Abstracts of talks and posters presented at IRAC 2022.

On this page


On the radar: offshore wind turbine curtailment informed by nocturnal bird migration predictions

Maja Bradarić, Bart Kranstauber, Willem Bouten, Judy Shamoun-Baranes

Each year millions of birds migrate nocturnally over the North Sea basin, an ecological barrier within the East Atlantic flyway designated for large scale offshore wind energy development. While wind energy is the key to reducing CO2 emissions in the region, it can be harmful to aerial wildlife through collisions and barrier effects.

Understanding factors that influence migration intensity and flight altitudes over sea can be crucial for informing mitigation measures, such as temporary wind turbine shutdowns. We used dedicated bird radars located at wind parks off the western Dutch coast to measure migration intensity, speeds directions and flight altitude distributions. Using statistical and simulation models we studied the influence of weather on migration intensity and flight altitudes.We reveal that in spring migrants select atmospheric conditions dominated by high-pressure systems and tailwinds from the estimated departure area in the UK. In autumn supporting winds at departures areas in the north of the Netherlands, Germany and Denmark mainly drive departure decisions. We find that birds tend to fly higher in spring compared to autumn and both distributions overlap with the planned wind turbines.

Furthermore, we demonstrate that winds, temperature, and day of year are the most important predictors for the fraction of birds flying within turbine height. Finally, we use this knowledge to predict intense migration nights offshore 48 hours in advance to inform wind turbine curtailments, aiming to decrease wind turbines’ adverse effects on migratory bird populations.

Long-term trends of insect migration patterns above the USA

Jason Chapman, Adriaan Dokter

We provide an overview of a long-term, continental-scale study of insect migration patterns, using data collected by the NEXRAD network of 143 weather surveillance radars across the contiguous United States. We restricted our analyses to daytime data collected during a one-month period in mid-summer (mid-June to mid-July), in order to reduce the possibility of contamination of the insect signal by migrating birds and bats. We show that while inter-annual variation is large, migration during this period was consistently most intense along the central flyway, where movement was typically directed northwards from Texas into the Great Plains region. Analysis of the long-term trends in mid-summer diurnal migration intensity show that over our 20-year study period (2000–2019) there were interesting spatial patterns, but no evidence of an overall decline in insect abundance. We discuss the implications of these results.

Quantifying phenological trends in aerial insectivore roosting behaviors in the Great Lakes region using weather surveillance radar

Yuting Deng, Maria Carolina T.D. Belotti, Daniel Sheldon, Subhransu Maji, Jeff Kelly, Elske Tielens, Wenlong Zhao, Zezhou Cheng, Victoria Simons, Kyle Horton

In response to climate change, organisms have been shifting their phenologies to adjust to changes in the emergence of resources. Yet understanding these patterns at large scales and across long time-series is often challenging. To address this challenge, and to capture phenology at a system scale, we use radar remote sensing, specifically the US weather surveillance radar (WSR) network, to collect data on species-specific communal roosts. Annually, the discrete roosting behaviors of Purple Martins (Progne subis) are captured, more specifically, it is the morning departure from ground-based roosting locations that are detected by radars. To automate the detection of these roosting events, we applied a custom machine learning detector to capture these large-scale behaviors from radar scans 30 minutes before sunrise to 60 minutes after sunrise across the Great Lakes region. Across 12 NEXRAD weather surveillance radar stations in the Great Lakes region, we used 20 years of remote sensing data to quantify the phenology in roosting behavior of Purple Martins at the roost, radar station, and regional scale. We show that Purple Martins’ roosting phenology (50th percentile) has advanced by 2.26 days per decade on the regional scale. Similar advancing signals were found at station scale, but not as readily evident at persistent roost scale. While broad long-term phenology shifts were evident, the timing of annual phenology also coincided with long-term average temperature at regional scales: the warmer the regional temperature is, the earlier the 50% passage date is. Our study represents one of the longest-term broad-scale phenology examinations of an aerial insectivore response to environmental change and provides a stepping stone for examining the phenology mismatch in trophic interactions at even broad spatial scales.

SEMAFOR project: remote sensing of avifauna using the French meteorological radar network speed talk

Thibault Désert, Jordi Figueras i Ventura, Baptiste Schmid, Vincent Delcourt, Cécile Bon, Camille Assali, Nicolas Gaussiat, Silke Bauer

In order to reduce the loss of biodiversity in a globally expanding wind energy sector, the de-risking of wind farm projects is a major issue. Billions of birds fly across France every year on their way to optimal habitats depending on the season. Monitoring of migratory movements have made possible to precisely establish the phenology of common species, the major sites of importance and the abundance of birds. However, these routes are still very complex to trace, even more overseas than on land. Thus, an accurate tool for bird detection is a strategic issue for project planning and the issuance of alerts of imminent passage of migratory birds. Weather radars capture any target backscattering its electromagnetic pulse: from precipitations to hordes of insects and migrating birds. A network of weather radars therefore offers the possibility to study and quantify biomass over long periods.

SEMAFOR project: The ambition of the project is to develop a real-time observatory of migratory birds at high resolution from the French weather radar network, and to propose a tool to forecast migratory passages. The SEMAFOR project brings together institutions specialized in radar and biodiversity: France Energie Marine, Météo-France, Biotope and Vogelwarte. Firstly, existing algorithms for bird detection will be adapted to raw data from the weather radar network operated by Météo-France. The resulting algorithm will be calibrated and validated by comparing its output to the ornithological radars from Biotope. Once validated, the detection outcomes of the algorithms will be analyzed to describe flows, spatial location and temporal evolution of migratory movements on a national scale during a complete life cycle. Secondly, a predictive model for the probability of passage of migratory birds, especially at sea, will be developed by the Swiss Ornithological Institute. The model will take into account real time measurements from weather radars, but also local meteorological and environmental parameters and orographic obstacles, as well as knowledge of the main migration routes. At IRAC 2022, we will report on preliminary calibration results of bird detection algorithms based on polarimetric quantities from S-, C- and X-band weather radars.

Macro-demography of North America’s migratory birds

Adriaan Dokter, Jacob Socolar, Frank La Sorte, Ken Rosenberg, Ali Johnston

Standardized surveillance of the US airspace by weather radars has detected alarming population declines of migratory birds. Global anthropogenic changes are thought to be responsible, however where in the annual cycle these changes affect migratory birds most strongly is poorly understood. To provide more insight into the processes underlying population changes, we examined the annual variation in migratory passage across the US using information from 143 radars in the contiguous US. We use this analysis to estimate the population size estimates of the migratory avifauna multiple time throughout the annual cycle, in order to pinpoint whether population fluctuations are related primarily to variability breeding season recruitment or overwinter survival. We also present complementary macro-demographic analyses based on data from the eBird citizen science project, which complement the radar analysis with species-specific information.

From data to action: BirdCast perspectives on transforming bird migration science to conservation planning

Andrew Farnsworth, Adriaan Dokter, Kyle Horton, Frank La Sorte, Cecilia Nilsson, Dan Sheldon, Benjamin Van Doren, Julia Wang

Aeroecology unifies scientific disciplines, human behavior, and technology, with great potential to inform conservation, particularly regarding application of weather surveillance radar data. Here, we present two examples of outcomes from BirdCast, a research consortium of core partners from Cornell Lab of Ornithology, Colorado State University, and University of Massachusetts Amherst, that employ data collected by the US WSR-88D network to develop conservation actions that reduce human impacts on nocturnally migrating birds. Lights Out Texas is a community-driven effort with the goal of increasing awareness of and action to reduce collisions and protect the billions of birds migrating through Texas annually that leverages the power of ecological forecasting and radar to promote decision making. Based on the peer-reviewed research, the campaign can tell municipalities when to turn off their lights using radar-based migration forecasts and real-time analysis, as well as historical perspectives on when peak windows of migration occur for targeted messaging about conservation action. Despite its infancy, the campaign has already successfully educated and increased awareness, leading to management actions and policy wins including public proclamations to indicate support and urge residents and businesses to turn out non-essential lights at night during critical bird migration periods. Another example is the recent adoption of bird-friendly legislation based testimony to the NYC Council Committee on Environmental Protection in support limiting light pollution to protect birds. This legislation was passed unanimously, representing a tremendous win: New York is the first city to enact legislation that protects birds from the hazards of glass (previously enacted law) and light pollution. Combining and collaborating to use game-changing technologies, e.g. cloud computing, machine learning, with publicly available radar data and research can create powerful opportunities that benefit diverse stakeholders. Models and visualizations that predict and display bird migration in real time, over tens of thousands of square kilometers highlight capabilities of tools with power to inform and prioritize action as never before. We provide background and potential utility for BirdCast approaches and its potential applications across the planet where weather surveillance radar networks exist.

