Monitoring bird migration with a fixed-beam radar and a thermal-imaging camera

Resumen

ABSTRACT Previous studies using thermal imaging cameras (TI) have used target size as an indicator of target altitude when radar was not available, but this approach may lead to errors if birds that differ greatly in size are actually flying at the same altitude. To overcome this potential difficulty and obtain more accurate measures of the flight altitudes and numbers of individual migrants, we have developed a technique that combines a vertically pointed stationary radar beam and a vertically pointed thermal imaging camera (VERTRAD/TI). The TI provides accurate counts of the birds passing through a fixed, circular sampling area in the TI display, and the radar provides accurate data on their flight altitudes. We analyzed samples of VERTRAD/TI video data collected during nocturnal fall migration in 2000 and 2003 and during the arrival of spring trans-Gulf migration during the daytime in 2003. We used a video peak store (VPS) to make time exposures of target tracks in the video record of the TI and developed criteria to distinguish birds, foraging bats, and insects based on characteristics of the tracks in the VPS images and the altitude of the targets. The TI worked equally well during daytime and nighttime observations and best when skies were clear, because thermal radiance from cloud heat often obscured targets. The VERTRAD/TI system, though costly, is a valuable tool for measuring accurate bird migration traffic rates (the number of birds crossing 1609.34 m [1 statute mile] of front per hour) for different altitudinal strata above 25 m. The technique can be used to estimate the potential risk of migrating birds colliding with man-made obstacles of various heights (e.g., communication and broadcast towers and wind turbines)—a subject of increasing importance to conservation biologists. Estudios previos, en donde no se ha hecho uso de radar, han utilizado cámaras infrarojas de imagen termal (CIT) y el tamaño de individuos como indicador, para detereminar la altura de vuelo. Sin embargo, este método puede dar origen a errores si las aves que vuelan a una misma altura varían en tamaño. Para subsanar esta dificultad y tomar medidas más exactas de la altura de vuelo y el número de individuos en una bandada, desarrollamos una técnica que combina un radar de rayos fijos con antena parabólica (RRF) con una cámara infraroja de imagen termal (RRT/CIT). El CIT provee de un conteo preciso de las aves pasando por un área circular fija de muestreo y el radar provee el dato preciso de la altura de vuelo. Utilizando una videograbadora digital, analizamos las muestras tomadas con la combinación RRT/CIT durante la migración otoñal noctura en el 2000 y el 2003 y durante la migración primaveral diurna del 2003, a través del Golfo de México. Utilizamos la cámara de video para hacer exposiciones en lapsos de tiempo en lo tomado por el CIT y desarrollamos criterios para distinguir entre aves, murciélagos e insectos, usando la huella dejada en el video y la altura del objeto. El CIT trabajo de forma eficiente tanto de dia como de noche, pero aún mejor cuando el cielo estaba despejado (cuando esta ausente la interferencia por la irradiación de calor de parte de las nubes). El sistema RRT/CIT, aunque costoso, es una herramienta valiosa para medir con presición las rutas migratorias y el número de aves moviéndose a diferente altura. Dicho sistema es de gran utilidad para determinar el riesgo de coliciones de aves migratorias con obstáculos construidos por el hombre a diferentes alturas (ej. torres de comunicación o turbinas de viento), asuntos de gran relevancia e importancia para la conservación de aves. Direct visual studies of nocturnal bird migration have historically been conducted using binoculars or a telescope either by viewing silhouettes of birds crossing the face of the moon (moon-watching technique; Lowery 1951, Lowery and Newman 1955, Newman 1956) or viewing birds (illuminated from the ground) passing through a narrow, vertical light beam with binoculars, telescope (ceilometer technique; Gauthreaux 1969), or an image intensifier (Gauthreaux 1979). Although moon-watching is better than the ceilometer technique for discriminating birds, bats, and insects aloft, moon-watching is restricted to a few nights around the full moon without clouds, and vertical light beams may influence the flight behavior of targets on nights when the light beam is clearly visible because of high atmospheric humidity. During daylight hours, a cloudless blue sky and haze may make it nearly impossible to detect high flying migrants by direct visual observation. To overcome these shortcomings, more recent visual studies of migration have incorporated