50 | Triennial Scientific Report
Monitoring bird migration by weather radar
Adriaan Dokter and Iwan Holleman
IntroductionWeatherandbirdmigrationareintimatelyrelated.Evolutionhasshapedthemigrationstrategiesofbirdssuchthattheyoptimallyrespondtoandmakeuseofweatherconditionsduringtheirlongdistancetravelsbetweenbreedingandwinteringgrounds.Bothwindandprecipitationstronglydeterminetheday-to-daytimingofmigrationandaltitudeusebybirds.
Spatiotemporalinformationonbirdmigrationisofinvaluableusetoscientistsandsocietyalike,butsofarnosensornetworkshavebeenestablishedthatcanmonitorbirdmovementcontinuouslyoverlargeareas.Inaviationbirdmigrationinformationisimportantforimprovingflightsafety.Especiallymilitarylowlevelflyinghasahighriskofen-routebirdstrikesandspatialbirdmigrationinformationisessentialforgeneratingreliableflightwarningstopilots.Comprehensivemonitoringofbirdmigrationatcontinentalscalescanalsoprovideinsightintomigrationpatternsandtheimpactonmigratoryflightofsynopticscalefactorslikeweatherandorography.
AspartofaninternationalprojectbytheEuropeanSpaceAgency(ESA)aimedatreducingcollisionsbetweenaircraftsandbirds,wehaveexploredthepotentialofoperationalDopplerweatherradarasabirdmigrationsensor.
Observing bird migration by Doppler weather radarOperationalweatherradarnetworksexistine.g.EuropeandtheUnitedStatesformeteorologicalapplications.ThesenetworkshavealargespatialcoverageasillustratedinFigure1,showingpartoftheEuropean
networkOPERA(OperationalProgrammefortheExchangeofweatherRadarinformation)1).Althoughweatherradarsareusedprimarilyforprecipitationmonitoring,alsobiologicalscattererscanbeobservedbythesesystems.Boundarylayerclear-airweatherradarechoesarecausednearlyexclusivelybyarthropods(most-lyinsects)andflyingbirds.Weatherradarsarethereforeapromisingsensorforprovidingbirdmigrationinformation.
Abirdmigrationquantificationalgorithmneedstoauto-maticallydistinguishbird-scatteredsignalsfromallotherechoesdetectedbyweatherradar.Birdechoesandotherclear-airsignalstendtobeconsiderablyweakerthanmeteorologicallyrelevantsignalsfromhydrometeors.Birds,insectsandhydrometeorsgiverisetosignalsinan
355 0 5 10
45
50
0 100 200
Kilometers
7
15
23
31
39
47
55
[dBZ]
EUMETNET/OPERA Radar Network
Wideumont
Trappes
Den Helder
De Bilt
Figure 1. Map of operational weather radars for part of Western Europe.
Radar sites are indicated by bullets, the weather radars used in this study
are labelled and coloured red. The OPERA reflectivit y composite is overlaid
for 19 April 2008 19.30 UTC.
Triennial Scientific Report | 51
overlappingreflectivityregime,whichmakesitchallengingtodistinguishthem.
Dopplerweatherradarsprovideinformationontheradialvelocityofscatterers,whichweusetoselectoutechoesrelatedtobirdmigration.Figure2showsweatherradarimagesduringintensebirdmigration(toprow)
andduringacasewithconvectiveshowers(bottomrow).Birdmigrationgivesrisetoacharacteristicallyhighspatialvariabilityoftheradialvelocityscandata,whichisnotobservedforechoesfromprecipitationorinsects.Unlikeprecipitationandinsects,birdsperformactiveflightwhichvariesinspeedanddirectionperindividual,causingahighervariabilityintheDopplervelocity.
0
90
180
270
0
90
180
270
0
90
180
270
0
90
180
270
Figure 2. Radar displays for reflectivity factor (left) and radial velocity (right) during a bird migration event (top rectangular box in green) and an event
with weak convective showers (bottom rectangular box in blue). The displays show the 1.2 degree elevation scan for a circular area of 25 km radius
around the weather radar in De Bilt. Contiguous areas of echoes identified by the algorithm are indicated within the reflectivity displays by red borders.
For the precipitation case (bottom) these areas removed from the data based on a low spatial variance of the radial velocities, while for the bird migration
case (top) these areas are retained based on a high spatial variance of radial velocities. After removing precipitation echoes average bird densities are
calculated from the remaining reflectivity data.