Autumn stopover hotspots and multi-scale habitat associations of migratory landbirds in the eastern U.S.

Fengyi Guo, Jeff Buler, Jaclyn Smolinsky, David Wilcove

Halting the global decline of migratory birds requires a better understanding of migration ecology. Stopover sites are a crucial yet understudied aspect of bird conservation, mostly due to challenges associated with understanding broad-scale patterns of transient habitat use. Here we use a national network of weather radars to identify stopover hotspots and assess multi-scale habitat associations of migratory landbirds across the eastern U.S. during autumn migration. We mapped seasonal bird densities over five years (2015 – 2019) from 60 radar stations covering 63.2 million hectares. At a coarse scale, we found that landbirds migrate across a broad-front with small differences in migrant density between radar domains. However, relatively more birds concentrate along the Mississippi River and Appalachian Mountains. At a finer scale, we identified radar pixels that consistently harbor high densities of migrants for all five years, which we classify as stopover hotspots. Hotspot probability increased with percent cover of all forest types, and decreased with percent cover of pasture and cultivated crops. Moreover, we found strong concentrating effects of deciduous forest patches within deforested regions to contain hotspots. We also found that the prairie biome in the Midwest is likely a migration barrier, with large concentrations of migrants at the prairie-forest boundary after crossing the agricultural Midwest. Overall, the broad-front migration pattern highlights the importance of locally-based conservation efforts to protect stopover habitats. Such efforts should target forests, especially deciduous forests in highly altered landscapes. These findings demonstrate the value of multi-scale habitat assessments for conservation of migratory landbirds.

Avian avoidance of offshore wind turbines measured by radar speed talk

Abel Gyimesi, Robert Middelveld, Jacco Leemans, Elisa Bravo Rebolledo, Daniel Beuker, Jente Kraal, Koen Kuiper, Ruben Fijn

The Dutch government targets to produce 16% of the generated energy sustainably by 2023. Consequently, it has been agreed that an additional 3,450 MW offshore wind energy should be realized on top of the 1,000 MW that has already been built. In 2016, a five-year governmental research program (Wozep) was launched to fill knowledge gaps in the ecological effects of offshore wind energy. One of the main objectives of the Wozep bird research is to develop a network of dedicated avian radars in offshore wind farms. As a first step in achieving this goal, in 2018 an automated 3D Fixed Robin radar, consisting of a horizontal and vertical radar, was installed in the offshore wind farm Luchterduinen at approximately 25 km from the Dutch coast, to continuously measure bird fluxes, avoidance-, as well as flight behaviour. To complement the radar observations with species information, we conducted visual observations locally in the wind farm from the beginning of 2019 up until the end of 2021. Based on the radar measurements in the same period, we have calculated overall mean traffic rates inside and outside the wind farm, pairwise at different distances from the radar, specified for different seasons, as well as diurnal periods. In addition, we have calculated flight intensities inside the wind farm at different distances from wind turbines, also defined at different time scales. Based on the observations collected during 46 field trips, we could identify associated species compositions. Consequently, the combination of radar and field measurements provided the opportunity to estimate values for species-specific avoidance rates for the macro- (avoidance of the entire wind farm) and meso-level (within the wind farm among the turbines). Answering such questions can improve our understanding of how birds react to offshore wind farms. Ultimately, such information is important for a sound design of eventual mitigation requirements in future wind farms, in order to be able to prevent high mortality levels among birds.

Spatiotemporal movements of insects across Europe, quantified using a vertical-looking radar network

Birgen Haest, Silke Bauer, Susanne Åkesson, Jason Chapman, Vincent Comor, Daniel Früh, Damiano Preatoni, Anna Nesterova, Judy Shamoun-Baranes, Felix Liechti

Each year, massive numbers of migratory animals move through the skies in pursuit of increased survival or reproductive output. While patterns of bird migration are increasingly well characterized, our current knowledge on insect movements remains fragmentary, even though the phenomenon surpasses other migrations in abundance and biomass, and has enormous implications for agriculture, economy, and human welfare. Over the past decades, the development of vertical-looking radar in particular has led to fascinating insights, but research has so far been limited to a single or a few locations. In a joint, cross-European effort, we created a network of 17 vertical-looking radars, stretching from Southern France to Finland, to quantify insect abundance and movements from March to October 2021. In this talk, we present patterns of migration traffic rates, biomass flux, and movement directions across these sites, and how they vary as a function of space, time, and habitat. We highlight patterns and differences in movement activity at various temporal scales, ranging from diurnal to annual seasonality. In addition, we identify the underlying drivers of these observed spatial and temporal patterns and variability by linking them to a set of environmental variables.

Patterns of flight altitudes for arriving spring migrants in the Gulf of Mexico region speed talk

Virginia Halterman, Adriaan Dokter, Andrew Farnsworth, Benjamin Van Doren

The Gulf of Mexico region serves as a major migration corridor every spring for billions of Neotropical migrants breeding in North America, with significant numbers of individuals arriving from over water into the southern United States. Technological and analytical advances in the last decade have allowed for regionwide, quantitative explorations of avian passage through this corridor. But this recent work has focused primarily on nocturnal bird migration, despite more than 75 years’ worth of direct and remotely sensed observations of diurnal migration in the region. Given the magnitude of movement through this region, we sought to characterize the diurnal component of migratory flights, particularly the altitude of arriving migrants from over water. Our study investigated and explored spatio-temporal altitudinal patterns in diurnal trans-Gulf migration over a 15-year period, examining changes in the average altitudes of migrants as detected by six WSR-88D stations and their relationships to prevailing wind patterns. On a daily basis, altitudes of migrants increased during the first few hours of diurnal periods, then plateaued, generally, until sunset; however, there was variation across the stations. Over the season a similar pattern of altitudinal increase occurred early in migratory periods, with a plateau beginning during peak movement periods and continuing through the end of spring movements. We found striking relationships with prevailing winds, particularly those of East-West component winds: eastern radar stations experienced increases in flight altitude as winds became more westerly, whereas western stations experienced the opposite pattern. Increasing southerly winds corresponded to increases in flight altitude across all stations, indicating that migrants flew at higher altitudes when tailwinds were present. Our results also highlight that the altitudes of arrival of migrants in the eastern Gulf were significantly lower than those in stations farther west, patterns that could correspond to migrants arriving from origin points closer to the radar stations that would be aloft for shorter periods before arrival. With projected changes in climate, habitat, and human activities, as well as declines in migratory bird populations, developing a greater understanding of movements of birds in this globally important migration corridor is critical. This is particularly important for relating the flight patterns of migrants to their passage and stopover.

Different migratory strategies in locusts and moths revealed by radar

Zhenhua Hao, Alistair Drake, Eric Warrant

An insect monitoring radar (IMR) was established at Bourke (30.0392° S, 145.952° E, 107 m above MSL) in inland NSW and provides a direct means of detecting population features including flight height, speed, displacement direction and orientation, and individual characteristics like body size, shape, and wing beat frequency. Utilizing these size and shape parameters and the fundamental wing beat frequency of the target, different types of nocturnally migrating insects (such as large moths and the Australian plague locust) could be classified according to previously established target-class selection criteria. Combining radar data with detailed meteorological information available from a high-resolution atmospheric model, we studied flight behaviour and its relationship to temperature and wind in migratory moths and locusts.

We found that these two insect taxa have markedly different responses to wind and temperature. Large moths generally select nights with favourable wind directions and fly downwind, but with orientations to the left or right of their track in an apparent attempt to fly in a direction that is not simply downwind, in this case a more south-eastward direction. In contrast, Australian plague locusts are less selective in wind and usually orient towards the southeast in spring and northwest in autumn; they nevertheless are carried in a wide range of directions by the wind. Large moths are able to tolerate a large range of air temperatures (> 10°C) while locusts are more selective, requiring warm nights with temperatures of at least 18°C. In both cases however, the flight speeds of both insects are usually less than the wind speed, so that their tracks are dominated by wind transport. We argue that these differences in behaviour are a manifestation of not only differing migratory strategies for survival in uncertain environments, but also differing sensory and navigational capacities.