passive infrared (IR) cameras (Buurma 1988, Winkelman 1992, Bruderer and Liechti 1994) that detect the heat generated by a target. These cameras make it possible to distinguish among birds, insects, and foraging bats (Zehnder et al. 2001). However, none of these techniques provides accurate information on the distance to target or altitude of flight. Liechti et al. (1995) found that the proportion of birds detected by tracking radar and thermal imager (TI) did not change with distance between 0.5 and 3 km, and the “very rough grouping of birds into three size classes by moon-watchers and IR-operators is closely related to the distances measured by the tracking radar.” Subsequent investigators (Fortin et al. 1999, Zehnder et al. 2001) have used the relationship between silhouette size and distance to group targets into various height classes when a TI is not used with radar. However, if many different-sized birds are migrating and target size is used to estimate height, then targets of different size (e.g., a warbler and a heron) flying at the same altitude will be classified as flying at different heights. Migration traffic rate (MTR, the number of birds crossing a mile of front [1609.34 m] per hour) is a standard metric of bird migration studies. Using a “very rough grouping of birds into three size classes” to estimate altitude does not permit accurate determination of MTRs because the altitude of individual birds is unknown. For accurate measures of MTRs, accurate information about the altitudes of birds crossing the vertical field of view is needed because sample space increases with altitude. High-resolution radar with a fixed, narrow, vertically pointing beam provides accurate measures of target altitude (Blokpoel 1971). Here, we report on a technique that combines vertically pointing, fixed-conical beam radar with a vertically pointing thermal imager (VERTRAD/TI). This approach (VERTRAD/TI) simultaneously provides information on the altitude and flight direction of each individual bird (and flock), and the information can be used to compute MTRs for different altitudinal strata. Because this technique can be used during day and night, daytime and nighttime MTRs can be compared. Equipment The TI (Radiance 1, Amber Raytheon, Goleta, CA) is an IR camera system with a full screen resolution of 640 × 482 pixels. The detector frame rate is 60/s. With a 100-mm lens, the field of view is 5.57° (horizontal screen dimension) and 4.19° (vertical screen dimension). We pointed the TI camera up a vertically pointing, narrow radar beam and oriented the TI so the top of the field of view was aligned toward north (Fig. 1A). The TI detected the path of a target (X and Y dimensions) and the radar detected its altitude (Z dimension). The VERTRAD/TI configuration. (A) Thermal imager (TI; left) and vertically pointing radar system (VERTRAD; right). The TI is attached to a tripod and leveled. The parabolic antenna is mounted to the radar transmitter/receiver. (B) Schematic of configuration that joins video from the thermal imager and the video of the fixed, vertically pointing radar beam. We used marine radar (Pathfinder Model 3400, Raytheon Inc., Manchester, NH) and replaced the typical open array antenna with a parabolic antenna (61-cm diameter) that produced a beam width of 4°. The vertically pointing antenna sat on top of a transmitter-receiver unit connected to a display unit (178-mm cathode ray tube) and rectifier by cables. The transmitter frequency was 9410 ± 30 MHz (3-cm wavelength) with a peak power output of 5 kW and a minimum range of detection of 25 m. At night, our settings were a 2778 m (1.5 nautical mile) range, a pulse length of 0.08 μs/3000 Hz pulse repetition frequency (range resolution of 12 m), and range marks every 463 m. During the day, we used a 5556 m (3.0 nautical mile) range, a pulse length of 0.35 μs/1500 Hz pulse repetition frequency (range resolution of 52.5 m), and range marks every 926 m. We used different settings for night and day because trans-Gulf migrants arriving at the Gulf Coast typically fly higher during the day than at night (Gauthreaux 1972). To combine information from the TI camera and the vertical radar display into a single video display, we configured components as shown in Figure 1B. The Radiance I thermal-imaging camera and the camera recording the radar beam (Model TRV-17; Sony Digital Handycam, Sony Electronics Inc., Park Ridge, NJ) were synchronized by the splitter (Model PXD310E; PIX/2 DSP Split Screen/Fader, MicroImage Video Systems, Boyertown, PA), with the output going to a date/time generator (Model MCG-2; Burst Electronic Inc., Albuquerque, NM) then to a mini-DV recorder (Model GV-D300, Sony Video Walkman, Sony Electronics Inc.), and lastly to a