52 | Triennial Scientific Report
Wedevelopedabirddetectionalgorithm3,4)basedonexistingwind-profilingalgorithmsforDopplerweatherradars,usingtheVolumeVelocityProfiling(VVP)tech-nique5).Atargetidentificationschemewasdevelopedtofilteroutnon-birdechoesfromtheradarvolumedata,basedonananalysisofthelocalvarianceinradialvelocity.Thefilteredreflectivitydataisusedtoconstructabirddensityaltitudeprofile,andananalysisoftheradialvelocitydatayieldstheaveragebirdspeedanddirection.Reflectivitywasconvertedtobirddensitybyassumingaconstantbirdradarcrosssectionof11cm2,whichwefoundtobeanadequateapproximationduringnocturnalmigrationoverWesternEurope,whichisstronglydominatedbypasserines(smallsong-birds).
Bird radar field campaigns for validationIncorporationwiththeSwissOrnithologicalInstituteweusedadedicatedbirdradarofthetypeSuperfle-dermaus(seeFigure3)tovalidatetheweatherradarobservationsofbirdmigration.Thededicatedradariscapableofdetectingthewingbeatpatternforindividualradartargets6)(seeFigure4).Basedontheseechosigna-tures,insects,birdsandhydrometeorscanbedistinguis-hedwithahighselectivity.Themobiletrackingradaristhereforeastate-of-the-artreferenceforvalidatingtheweatherradarbirdobservations7).
Threefieldcampaignswereorganizedtovalidatetheweatherradarbirdobservations.Thebirdradarhas
beenstationedwithinthemeasurementvolumeoftheweatherradarinDeBilt,theNetherlandsfrom19Aug-16Sep2007,inWideumont,Belgiumfrom18Sep-22Oct2007andinTrappes,Francefrom10Mar-9May2008,thuscoveringafullautumnandspringmigrationseason.InFigure5thebirddensitiesaltitudeprofilesdetectedbybirdradar(toppanel)andweatherradar(middlepanel)aredisplayedfortheperiodof11-16October2007,whenthebirdradarwasstationedinTrappes.Wefindaremarkablecorrespondenceinthedetectedbirddensitiesbythetwosensorsduringallfieldcampaigns.Thealtitudedistributionsandabsolutenumberofdetectedbirdsmatchquantitatively.Weatherradarcanthusbeusedasareliablesensorforbirdmigrationquantification.
The effect of wind on bird migrationTheeffectoftheenvironmentalwindonflightaltitudesbecomesevidentbycomparingtheweatherradarbirddensityprofile(Figure5middlepanel)withthewindprofilescalculatedbytheHIRLAMnumericalweatherpre-dictionmodel(Figure5bottompanel).Wefindthatbirdsadjusttheirflightaltitudetomakeoptimaluseoftailwindsalongthepredominant(south-westerly)migratorydirection.Forexample,theHIRLAMwindprofile(bottompanelFigure5)showsthatonOctober4windconditi-onsathighaltitudewereunfavourable,wherestrongwesterlywindscausenegativewindeffects.Migrationthereforedoesnotextendabove1km(seemiddlepanelFigure5).Ontheotherhand,onOctober6tailwindsaremostfavourableabove1kmandasaresultalargefractionofmigrationtakesplaceathighaltitude.
Birdmigrationpatternsobservedatsinglesitescanstronglydependonweatherconditionselsewhere.Thisappliesparticularlytonortherntemperateclimate,wherefrequentpassagesofhighandlowpressuresystemscausealargespatialvariabilityinweather.Usingfourweatherradars(seeFigure1)we
Figure 3. The bird radar Superfledermaus equipped with a camera moun-
ted parallel to the radar antenna.
Figure 4. Wingbeat pattern of a small songbird with regular phases
of wingbeats (frequency around 15 Hz) and pauses as recorded by the
Superfledermaus bird radar. The single wingbeats are clearly visible. The
envelope of increasing and decreasing amplitude (vertical axis) of the
signal along the time axis (horizontal axis) reflects the birds flight entering
the radar beam at one edge, flying through the centre and leaving at the
other edge.