Weather surveillance radar for biodiversity monitoring in the UK

Christopher Hassall, C. Hassall, M. Lukach, F. Addison, T. Dally, E. Duncan, W. Evans, M. Mungee, W. Kunin, J. Chapman, R. Lovelace, R. Neely III

Concerns have been raised about potentially widespread declines in insect abundance and diversity, but progress in evaluating these claims has been hampered by the lack of standardised monitoring data. Weather surveillance radars (WSRs) routinely detect insects, but since animals are not of interest to meteorologists, they are discarded as unwanted “noise”. That “noise” is a veritable treasure trove of information on insect diversity and abundance, but what is required is a way to link what a radar sees to the insects that we wish to monitor. This presentation will describe the ongoing work in the “BioDAR Project” that brings together ecologists and radar scientists in the UK to collaborate on a programme of work that will produce, test, and disseminate computer algorithms to turn radar noise into high-quality biological data. We will discuss three main areas of work: (i) the generation of a database of insect models to predict their reflections in radar data; (ii) empirical analysis demonstrating that radars can provide a proxy for in situ insect abundance and diversity; and (iii) maps of aerial insect biodiversity and abundance that can be used to investigate pressing issues in conservation. Our aim is to generate a biodiversity indicator that can be used to track insect populations in parallel with more traditional, ground-based measures of biodiversity.

High concentrations of migratory birds in Swiss-alpine valleys more frequent in pre- than postbreeding migration speed talk

Simon Hirschhofer, Peter Ranacher, Baptiste Schmid, Robert Weibel, Felix Liechti

Many migratory bird species increasingly suffer from population declines. Efficient conservation measures require further knowledge on migration strategies and route choices. The Alpine arc is one of the main environmental factors influencing the migratory divide within the European flyway. Therefore, understanding how birds react to this large structure is of great interest. We studied Alpine bird migration with three dedicated bird-radar devices. One was operating in the Swiss lowlands close to the northern foothills of the Alps, the other two were placed in two central Alpine valleys.

We found a high degree of alignment of the flight directions of birds inside the two Alpine valleys. This indicates that birds orient themselves on the course of those valleys and use them as passages to avoid high climbs over peaks and ridges. We detected the highest bird volumes in the Alps during the spring season, which stands in contrast to the lowlands, where autumnal migration was stronger. A correlation analysis between the sites revealed that the Alpine sites were highly connected to each other and to the lowland site during spring, but not during autumn. Previous work on the correlation of migration intensities across the Swiss lowlands showed high between-site connectivity during the autumn season, but not during spring. These results indicate that different flight routes over or around the Alps are preferred in spring than in autumn respectively. In autumn, a large proportion of migratory birds avoid crossing the Swiss Alps. When flying south, migratory birds tend to follow the Swiss Plateau towards the southwest. In spring instead, the concave, arched shape of the Alps could have a funnel effect on the returning birds towards south-facing valleys and eventually encourage a crossing of the mountains.

We therefore emphasize the great importance of high concentrations of migratory birds during spring migration in Alpine valleys and over passes. For the protection of migratory birds, consideration of such bottlenecks is of particular importance, especially with reference to the fact that additional mortality in spring has a much greater impact on breeding populations than during autumn migration.

Fireworks disturbance across bird communities

Bart Hoekstra, Willem Bouten, Adriaan Dokter, Hans van Gasteren, Chris van Turnhout, Bart Kranstauber, Emiel van Loon, Hidde Leijnse, Judy Shamoun-Baranes

Fireworks are important parts of celebrations and festivals globally. While public health concerns about fireworks are growing, little remains known about their effect on wildlife. The sudden synchronized lighting of fireworks on New Year’s Eve causes a strong flight response in birds. We show how weather radar and nationwide counts of birds wintering in The Netherlands can be used to quantify how flight response differs across bird communities and habitats, and determine the distance-dependence of this relationship. Accounting for differing radar cross-sections, we found that tolerance to abrupt disturbance is related to body size, with communities dominated by large-bodied species responding more strongly than those with smaller birds. On average, 100 times more birds were in flight during NYE than on regular nights, with local increases of 4-5 orders of magnitude. We found the effects of disturbance become indistinguishable from flight activity on regular nights only at around 10km from fireworks. At half that distance the average flight response is still elevated, but minimized and reduced to a factor of 10 above normal levels. Given the pervasive nature of this disturbance, mitigation can only be achieved by establishing large firework-free buffering zones or concentrating fireworks in urban centres. With millions of birds wintering in densely populated areas, conservation action should prioritize the most disturbance-prone, larger-bodied, bird communities.

Using radar to mitigate birdstrike risk in costal airport on Bohai Bay speed talk

Lili Jia

Bohai bay, located on the East Asian-Australasian Flyway, is proved to be critical sites for migrant waterfowls and raptors.Every year millions of birds flying over Bohai Bay during night for wintering or breeding, which poses high risk for aircraft safety. Despite that some species migration and birds biodiversity research have been carried out in civil airport of China, birds movement pattern, especially nocturnal movement pattern and how to mitigate the potential risk of wintering birds to aircraft is rarely studied.

In this study, we used a set of horizontal and vertical birds radar to monitor annual birds movement in an international airport along Bohai Bay, and testified methods to mitigate birdstrike risk.Combing with field survey, we found the flying pattern and roosting sites of wintering crows. We found the crows daily movement pattern is closely related with local sunrise and sunset pattern. Based on on-time monitoring and observations of crows behaviors, we used birds repellent devices to stop the crows fly across the taking-off and landing space during morning. Besides that, on the basis of analyzing the dominant species’ flying route and distribution hot spot, we assessed the potential birdstrike airspace. Meanwhile we conducted before-and -after experiments by using a self-developed software system to combine radar detection data with birds repellent devices and adjust the deployment of the devices to optimize the performance. In addition, we also analyzed nocturnal birds movement pattern and birds abundance during non-migration season and migration season, and analyzed birds migration pattern with meteorological factors and proposed using birds detection-warning-repellent system to monitor birds migration dynamic and mitigate potential risk. Bird migration timing and flyway is deeply impacted by meteorological factors.More accurate real-time birds migration pattern and early warning performance could be realized if local birds radar could be integrated with network of weather radar and network of global birds observation data set in the future.

Bin-based polarimetric echo classification for spatially flexible aeroecological purposes in combination with citizen science

Jarmo Koistinen, Nadja Weisshaupt

The demand for biomass monitoring by weather radars at various spatial scales requires reliable high-resolution target identification in a variety of environmental settings and for various types of migrations of aerial biota. Another challenge in using weather radars more effectively in ornithological research is the unknown species composition in the radar volume.

In the present work, we show the application of a novel supervised Bayesian classification methodology based on polarimetric moments from C-band dual-polarisation weather radars in aeroecology supported by citizen science data. The key achievement of the methodology is the quantification of different types of aerofauna in each range gate at a spatial resolution of about 1 km. As a second important benefit the methodology simultaneously separates a range of meteorological and non-meteorological target types. The bin-based methodology allows for spatially more flexible high-resolution applications in a broader set of migration types compared to existing layer-based approaches. The method enables spatially distributed retrieval of bird and insect information also in cases when precipitation occupies parts of radar scan volumes. We compare the performance of the polarimetric methodology to established standard methods used in radar biology and discuss its extended potential for biodiversity monitoring in different types of migrations which unveil novel areas of application.

Incorporating seasonal differences in phenology is essential for accurate bird migration forecasts

Bart Kranstauber, Willem Bouten, Hans van Gasteren Bouten, Judy Shamoun-Baranes

During their annual migrations billions of birds may encounter human infrastructure. To mitigate the risks of encounters for both humans and birds, temporary measures such as wind turbine shutdowns and flight safety warnings are implemented. To implement these mitigation measures forecasts for the expected intensity of migration are essential. We use weather radar data to monitor migration as these are an especially powerful tool since they operate continuously and cover a large area. Using a phenological model as a baseline, we develop a predictive model for bird migration designed for flight safety warnings for military aviation. Using 10 years of migration data within the Netherlands we predict the density of migrants based on weather conditions and the phenological model. Predictions from the ensemble model are evaluated for years omitted from the dataset and compared between the spring and fall migratory season. We find differences in both the important environmental variables and the predictability between seasons. In autumn the accumulation of migrants due to wind conditions is more important and environmental conditions contribute more to consistent predictions. We further investigate variation in diurnal migration phenology across Europe with respect to major migration barriers and season. We find spring migration is more uniformly distributed across the night in contrast to autumn migration which happens more in the first half of the night. Peaks around sunset are common in coastal areas. These findings suggest that underlying differences in phenology can and should be incorporated into mitigation measures to reduce conflicts between migratory birds and human activities or infrastructure.

