal mean headings. This work will take advantage of the Next Generation Weather Radar Network (NEXRAD) to monitor aerial insect movement and magnitude across the West Coast. The heterogeneous windscape of the West Coast allows for the comparison between weather radar stations in the same region with likely similar aerial insect species compositions but differ greatly in the average wind heading and variability. These data will not only be useful in understanding how aerial insect migrations are influenced by complex geography but also elucidate their phenology and magnitude. The West Coast is very valuable for both agriculture and conservation, making knowledge of species migratory routes important for basic and applied questions.

Simultaneous offshore and inshore radar recordings bring new insights into bird migration at the Gulf of Lions scale

Vincent Delcourt, Camille Assali, Dorian Chauvin, Vincent Liebault, Cyprien Daide, Hélène Schopper, Nicolas Delelis

Birds Other radar Applied radar aeroecology Spatial temporal movement processes

While bird migratory displacements are informed by visual counts since decades, at-sea distribution of migratory bird in the Western Europe, and especially across the Mediterranean Sea, is poorly known. However, significant development of offshore windfarms is planned in the French waters of the Mediterranean Sea, and especially in the Gulf of Lions, raising crucial needs to acquire improved knowledge about bird migratory patterns at this scale.

Combining for the first time concomitant inshore and offshore radar surveys conducted during 3 years within the MIGRALION programme, we got an unprecedented view of bird migratory movements at the Gulf of Lions scale.

Data was collected with 3 vertical looking radars, deployed simultaneously on the coast (Birdscan MR1, 2 units) and offshore (marine radar, 1 unit). The offshore radar was especially installed onboard a dedicated vessel, conducting more than 1000 km-long transects over the whole Gulf of Lions during 3 days and nights (2 times in spring, 2 times in autumn). Surveys were conducted during 3 years, covering 3 spring migrations and 3 fall migrations. Birds were recorded up to 1500 m agl/asl, informing fluxes (Migration Traffic Rates) and flight heights.

Data analysis provided direct comparison between at-sea and inshore movements and gave new insights about bird migration at sea. Cross-analysing different radar sources across the Gulf of Lions allowed us to inform characteristics of migration flows in terms of phenology, spatial and altitudinal distribution of birds. In addition, this comparative study allowed to evaluate the representativity of offshore marine radar surveys in characterizing at-sea bird migration, while inshore ornithological radar could provide precise quantification of migration on the coast with continuous data recording over the whole 3-years programme.

While windfarm development in currently conducted in the area, the observed proportion of the avian migrating community potentially at risk at the Gulf of Lions scale is evaluated, in the light of these new elements on migrating birds at-sea distribution.

Synoptic state alters flock migration patterns through a bottleneck in the Levant

Yuval Werber, Asaf Hochman, Nir Sapir

Birds Applied radar aeroecology Spatial temporal movement processes

Bird migration is known to be affected by weather conditions, but little is known about the relationship between synoptic conditions spanning hundreds of kilometers and bird migration. Synoptic types can be considered a capsule of all meteorological variables, including pressure, wind intensity and direction, temperature, and humidity. In the Eastern Mediterranean, each day is classified into one of several synoptic types: Cyprus Lows, Persian Troughs, Highs, Red Sea Troughs, and ’Sharav’ Lows. These alternate along the migration season, creating a temporal mosaic in the meteorological landscape where widely different atmospheric conditions prevail daily.

The diurnal migration of large soaring birds is known to depend heavily on specific weather conditions, particularly for thermal soaring. Given their size and the length of their migration, many are physiologically unable to complete their journeys without significant support from rising air and tailwinds. It is consequently expected that these migrants should be able to distinguish between poor and fine weather, make use of supportive conditions, and avoid detrimental ones.

Here, using the daily weather-type classification of 16 migration seasons, we examine diurnal flock migration across different weather types in a bottleneck for soaring migrants on the Eurasian-African flyway and explore the potential of synoptic classification for migration forecasting and aviation safety applications.

Our results show that diurnal flock migration occurred across all synoptic types. However, migration had distinctive characteristics in each type: migration altitudes, flock sizes, and the overall number of flocks differed between weather types, producing consistent patterns across multiple years and various environments. By applying predictive modelling across a critical leg of the flyway at 1 km2 resolution, we demonstrate that the spatial features of migration differed substantially between weather types. Birds preferred to migrate along the coast in high-pressure systems and concentrated along the Great Rift Valley to the east in low-pressure systems. Our results suggest that birds strongly respond to synoptic-scale weather conditions and demonstrate the potential of harnessing widely used meteorological methodologies to predict bird migration, helping to manage its practical implications of bird migration for human life.