Weather radar can be used as a reliable sensor for bird migration quantification
Triennial Scientific Report | 53
02-Oct-07 03-Oct-07 04-Oct-07 05-Oct-07 06-Oct-07 07-Oct-07
Height-integrated bird density and wind profile
02-Oct-07 03-Oct-07 04-Oct-07 05-Oct-07 06-Oct-07 07-Oct-07Time
0
50
100
150
Bird
den
sity
[bird
s/km
2 ]
1
2
3
4
Hei
ght a
gl [k
m]
Bird radar
02-Oct-07 03-Oct-07 04-Oct-07 05-Oct-07 06-Oct-07 07-Oct-070
1
2
3
4
Hei
ght a
gl [k
m]
Bird radar
02-Oct-07 03-Oct-07 04-Oct-07 05-Oct-07 06-Oct-07 07-Oct-070
1
2
3
4
Hei
ght a
gl [k
m]
02-Oct-07 03-Oct-07 04-Oct-07 05-Oct-07 06-Oct-07 07-Oct-07
Weather radar
02-Oct-07 03-Oct-07 04-Oct-07 05-Oct-07 06-Oct-07 07-Oct-07
1
2
3
4
Hei
ght a
gl [k
m]
Weather radar
02-Oct-07 03-Oct-07 04-Oct-07 05-Oct-07 06-Oct-07 07-Oct-07
1
2
3
4
Hei
ght a
gl [k
m]
02-Oct-07 03-Oct-07 04-Oct-07 05-Oct-07 06-Oct-07 07-Oct-07
1
23
5
1015
25
50
100
200
400
Den
sity
[Bird
s / k
m3 ]
-16
-12
-8
-4
0
4
8
12
Ref
lect
ivity
fact
or [d
BZ]
Figure 5. Comparison of the bird densities altitude profiles determined by bird radar (top panel) and weather radar (middle panel). Integrated bird densi-
ties over all height layers are displayed in the lower panel for both weather radar (red) and bird radar (blue) (left vertical axis). The period between sunset
and sunrise is shaded in grey. Wind barbs in the lower panel show the wind profile from the HIRLAM numerical weather prediction model (right vertical
axis). Wind barbs in the middle panel indicate the bird speed and direction as retrieved by the weather radar algorithm. Each half flag represents 10 km/h
and each full flag 20 km/h.
monitoredthebirdmigrationflywayatseveralsitessimultaneously.Weatherradarbirdobservationsforthenightof19-20April2008aredepictedinFigure6.ForthisnighttheOPERAreflectivitycompositeisoverlaidonFigure1togiveanimpressionofthelargescaleweatherconditions.
Ateachsite,thetimingandaltitudeprofileofbirdmigra-tionisobservedtobeinfluencedbyweatherconditionsatlocationspassedearlierduringamigratoryflight.Asanexample,migrationatthemostsouthernsite(Trap-
pes)ischaracterizedbystrongdepartureandascenttohighaltitudesaround2.5km,wherebirdsexperiencefavourabletailwindsalongthepreferredmigratorydirection(NEinspring).Lowaltitudemigrationisavoidedbecauseofunfavourableeasterlylowlevelwinds.Atoneradarsitenorth-eastofWideumont,weobservetheappearanceofamigrationlayercentredaround2kmheightafter22UTC.Thismigrationlayerappearsafter4hoursofflighttime,whichcorrespondswithameasuredbirdgroundspeedof70km/htobirdshavingtravelledoveradistanceofabout280kmtowards
54 | Triennial Scientific Report
19-Apr-08 18:00 20-Apr-08 00:00 20-Apr-08 06:000
50
100
150
19-Apr-08 18:00 20-Apr-08 00:00 20-Apr-08 06:000
50
100
150
18:00 20:00 22:00 00:00 02:00 04:00 06:00 08:00 10:000
1
2
3
4
18:00 20:00 22:00 00:00 02:00 04:00 06:00 08:00 10:000
1
2
3
4Den Helder
19-Apr-08 18:00 20-Apr-08 00:00 20-Apr-08 06:000
50
100
150
18:00 20:00 22:00 00:00 02:00 04:00 06:00 08:00 10:000
1
2
3
4
18:00 20:00 22:00 00:00 02:00 04:00 06:00 08:00 10:000
1
2
3
4De Bilt
19-Apr-08 18:00 20-Apr-08 00:00 20-Apr-08 06:000
50
100
150
18:00 20:00 22:00 00:00 02:00 04:00 06:00 08:00 10:000
1
2
3
4
18:00 20:00 22:00 00:00 02:00 04:00 06:00 08:00 10:000
1
2
3
4Wideumont
19-Apr-08 18:00 20-Apr-08 00:00 20-Apr-08 06:000
50
100
15018:00 20:00 22:00 00:00 02:00 04:00 06:00 08:00 10:000
1
2
3
4
18:00 20:00 22:00 00:00 02:00 04:00 06:00 08:00 10:000
1
2
3
4Trappes
19-Apr-08 18:00 20-Apr-08 00:00 20-Apr-08 06:000
50
100
150
1 2 3 5 10 15 25 50 100 200 400Density [Birds / km3]
-16.0 -12.0 -8.0 -4.0 0.0 4.0 8.0Reflectivity factor [dBZ]
Den Helder
De Bilt
Wideumont
Trappes
Hei
ght A
GL
[km
]bi
rd d
ensi
ty [b
irds/
km2 ]
Time [UTC]