Patterns in high-altitude bat movement over Texas revealed by radar speed talk

Jennifer Krauel, Gary McCracken, Andrew Farnsworth, Birgen Haest, Kyle G. Horton, Felix Liechti, Cecilia Nilsson, Baptiste Schmid

Movements of insectivorous bats foraging in open space and at high altitudes are very poorly understood. Advances in technology such as radar and telemetry have provided some clues. For example, Brazilian free-tailed bats (Tadarida brasiliensis) fly over 3km above ground level (AGL) after leaving their cave roosts, but we do not know if they maintain those altitudes while foraging, or if they forage in proximity to other bats. Some movement patterns may depend on factors related to insects, such as their location, abundance and diversity, whereas other patterns may depend on factors related to bats themselves such as phenology of juvenile flights and density of foraging bats. These patterns vary within and among nights and seasons. We installed a vertical radar in an area with many nearby colonies of T. brasiliensis in southern Texas, where large numbers of bats forage over agricultural fields. We analyzed data from three seasons in 2018 and compared them to data from a nearby weather radar. We characterized bats’ behaviors in several ways, including finding that bat activity peaked in most seasons at approximately 200m above ground level though bats were active to at least 1600m from April-November. We report distributional patterns of bat activity between 50-3,000 m above ground level. Understanding these foraging movement patterns is crucial for bat conservation efforts, because flights within range of large maternal colonies occur at altitudes matching threats from growing wind energy facilities.

Tracking desert locust swarms using dual-polarisation weather radar

William Kunin, Thomas Dally, Christopher Hassall, Maryna Lukach, Ryan Neely III, Freya Addison, Mamoon AlSaraiAlalawi, B. Arul Malar Kannan, K.C. SaiKrishnan

Desert locusts (Schistocerca gregaria) are the most dangerous of all migratory pest species globally; a typical 1 km2 swarm eats as much food per day as ca. 35,000 people. The damage caused by locust swarms makes tracking their movements a matter of substantial public interest. Dual-polarisation weather radar may provide a novel tool for tracking locust movements. Radar signals respond not only to the presence of materials in the air column, but also to the size, shape and orientation of the objects detected; the large size and elongated shape of locusts, combined with their tendency to fly above ground vegetation in huge swarms containing many millions of individuals, makes them ideal subjects for radar detection. This talk will describe our successful efforts to detect, discriminate, and track airborne locusts in Oman and India during the desert locust outbreaks of 2018-2021, as part of the Bill and Melinda Gates Foundation “PestDAR” project. These methods provide an initial data analysis pipeline for the integration of locust detection into weather radar analysis. We also demonstrate that swarm characteristics can be extracted, including approximate area, density, and the speed and bearing of the swarm while in range of the radars. With a growing weather surveillance radar network across regions of locust activity, we hope that these methods will be of use to the wider locust monitoring and mitigation community.

Integrating deep learning with mechanistic modeling for spatio-temporal migration forecasts based on weather radar networks

Fiona Lippert, Bart Kranstauber, Patrick Forré, Emiel van Loon

Weather radar networks make it possible to continuously monitor migratory movements of birds, bats and insects over unprecedented geographical expanses. This provides opportunities for developing large-scale migration forecasting systems that predict spatio-temporal movement patterns in near-real time and thereby allow for systematic prioritization of conservation and conflict mitigation efforts. However, the irregular and sparse spatial distribution of weather radars poses a challenge to spatio-temporal predictive modeling of migratory movements.

We tackle this challenge with a hybrid approach between data-driven and mechanistic modeling. The resulting model, called FluxRGNN, is a recurrent graph neural network that is based on a generic mechanistic description of population-level movements over the Voronoi tessellation of radar locations. Unlike previous approaches that rely solely on local associations between environmental conditions and migration intensity, FluxRGNN integrates both the spatio-temporal dependencies between the radars which arise from the movement process and environmental responses of migrants.

We use our model to make 72-hour forecasts of nocturnal bird migration over Western Europe using vertically integrated measurements from the European weather radar network. With its spatial and temporal components, FluxRGNN predicts future bird densities more accurately than local regression models. Due to the underlying mechanistic model, it can additionally disentangle aerial movements from local take-off and landing dynamics, providing a comprehensive picture of the migration process. We validate the model based on simulated data and show how it can be used in practice to generate interpretable migration forecasts that allow stakeholders to, for example, reason about the spread and accumulation of pathogens like avian influenza. Eventually, hybrid modeling approaches like FluxRGNN could help advance our understanding of migration systems and identify inconsistencies in the radar measurements.

Impacts of the differential phase upon transmission on radar variables from birds

Valery Melnikov, Precious Jatau, Tian-You Yu

Polarimetric weather radars are a powerful tool for studying flying birds. Most of the radars employ Simultaneous Transmission And Reception of orthogonally polarized waves (STAR radar design). In this design, the phase between the polarized waves is not controlled and can be arbitrary upon transmission. It is shown in this communication that the differential phase upon transmission (DPT) affects the polarimetric radar variables from flying birds. Results of the DPT-impacts has been obtained using genuine shapes of birds’ bodies and software for scattering electromagnetic waves from arbitrary shaped scatterers. Results for flying Canadian Gees, Mallards, and Purple Martins are discussed. The DPT strongly affects values of differential reflectivity, the differential phase, and correlation coefficient measured by STAR radars. Analyzed radar data include observations of migrating birds from adjacent NEXRAD radars having different DPTs. The DPT for a given radar should be considering in the quantitative analysis of radar data. An attempt of identification of the mean birds’ size using STAR radar variables is discussed.

Aerial and terrestrial biomass flows of migratory birds across the US

Raphaël Nussbaumer, Adriaan Dokter

Quantifying and characterising the biomass flows of bird migration, and their interaction with terrestrial habitats, is critical to understand the influence of mass migration on habitats, ecosystems and human societies, especially in light of rapid population declines of common nocturnal migrants.

While the use of radars has allowed researchers to study how birds move through the airspace, the movement of birds from and to the ground has received, in comparison, little attention. Using a model inspired from the field of fluid dynamics, we estimate the flows of birds taking-off, flying and landing each night across continental USA over 25 years. Cumulating these fluxes, we can summarize the change in the number of birds on the ground over the seasons and the entire year, quantify the magnitude of subsequent waves of bird migration between nights in the air and on the ground, and identify regions that see major biomass movements.

Investigating terrestrial-aerial relationships in an integrative framework has the potential to quantify how aerial movements produces fluctuations in bird numbers on the ground at a daily timescale, and we evaluate the extent to which changes in arrival and departure lead to differences in reporting rates by field observer of the citizen-science project eBird . Providing insight into the connection between departure, movement and landing also provides insight into the environmental drivers behind stopover and migration decisions, and will contribute to conservation decisions for prioritizing important stopover habitats that is informed by the duration in which birds use these habitats.

Using spatio-temporal information in weather radar data to detect communal bird roosts

Gustavo Perez, Wenlong Zhao, Zezhou Cheng, Subhransu Maji, Daniel Sheldon

Weather radar data provide detailed information to study animals in the atmosphere. Specifically, the US weather radar network holds more than twenty-five years of archived data. We propose a detection system for communal bird roosts that leverages temporal information from consecutive weather radar images.

Recent advances in deep learning allow transfer learning from convolutional neural networks pre-trained on large-scale image datasets to different domains and tasks. However, it is not clear how to use transfer learning for a new task where the shape of the data differs from the images used for pre-training, since the learned parameters are linked to the network architecture and the dimension of the inputs. In particular, a radar volume scan has many more channels than a natural image. A way around this is to manually select a subset of channels at the cost of a potential loss of information from the discarded channels. We propose a simple linear adaptor layer to map a larger number to three channels to take advantage of pretrained models while exploiting all the information available in the radar data. In addition to the information available in a single time frame of the radar data, we use consecutive frames as additional channels to train our network. We show that using temporal information reduces false positives caused by rain and static structures with little additional memory and computational cost.

Two radars in the desert - Scrutinizing bird migration dynamics over different timescales using retrospective comparison and operational prediction

Nir Sapir, Inbel Schekler, Yuval Werber, David Troupin, Yoav Levy, Felix Liechti

Thirty years ago, two superfledermaus radars were deployed in two different locations in the desert of southern Israel. These radars provide the only comprehensive information we have until today regarding the properties of bird migration in this region, which is located at the heart of one of the largest migration flyways in the world. In the last two years, we deployed two Birdscan-MR1 radars in the exact same localities, for two different goals. First, we wanted to compare the number of birds that migrate in this region nowadays and 30 years ago. To do that, we examined how the two radar systems, which have different beam properties and mode of operation, have been detecting targets, classifying them and estimating the number of passing birds. Our findings are very concerning, estimating a decrease of about 30% in the number of migrants over the last 30 years period, reflecting a roughly 1 percent average annual decline. If this trend will also continue in the future, we expect a worrying disappearance of bird migration in the coming decades, at least in this region. Second, we were interested to predict the intensity of migration of large birds a few days in advance, to provide warning regarding the risk of aircraft collision with birds in this region where low-altitude training of military aircrafts takes place. For this purpose, we developed a statistical model that explained the intensity of the migration of large birds by several meteorological factors and the date, using data from only a single migration season (spring 2020). The model produced quantitative prediction of bird Migration Traffic Rate up to 3.5 days ahead of time. To test its performance, we compared model prediction with actual recorded MTR from the radars as has been collected in the spring of 2021. The correlation between model prediction and actual bird MTR had an average correlation coefficient of 0.65. We will be operating these radars for the next several migration seasons for increasing the data used by the model, hopefully increasing its predictive ability. Developing a precise prediction model of large bird migration is essential for safe training by minimizing bird-aircraft collisions, thereby saving money and more importantly, human lives in addition to enhancing our understanding of the causes, mechanisms and patterns of large bird migration in this important migration flyway.