north-east.ThesebirdsmustthereforehavedepartedinthevicinityofTrappesinnorthernFrance,wherebirdschoseaflightaltitudeof2.5km.SinceWideumontislocated500mabovemeansealevel,the2kmaltitude
bandcloselymatchesthecruisingaltitudeafterdepar-turenearTrappes.Birdshavethusmaintainedaconstantflightaltitudeduringtheirmigratoryflightwithrespecttosealevel.ThealtitudeprofileobservedinWideumontisclearlyaffectedbytheparticularwindconditionsatthemoresouthernsiteofTrappes.InDeBilt,northofWideumont,nobirdsdepartintheearlynight.ThisisexplainedbytheweakocclusionfrontslightlysouthofDeBilt,whichblocksmostmigration.Withthisfrontweakeninginactivity,migrationcondi-tionsbecamemorefavourableoverthecourseofthenight,andafter00UTCwedoobservethepassageandarrivalofmigratingbirds.NomigrationisobservedonthemostnortherlylocatedweatherradarinDenHelder.Theextenttowhichbirdsareabletoadapttheirflightaltitudesinresponsetochangingmeteorologicalcondi-tionswillbetopicoffutureresearch.
ConclusionWefindthatDopplerweatherradarishighlysuccessfulindeterminingquantitativebirddensitiesandaverageflightspeedsanddirectionsasafunctionofaltitude.Wefindthatweatherradarreflectivitycanbequantitativelycorrelatedtothebird-densitiesdeterminedindependentlybybirdradar.Thedevelopedmethodsforbirddetectionandquantificationcanbeeasilyextendedtofulloperationalweatherradarnetworks.WiththeestablishmentoftheOPERAdatacentreforradardatawithinthecomingtwoyears1)theestablishmentofacontinent-widebirdmigrationsensornetworkinEuropeiswithinreach.Suchanetworkcanenableimportantapplicationsbothinflightsafety,health(aviandiseasespread)andenvironmentalimpactassessments.
Figure 6. Bird densities as a function of time and altitude at different
weather radar sites (see Figure 1) for the night of 19 April 2008. The wind
barbs represent the measured bird ground speed and direction. The lower
panel shows the height integrated bird densities over 0.2-4 km altitude.
The period between sunset and sunrise is shaded in grey and civil twilight
is shaded in light grey. The pink boxed inset on the right hand side of each
panel shows the HIRLAM wind profile at 00 UTC. Around 00 UTC a double
layered bird density profile is observed in Wideumont. We attribute the top
band to birds that departed in the vicinity of Trappes, while the lower band
results from birds departed at closer distance.
The establishment of a continent-wide bird migration sensor network in Europe is within reach
Triennial Scientific Report | 55
References1) Holleman I., L. Delobbe and A. Zgonc, 2008. Update on the european weather radar network (OPERA). Proc. Fifth European Conf. on Radar in Meteorology and Hydrology, June-July 2008, Helsinki, Finland, 5pp2) Holleman, I., H. van Gasteren and W. Bouten, 2008. Quality assessment of weather radar wind profiles during bird migration. J. Atm. Oceanic Technol., 25, 2188-2198, htpp://journals.ametsoc.org/doi/pdf/10.1175/2008JTECHA1067.13) Dokter, A.M., F. Liechti, H. Stark, L. Delobbe, P. Tabary and I. Holleman, 2010. Bird migration flight altitudes studied by a network of operational weather radars. J. R. Soc. Interface 7, doi:10-1098/rsif.2010.01164) Gasteren, H. van, I. Holleman, W. Bouten, E. van Loon, and J. Shamoun-Baranes, 2008. Extracting bird migration information from C-band weather radars. Ibis, 150, 674-686.5) Waldteufel, P. and H. Corbin, 1979. On the analysis of single doppler radar data. J. Appl. Meteor., 18, 532-542.6) Zaugg, S., G. Saporta, E. van Loon, H. Schmaljohann and F. Liechti, 2008. Automatic identification of bird targets with radar via patterns produced by wing flapping. J. Royal Soc. Interface, 5, 1041-1053.7) Schmaljohann, H., F. Liechti, E. Bchler, T. Steuri, and B. Bruderer, 2008. Quantification of bird migration by radar - a detection probability problem. Ibis, 150, 342-355.