Automatic detection of bird flocks by weather radars

Inbal Schekler, Ilan Shimshoni, Tamir Nave, Nir Sapir

Application of automatic remote sensing tools such as radars allow gathering important information on bird migration at large spatial scales, ranging from tens to thousands of kilometers. However, most of the algorithms that have been developed to date focused on the patterns of nocturnal passerine migration. Many large soaring birds, such as raptors, storks, cranes and pelicans, migrate during the day, and have very different migration characteristics than those of passerines. Identifying and quantifying the migration of large soaring birds is important not only for better knowledge of their aero-ecology but also because they pose serious threat to aerial transportation. Indeed, strikes of large birds have caused many severe and sometimes devastating accidents with severe economic consequences and risk to human lives. In this study we tried to tackle this gap by applying machine and deep learning methods. We tagged thousands of radar images from three different weather radars in Israel. Using these images, we developed an algorithm that uses three different parameters from the radar data, as well as images from previous radar scans that facilitate the detection of bird flock movement at a range of 50 km from the radars. We present our preliminary results regarding the performance of the model in detecting bird flocks, suggesting recall of 0.9 and precision of about 0.6. Hence, the algorithm we have developed identifies more than 90% of the flocks that were passing through the radar coverage volume, while having a rather high false positive rate of detection, identifying a relatively large number of non-bird echoes as bird flocks. We expect to complete and improve the development of this tool within the coming months and suggest that the application of this model will revolutionize our capacity to detect, quantify and research soaring bird migration at a continent-scale, which is critically important for aircraft flight safety.

Prospects for monitoring bird migration along the East Asian-Australasian Flyway using weather radar speed talk

Xu Shi, Joshua Soderholm, Zhijun Ma, Cheng Hu, Jason Chapman, Richard Fuller

Each year, billions of birds migrate across the globe, and interpretation of weather radar signals is increasingly being used to document the spatial and temporal migration patterns in Europe and America. Such information is lacking in the East Asian-Australasian flyway (EAAF), one of the most species-rich and threatened flyways in the world. Logistical challenges limit monitoring of migratory birds in the EAAF, resulting in knowledge gaps on population status and site use that limit evidence-based conservation planning. Weather radar data have great potential for achieving comprehensive migratory bird monitoring along the EAAF. In this study, we discuss the feasibility and challenges of using weather radar to complement bird migration surveys in the flyway. We summarize the location, capacity and data availability of weather radars in different EAAF countries, as well as the spatial coverage of the radars with respect to migrants’ distribution ranges and migration hotspots along the flyway. More than 430 weather radars exist in the EAAF countries, covering half of bird species’ migration and wintering ranges on average, as well as 40% of important shorebird sites. We suggest that weather radar networks could monitor bird movement over the full annual cycle along the EAAF region, and provide estimates of migration traffic rates, site use and long-term population trends, especially in remote and less-surveyed regions. Weather radar networks would greatly complement existing ornithological surveys and help understand the past and present status of the avian community in a poorly documented flyway.

Spatial patterns in diurnal aerial insect biomass and landscape level drivers across the continental United States

Elske Tielens, Jeff Kelly

Aerial movement of high-flying insects is common and can provide significant transport of biomass, nutrients, and disease vectors. While the importance of regional insect movement for ecosystem services has been established, the magnitude and spatial characteristics of this flow above the continental United States have not been quantified. In particular, we lack standardized datasets of insect abundance and biomass at the regional scale, with existing data biased towards areas and species of conservation concern. To address this, we generated a ten year data set of diurnal high-altitude insect biomass across the continental United States from weather surveillance radar, collected from 135 widely distributed radar stations and spanning from 2012-2021. Using this standardized data set, we explore spatial patterns in aerial insect biomass. We also explore the spatial context of localized increases and declines in insect biomass.

Across the continental US, aerial insect biomass decreased with latitude and increased unimodally with longitude, with greatest biomass above the plains region. Temporal trends at the continental scale showed stable aerial insect biomass across years with high interannual variation. Long term trends varied across stations and biomes. While most stations showed increases in insect biomass, stations showing declines were more common in croplands and in the central and Mississippi region. Lastly, we use gradient-boosted trees to to predict insect reflectivity on radar across the United States from atmospheric conditions reported by the North American Regional Reanalysis. The model explains 72% of variation in noontime insect aerial biomass, with the greatest amount of variation explained by date, year, latitude, and longitude.

Our results give insight into the spatial dynamics of diurnal high-altitude insects above the United States. Moreover, we find that weather radar data can provide a standardized macroscale surveillance of the aerial biomass and movement of insects, with potential to address a broad range of questions in insect ecology.

Linking acoustic monitoring and radar to capture the timing and intensity of bird migration

Benjamin Van Doren, Vincent Lostanlen, Aurora Cramer, Justin Salamon, Adriaan Dokter, Steve Kelling, Juan Pablo Bello, Andrew Farnsworth

Monitoring small, highly mobile organisms poses a variety of challenges. Migratory birds undertake nocturnal movements of thousands of kilometers, often over inaccessible and inhospitable geography. Acoustic monitoring can complement existing remote sensing methods, such as Doppler weather radar, by providing information about individual behavior and species identities. However, the need for expert humans to review audio and to identify vocalizations are barriers to application and development of acoustic technologies. Here, we use an automated acoustic pipeline to monitor a complete autumn migration season. We validate measures of migration timing and intensity with contemporaneous data from Doppler weather surveillance radar and visual observations from citizen scientists. A model combining acoustic and weather data explained 75% of variation in radar-derived migration passage. Including acoustic data in the model decreased prediction error by 33%. Seasonal migration timing estimated by acoustic sensors was consistent with timing metrics derived from community science (i.e. citizen science) data at both the family and species levels. Automated acoustic technologies can document the magnitude and species composition of avian nocturnal migratory flights and can complement weather radar and human observation, especially in inaccessible and inhospitable locations.

Identifying fine-scale flight behaviour in 2D bird radar tracks: marine thermal soaring on the North Sea speed talk

Jens van Erp

Surveillance radars equipped with specialized tracking software can be used to track individual birds and quantify bird flight in a specific area. These observations are often used to provide aggregate statistics for operational purposes, like time-series of counts, average flight direction or average airspeed. The individual tracks do however contain much more information which can be used to investigate intricate flight behaviour of birds.

We show how the information from 2D bird tracks captured by the RobinRadar 3D-fix bird radar can be used to identify marine thermal soaring and the environmental conditions supporting this behaviour on the North Sea. Firstly, the radar tracks are filtered to obtain high-quality bird tracks by using radar and track properties. Secondly, we transform the bird track geometry to reflect the bird’s flight path relative to the air and use thermal soaring properties as measured in GPS tracks to assess the stereotypical circling behaviour often observed in thermal soaring. The method successfully identified thermal soaring at sea and we show the behaviour occurs when sea surface temperature is higher than the air temperature, which is facilitated by north-westerly winds carrying in cold air from higher latitudes. The methods we propose for extracting fine scale flight behaviour from radar should be applicable at other locations on land and at sea, and could be extended to identify different types of behaviour such as foraging flight. We briefly discuss the relevance of this methodology for interactions between birds and wind energy.

Validation of local bird radars used in military aviation

Hans van Gasteren, Karen Krijgsveld

Birds and aircraft share the same environment and can meet each other at the same time and position, resulting in so-called bird strikes. For military aircraft, these strikes occur in similar ratios locally (during take-off and landing) and en-route (during low-altitude flights). To prevent bird-aircraft collisions en-route, the Netherlands Air Force has been using radar systems since 1978 as an effective warning system for high bird migration intensities (https://www.flysafe-birdtam.eu/).

To help minimize local bird strikes, 7 bird radars were installed on RNLAF air bases and bombing range. We evaluated the accuracy and efficiency of these radars, by visually observing local bird flight activity on the air bases whilst simultaneously recording the corresponding radar results. A total of 2.600 bird flights were thus recorded during 80 hours of observations.

Here we present the result of this validation study. We show the effectiveness of the radars in detecting local bird movements and in classifying sizes of birds. In addition we show how we use bird radars to monitor seasonal migration activity as well as local flight movements of birds between roosting and foraging areas.

Flying across the sea – possibilities and limitations of weather radar data for the understanding of bird migration over the Baltic Sea

Arndt Wellbrock, Natalie A. Kelsey, Ommo Hüppop

For millions of landbirds migrating between Fennoscandia/Siberia and Central Europe within the East-Atlantic flyway system, the Baltic Sea represents an ecological barrier. To assess and mitigate human impacts on migrating birds, e.g. by offshore wind farms, it is necessary to identify the main migration routes across the different parts of the Baltic Sea. For this large-scale purpose, operational weather radar provides the only method to scan the airspace in a sufficient spatial and temporal resolution, especially for the nocturnal migration of passerines, but also of waders, ducks and geese. Using weather radar data from countries neighbouring the Baltic Sea, which provide data to the Operational Programme for the Exchange of Weather Radar Information (OPERA), we investigate the spatial-temporal patterns of bird migration across and along the Baltic Sea. We will present first results on the main offshore and coastal migration areas. For the peak of migration, we will show bird densities and migration traffic rates exemplarily for selected areas in order to detect main hotspots of migration under possible risk of human impacts. Moreover, we will discuss to what extend weather radar data can be used to interpolate movements across the sea as there is a limited spatial coverage of weather radar stations in different parts of the Baltic Sea, and technical difficulties (sea clutter, variable quality of radar data provided by national weather services) might impair quantifications. Our findings should help to focus future observation studies and environmental management on areas which were out of one’s “radar” so far.

Finding bats in RADAR data: a novel approach based on Ecology, movement and machine learning

Yuval Werber, Nir Sapir, Hadar Sextin, Yossi Yovel

The task of distinguishing bats from birds is a major challenge for undertaking taxa-specific applications using data from RADAR systems. Yet, visual detection of high altitude nocturnal activity of vertebrates is almost impossible, impeding the collection of training datasets for classifier development. Specifically, since bat foraging movements and long distance migration, which are highly influential for humans and the environment, are still largely understudied, there is an urgent need to develop bat classification capabilities. Here, we used patterns of aerial vertebrate phenology and behavior to assemble a large dataset of bat activity above a Birdscan MR1 vertical looking RADAR that is positioned in the Hula Valley in north-eastern Israel using clustering algorithms.

Our approach consisted of two stages. First, we labelled the data “Day” or “Night” according to times of sunrise and sunset, and used k-means analysis consisting of three clusters (aiming for insect, bird and bat clusters – the first two groups being both diurnal and nocturnal) in an attempt to detect a group which is active only during the night, presumably consisting of bat-only data. Once the existence of a predominantly nocturnal group was established, we proceeded to optimize the separation of the groups by filtering the dataset so that a bat labelled training data could be refined and used for classification into “Bat” and, “Not bat” signals. This was done by filtering out bird and insect activity based on phenology, wing flapping characteristics, morphology, and detection altitude thresholds of bat sized objects.

Using this novel dataset, we trained and tested 4 different classifying approaches on different combinations of RADAR feature that displayed good separation in the first stage and compared their performance. Data analysis reveals that the characteristics differentiating bats from other nocturnal aerial fauna are mainly related to size, speed and wing flapping. Based on these parameters we report successful separation of bats in RADAR data with over 90% accuracy using the bagged tree algorithm. In accordance to the unique anatomy of bat wings, it seems from our results that bats may have a typical wing beat pattern (pulse width, pause length relative to wing flapping frequency and speed), which is important in the separation between bats and similar objects.

A bioenergetic model to facilitate migratory landbird conservation along the northern Gulf of Mexico coast

Theodore Zenzal, Jeff Buler, James Cronin, Lori Randall, Wylie Barrow, Barry Wilson, Jaci Smolinsky, William Vermillion, Randy Wilson

In order to implement sound and effective conservation strategies for migratory landbirds, we need to know where birds stopover and how those stopover sites function (i.e., rest or refuel). However, even such basic information to inform conservation during the migratory period is limited, despite many migratory species in decline and mortality during migration being substantial. Conservation actions may be most critical at the edge of large ecological features, where birds must negotiate a potential long-distance, non-stop flight while contending with the normal challenges of migration. To provide managers with a decision support tool and to understand the function of stopover habitat for migratory birds, we developed an energetic model to determine whether there are sufficient food resources to support forest landbirds during their migratory stopovers in spring and fall across the U.S. Gulf of Mexico coast. To build the model and assess energy availability and demand, we used a combination of 1) field collected data on invertebrate, fruit, and nectar abundance, bird stopover duration, and forest composition, 2) bird food item energy content from the literature or lab analyses, 3) migrant forest landbird densities derived from weather surveillance radar, and 4) the 2016 NOAA Coastal Change Analysis Program land cover data. These data were used in a spatiotemporally explicit model to predict the energetic value of broad-leafed forests, estimate the energetic demands of migrants, and identify areas of energetic surplus or deficit within the region. Specifically, we estimated an average day’s food energy availability, bird energy demand, and their difference (i.e., energy surplus or demand) at a 240-m resolution across the entire U.S. Gulf coast for 4 semi-monthly periods during spring migration and 4 semi-monthly periods during fall migration. Furthermore, we estimate energy demand for circum-Gulf only flights, trans-Gulf only flights, and a hybrid approach (25% trans-Gulf and 75% circum-Gulf) based on existing radio telemetry data. By integrating field-based data collection with remotely sensed weather surveillance radar data, we have used a co-production process to develop a decision support tool that can prioritize important stopover areas and identify resource gaps to implement successful conservation practices.


Between sea and desert: soaring migrants beneficially modulate their flight in relation to sea-breeze

Paolo Becciu, David Troupin, Leonid Dinevich, Yossi Leshem, Nir Sapir

Environmental conditions influence the density and movement of flying migrants at different spatiotemporal scales. Soaring land migrants are often funnelling in so-called migratory bottlenecks and corridors, which are characterized by geographic features that allow them to minimize their cost of transport, compared to nearby areas. When flying over a migratory corridor, birds are still at the mercy of the weather conditions and could be subjected to adverse winds that may increase their flight energetics, slow them down or even terminate their flight. Millions of migrating soaring birds funnel between the sea and the desert twice a year along the Mediterranean coast of the Levant region. Especially during autumn migration, the birds encounter westerly winds due to the daily sea-breeze circulation. We were interested to understand how migrating soaring birds cope and/or exploit a changing lateral wind flow (sea-breeze) during their southward journey parallel to the coastline. We used automatically measured diurnal bird tracks recorded by the MRL-5 weather radar dedicated to bird tracking, which is located at Latrun in central Israel (34.98N, 31.84E) during eight different years between 2005 and 2016. Our results show the diurnal sea-breeze development, confirming that low-speed westerly wind increased in speed during the afternoon, as well as a clockwise rotation of wind direction from 90 to 135 degrees. This resulted in a constant increase of wind support (from 0 to 3 m/s) and a more variable eastward crosswind with increasing speed between morning to early afternoon, followed by a rather stable maximum speed for 1-2 hours and decreasing speed before sunset. Notably, birds decreased their airspeed throughout the day and modulated sideways speed by increasing it towards the direction of the wind as the crosswind speed increased, and decreasing it when crosswind speed decreased, allowing a relatively stable groundspeed and flying direction throughout the migration day. Our work highlights the flexible flight behaviour of soaring migrants, thereby enhancing the understanding of the movement ecology of soaring migrants, as well as its practical aspects, including bird-aircraft collision risks.

Combining radio-telemetry and radar measurements to reveal how insect abundance affects the movements of an aerial insectivore

Itai Bloch, David Troupin, Sivan Toledo, Ran Nathan, Nir Sapir

The movement properties of foraging animals are known to vary in relation to prey abundance, as described for terrestrial and marine environments. Yet, due to technological and logistical limitations, there is still lack of information regarding the foraging movements of aerial foragers. Specifically for insectivores, it is well established that insect abundance aloft is highly variable over time but it is not clear if and in what ways aerial insect predators respond to these dynamics. We used BirdScan-MR1 radar to estimate the Movement Traffic Rate (MTR) of insects over the Hula Valley in northeastern Israel and used tiny transmitters of the ATLAS system attached to Little swifts (Apus affinis) to investigate specific foraging movement properties. The unique combination of these two advanced systems allowed examining, for the first time, how the dynamics of insect densities affect the movement of aerial insectivores. We examined how insect MTR affected eight different movement properties of Little swifts during the breeding season. We found that insect MTR varied between different days by up to one order of magnitude. We found a decrease in the mean distance of the birds from the breeding colony but an increase in the length of daily foraging trajectory in days with higher insect MTR. Also in days with higher insect MTR, the rate of visits to the colony was lower and the arrival at the breeding colony at dusk was earlier. Additionally, the distance between individuals increased with higher insect MTR. We found no effect of insect MTR on the time of morning departure from the breeding colony, the flight speed of the birds, and the duration of visits at the breeding colony. The study shows that variation in the abundance of aerial insects influence key foraging movement properties, with possible impact on the breeding characteristics and potentially the reproductive output of swifts. In addition, it demonstrates the importance of studies that combine radar and biotelemetry methodologies for enhanced understanding of different ecological processes by providing a thorough exploration of predator-prey relationships in the airspace.

Bird migration forecast: developing radar-based forecasts of bird migration over Switzerland

Tom Carrard, Baptiste Schmid, Lukas Gudmundsson

Bird migration represents a massive flow of biomass that takes place twice a year between their breeding and non-breeding ground. The development of human infrastructure has led to increased conflicts in the aerial space. Artificial lights, high buildings, communication towers, and other infrastructures are thought to be responsible for an increased mortality during bird migratory flights. On the other hand, bird strike represents a source of danger and leads to economic loss. Forecasting mass movements of migratory birds can mitigate these negative interactions in the aerial space.

Existing forecasting systems combine weather radar measurements of migration intensity with meteorological data. This master thesis project aims to develop a forecasting system for nocturnal bird migration intensity for the Swiss Lowland using six years of dedicated small-scale radar monitoring (BirdScan MR1) in Sempach, Switzerland. Statistical models will be trained on migratory traffic rates in combination with meteorological data from ERA-5 reanalysis. The best performing model will be used to develop and display a real-time migration forecast using weather forecasts from ECMWF.

CROW: Visualize bird migration in your browser

Peter Desmet, Nicolas Noé, Robin Brabant, Maarten Reyniers

Every spring and autumn, millions of birds migrate over Europe. They mainly do this at high altitudes and at night, making this phenomenon largely invisible to us. But not for weather radars! We developed the open source web application “CROW” so you can explore these data directly in your browser. CROW pulls vertical profile data (vpts) from a public repository, calculates migration traffic rate (MTR), bird density and other variables, and visualizes these as interactive charts. The application can be hosted on a static file server and only visualizes data from one radar at a time, making it highly portable and scalable.

CROW was jointly developed by the Research Institute for Nature and Forest (INBO) and the Royal Meteorological Institute of Belgium (RMI) in collaboration with the Royal Belgian Institute for Natural Sciences (RBINS), with financial support from the Belgian Science Policy Office (BelSPO valorisation project CROW). It is deployed at https://www.meteo.be/birddetection to show bird migration in real time across the Benelux. We are planning to deploy it for data in the ENRAM data repository (https://enram.github.io/data-repository/) as well.

Aloft and on the ground: do bird radar data and constant effort bird ringing data of migrants in Eilat, Israel, correlate?

Tsafra Evra, Inbal Schekler, Noam Weiss, David Troupin, Nir Sapir

Israel lies at a bottleneck along one of the world’s most important migration flyways. Thanks to that, during migration season many birds can be seen around the country, both in the sky and on the ground. Monitoring of migrating birds is done in various methods, with constant effort trapping and bird radar surveillance being amongst them. The International Birding and Research Center in Eilat (IBRCE) is one of the birding centers in Israel that partakes in a constant effort trapping and ringing of birds over the last 8 years. Yet, it is unknown whether bird densities trapped on the ground correspond to the densities of actively migrating birds aloft. For exploring and measuring aerial bird densities we used data from a Birdscan MR-1 bird radar (Swiss-birdradar. Winterthur, Switzerland) and examined if there is a correlation between radar data in the form of Migration Traffic Rates (MTR) of nocturnally migrating birds in the area, and the number of birds trapped in the IBRCE. To answer this question, we used the constant effort ringing data from the migration seasons of 2016 (February-May and August-November) and compared it to data from the Birdscan-MR-1 radar that was positioned 3 km NNE of the IBRCE in 2016. Our preliminary findings show a positive correlation between the mean MTR of birds during the night and the number of birds trapped in IBRCE in the following day, but only during the fall migration season and not during spring, indicating a season-specific relationships between aloft and on-ground bird densities.

Migratory behavior of the fall armyworm in Israel and its outbreak potential in Europe

Fanqi Gao, Yuval Werber, Nir Sapir, Regan Early, Jason Chapman

The fall armyworm (Spodoptera frugiperda) is a major migratory noctuid moth originating in the Americas. The caterpillars of this pest can damage over 180 species of plants. The fall armyworm’s preferred host plant (corn) suffers tremendous damage and the yield can be reduced by up to 70%, potentially causing billions of dollars of economic loss. The fall armyworm has recently invaded almost every country in Africa and Asia, even as far as Australia. Although it has not yet reached Europe, it is already frequent throughout Israel. As Israel provides a natural link between Europe and Africa, there is concern that fall armyworms may soon reach southern Europe. In order to monitor the population trends, migration routes and the likelihood of invading Europe, research in this region is urgently needed. My PhD project will make use of data from a vertical-looking BirdScan radar and trapping in the Hula Valley of North Israel, and will focus on predicting the migratory routes and outbreak potential of fall armyworm in Europe. In this presentation, I will introduce my project and the main objectives of our research, and share some preliminary results.

Climatic drivers of Bracken Cave (USA) bat migration phenology and demography

Birgen Haest, Jennifer J. Krauel, Silke Bauer, Felix Liechti, Charlotte Wainwright, Phillip Stepanian

Changes in the timing of seasonal migrations have been observed across biological taxa, including birds, mammals, and insects. For birds, strong links have been shown between changes in migration phenology and changes in weather conditions at the wintering, stopover, and breeding areas. For other animal taxa, the current understanding of influences of climate change on migration timing remains limited. Many migratory animal populations also show strong changes in population sizes over the last decades, but in most cases the exact role of climate change remains to be teased apart from other driving factors. For many taxa, such as insects and bats, a major obstacle in furthering our understanding of how climate change affects both phenology and demography is the lack of long-term phenology datasets.

Bracken Cave in Texas (USA) holds one of the largest bat colonies of the world. Using weather radar data, a unique 23-year (1995–2017) long time series was recently produced of the nightly population estimates of Brazilian free-tailed bats (Tadarida brasiliensis) at Bracken Cave. Here, we use this dataset in combination with gridded weather data across North America to identify the combinations of weather variables, locations, and time periods that likely drive changes in the seasonal (1) migration phenology and (2) demography at Bracken Cave. We found spring migration phenology to be mainly driven by wind conditions during migration at likely wintering or spring stopover areas to the west of Bracken Cave, while autumn migration timing was mainly affected by precipitation to east and north-east. Variation in summer population sizes at Bracken Cave was largely explained by fluctuations in maximum summer and autumn temperatures from the previous year in central and southern Mexico. Winter population estimates were positively associated with prior spring and summer maximum temperatures in states to the north of Texas. These results illustrate how some of the remaining knowledge gaps on the influence of climate change on bat migration and abundance can be addressed by analysing historical weather radar data.

Developing automatic radar-controlled wind turbine shutdown to reduce collision mortality of local breeding birds.

Jonne Kleyheeg-Hartman, H.A.M. Prinsen, A. Potiek

Along the western edge of the Rotterdam harbor on the Dutch west coast, a wind farm of 22 wind turbines is being developed. This onshore wind farm is neighboring the largest breeding colony in the world for Lesser Black-backed Gull (±19,000 breeding pairs). One of the mitigating measures that is being applied concerns wind turbine shutdown for locally breeding Lesser Black-backed Gulls and Herring Gulls, using a dedicated Max® 3D bird radar within the wind farm. Each individual wind turbine will be stopped for a maximum of 50 hours per year. Because of this upper limit in shutdown time, it is important to employ the available hours of shutdown at moments with the highest collision risk to achieve the largest possible reduction in collision mortality. This asks for a very precise definition of the decision rules for shutdown and will ask the utmost of the capabilities of the bird radar.

Automatic radar-controlled shutdown for locally breeding gulls is very innovative and requires an intensive developmental process. The goal is to automatically stop individual wind turbines when the radar detects high-risk flights of the target species. The challenge is to define a ruleset with which the radar can automatically identify these high-risk flights. This means that we will have to 1) define specific characteristics of radar tracks belonging to the target species, and 2) define risk-areas within which flights of gulls might lead to a collision.

Bird radars are not (yet) able to automatically identify birds at species(group) level, so without further fine-tuning the wind turbines might end up being stopped for all kinds of larger bird species. The Max® 3D bird radar, however, records many track characteristics in high detail. In the breeding season of 2022, we start with site validation of radar tracks of gulls using visual observations and link information about species and bird numbers to radar tracks. With a large database of gull tracks we aim to identify characteristics of tracks of high-risk flights of the target gull species. These are used in the ruleset for shutdown. On our poster we will show the steps we take in the development of species- and turbine-specific radar-controlled shutdown.

This project offers a great opportunity to learn and develop state-of-the-art shutdown strategies that can significantly contribute to a wild-life friendly global energy transition.

Distinguishing birds from waves based on radar track characteristics

Jacco Leemans, Camiel Heunks, Abel Gyimesi, Jonne Kleyheeg-Hartman

Direct measurements on avian collision mortality in wind farms located offshore or in inland waters is challenging. Namely, fatality searches are not possible as victims fall in the water and sink. Therefore, currently collision risk models are commonly used to predict the number of casualties in these wind farms. The outcomes of these models are known to be sensitive to assumptions made about the behavior of the species concerned. However, data on avoidance behavior, fluxes, flight speeds and flight altitudes are scarce and could be highly location specific. Such data can be collected by specialized bird radars. These radars are capable of simultaneously tracking multiple birds in a relatively large area during large periods of time, including nighttime. However, in aquatic environments even the most advanced bird radars have difficulties in filtering out radar echoes caused by waves, which can substantially reduce the quality of the collected radar data and thus its usability. In 2021, a full 3D Max® bird radar was installed to monitor the effects of a large wind farm located in the IJsselmeer, the largest lake in the Netherlands with a surface area of 1133 km². This type of bird radar automatically collects and stores a large amount of characteristics of every track within its reach. During field visits, observers have identified the source of radar tracks, including birds and waves. The analysis of these tagged radar tracks indicates that waves can be distinguished from birds based on a combination of several radar track characteristics, like radar cross-section, track quality, track length, directionality, and altitude. Based on the first results of a model that classifies tracks based on these characteristics, we expect that more than 90% of waves can be filtered out. Such model could therefore considerably improve the quality of the collected data. Moreover, this opens up opportunities for (improved) real-time classification of waves, which could significantly increase the effectiveness of mitigation measures like radar assisted shutdown on-demand of wind turbines in periods with high bird activity.

Assessing the impact of fireworks on urban birds using L-Band staring radar

Joseph Wayman, George Atkinson, M. Jahangir, M. Antoniou, J. Reynolds, J. Sadler

Fireworks and other pyrotechnics have long been known as a source of disturbance for wildlife, highlighted by the fact that they, accompanied by their loud noises at discharge, are often used as wildlife deterrents. Consequently, events that utilise fireworks have the potential to impact local wildlife detrimentally and it is, therefore, important to investigate how fireworks affect local animal populations. Here we provide an initial exploration of the effect of fireworks on urban birds using an L-band staring radar as a tool to capture and characterise bird flight tracks within the city of Birmingham, UK. The radar covered a 90-degree sector out to a 5km range, from ground height to a 900m altitude. To examine how flighted behaviour changed during evenings/nights of excessive firework use in the area, we compared quantitatively the number, speed, and height of birds aloft during the same period in the days preceding and after New Year’s Eve (i.e., 28/12/2021 – 02/12/2022) and New Year’s Eve night (when fireworks were let off in large numbers at midnight and directly after). There were significantly more tracks in the period directly after midnight than in the same period on preceding and subsequent evening (389 compared to 20 and 25 respectively), observable as a peak in the number of targets aloft. Birds during this period of disruption flew higher than in the same period on preceding and subsequent evenings, but they did not fly for longer. Radar data indicated that birds flew at higher elevations and in greater numbers during times when they were subjected to intense disturbance from firework usage, at a point in their diurnal cycle when many would otherwise be roosting. This energy expenditure may have implications for avian fitness and survival, especially during the winter months when fireworks are typically set off and daily energy expenditure of birds is higher due to elevated thermoregulatory costs. Further work and long-term monitoring looking at which species are most impacted by such forms of disturbance are vital to identify mitigations such as more intense regulation of fireworks in urban spaces.

The impact of hydrometeors on bird migration as observed by various remote sensing systems

Nadja Weisshaupt, Mercedes Maruri, Maxime Hervo, Jarmo Koistinen

Meteorological conditions are a key modulator of bird migration dynamics. Besides wind, precipitation is considered a decisive antagonist to bird migration intensity. However, the role of precipitation might not be so straightforward and its detriment to migrants could be also conditioned by different degrees of visibility in precipitative conditions at various altitude layers. For example, spatial analysis of classified polarimetric weather radar observations in rain situations show that bird migration is present all around a precipitation area. Rain and bird echoes even coalesce, and birds may continue also below the clouds and thus in the rain. Besides, ceilometer measurements prove that low-level Stratus clouds with poor visibility are often present below the precipitating clouds. However, there is no research that has actually looked at bird migration during precipitative conditions in altitude and in relation to the visibility in precipitation. Also, the role of precipitation in migration studies is typically deduced from ground-based rain gauges, often with a coarse temporal resolution and ignoring conditions in altitude and time. The reason for that is obviously the failure of monitoring devices in precipitative circumstances, e.g. in most radars, bird signals are masked by rain and drizzle, and thus aeroecologists’ only mass migration study tool is rendered futile. However, time series data of low-frequency radar wind profilers depict bird echoes even in different kinds of precipitation. So, this provides an unprecedented opportunity to study the impact of various types of hydrometeors on migration dynamics in combination with other meteorological data from complementary remote sensing systems. Here we present preliminary findings of bird migration in different types of precipitative and cloud conditions as observed by radar wind profiler, ceilometer and rain gauges from two migration seasons.

Dealiasing of radial velocities based on interleaved dual-PRF measurements and bin-based polarimetric echo classification

Nadja Weisshaupt, Jarmo Koistinen

Weather radars provide amongst others measurements of radial velocities of targets. Radial velocities are reliable as long as a certain target type does not exceed the maximum unambiguous velocity, so-called Nyquist velocity, determined for a given pulse repetition frequency (PRF). Velocities of targets exceeding this limit are folded and require corrective dealiasing. Many dealiasing approaches have been developed which work well for wide-spread targets with uniform direction and speed, such as rain or insects (passive wind tracers). However, in many operational sweeps with low PRF these methods do not work well for birds with spatially heterogenous directions and speeds. Also, interference between mixed target types may affect the outcome of dealiasing, e.g. insect or rain contamination in a bird sample. Bin-based polarimetric target classification can help reduce the risk of mixed velocities from different target classes in one volume, though it does not avoid mosaic aliased velocity patterns or the Nyquist dilemma. On the other hand, non-polarimetric interleaved dual-PRF sweeps will give much better velocity estimates although they lack polarimetric information. Here we propose a novel method using a combination of interleaved elevation sweeps of bin-based polarimetric classification and dual-PRF measurements to circumvent dealiasing during bird migration and to obtain radial velocities for the calculation of migration traffic rates.

RADAR in a multisensory setup: a case study to assess the impact of an aerial deterrent

Yuval Werber, Gadi Hareli, Omer Yinon, Nir Sapir, Yossi Yovel

Remote sensing technologies operate within a certain scale at a given accuracy, preventing comprehensive data collection of animal aeroecology. To overcome this problem, we adopted a multi-sensor approach by uniquely combining RADAR, LIDAR and ultrasonic recording to test a novel technology for reducing bat mortality from wind turbine blades. Wind energy, a rapidly growing industry, induces mortality of flying animals and bats in particular. We tested an airborne deterrent which produces a pulsating combination of strong auditory and visual signals. The device was flown at an altitude of 100 m, to examine its impact at heights that are typical of many wind farm facilities. We used LIDAR to assess the device’s impact at altitudes below the deterrent height, corresponding to turbine nacelle altitudes, while RADAR assessed its influence at heights above 100 m, higher than turbine nacelle. Ground level and high altitude automated ultrasonic acoustic recorders carried by an aerostat were used to describe bat activity in the research site during the experiment up to 500 m AGL.

We recorded intense activity of multiple bat species throughout the experiment, revealing altitudinal and hourly activity trends and substantiating the intense use of bats, but not birds, of the site during the experiment. We revealed a significant ~40% decrease in activity below, and a significant increase above the deterrent’s flight altitude during the device’s operation compared to controls, implying that when exposed to the deterrent, bats avoided it by increasing flight altitude. We concluded that the deterrent proved highly efficient in reducing bat activity in a designated airspace. Interestingly, the sole use of LIDAR or RADAR for monitoring the effects of the device would have produced conflicting results, failed to disclose the actual pattern, and could critically bias our results and conclusions. Our work exemplifies the importance of undertaking comprehensive studies which are rarely feasible using a single sensor. We suggest that stationary RADAR operation sites can act as large scale “aerial field laboratories” where other technologies can be deployed to complement detection capabilities towards specific aims or during routine operation.