WG 14 B/W video monitor (Hitachi Kokusai Electric, Inc., Woodbury, NY). Data collection We collected data at Pendleton, South Carolina (34°39′N, 82°47′W), during the evening of 26–27 September 2000, and at Wallops Island, Virginia (37°53′N, 75°30′W), about 6 km south of the Wallops Island Flight Facility (NASA) on nine dates in 2003 (16 October–3 November) when weather conditions (no rain and relatively clear skies) allowed data collection. Daylight observations were made on 1 May 2003 at McFaddin National Wildlife Refuge, Texas (29°40′N, 94°04′W). Data analysis We used a Model 443 video peak store (VPS; Colorado Video Inc., Boulder, CO) to analyze the video tapes. This device stores pixels in a new video frame if they are brighter than the corresponding pixels already stored in frame memory and, over time, a light target moving against a dark background produces a bright track (Figs. 2A–D). We analyzed these tracks using the following criteria: (1) bright targets showing a straight track with modulation along the track (from wing-beat patterns) in the VPS and echo modulation in the vertical radar beam were classified as birds (Figs. 2A and B), (2) gray and dull targets with no visible modulation along the VPS track and no modulation in the vertical radar beam or targets appearing as circles in the VPS with no echo in the radar beam were classified as insects (Fig. 2C), and (3) bright targets showing tracks with sharp turns and occasional pauses with little modulation in the VPS track and echo modulation in the vertical radar beam were classified as bats (Fig. 2D). Because the direction of target movement is not evident in a completed VPS track (time exposure), it is important to note the direction of target movement while generating a VPS track. With multiple targets, it is also difficult to associate the altitude of a target with its track in the completed VPS image (e.g., Fig. 2D), and track-altitude associations should be noted as VPS images are being generated. Examples of VPS images showing tracks and their altitudes. The display of the vertical radar beam is on the left and range marks appear as light dots along the beam. The display of the thermal imager is on the right. Because the top of thermal imager is toward the north, and the thermal imager is pointing upward, east is on the left and is on the of the that the time between each in the track is The time for each image is a few on the of the target. (A) showing in thermal radiance as a of and flying 25 m and in echo at the of the radar beam. (B) bird track moving toward the at an altitude of m the wing-beat modulation dots thermal are produced when the are and dots thermal when the are single track. that the thermal radiance of the target is the track is and is no wing-beat The echo from this is in altitude to be in the vertical radar. The echo in the radar beam 463 m does not to this track. bird tracks is to from is m and the is m a track of a foraging at m and a straight track that is to from The bird tracks and track have thermal radiance while the track For data collected on 26–27 September 2000, we analyzed samples every from to For data collected during the spring of 2003, we analyzed samples every from to We used the VPS to every track in a sample of the video record and the altitude of the echo in the radar beam the direction of and of the target using the above Data were then into a along with and we analyzed samples from to and from to on each of nine nights from to 3 2003. These were because was also being measured at the same time, and wind data were needed to bird from by of radar in We used a frame with 640 × resolution Video Inc., CA) to VPS images of each target track that through a circular sample area on the monitor an sample of 4.19° or the vertical of the images were in and with Inc., to more We used circular to analyze the straight track data and direction of The tracks of foraging bats were not straight and were not To compute MTRs the number of birds crossing 1609.34 m or statute mile of front per we measured the altitude of each bird that through the radar beam to the 12 m at night and 52.5 m during the To the number of birds crossing the 1609.34 m migration we and to each bird based on its altitude because the sample area of the beam with altitude. We then the number of birds for time to the rate of per We MTRs for birds at altitudes and then did so for birds above a altitudinal This allowed to the of migration between 25 and m with many man-made as communication towers and wind with that above m. Although the TI detected bird targets 25 these were not in our because we did not the altitude of these our of and altitudinal targets detected by and TI were fall 2000 and 2003 The VERTRAD/TI system detected birds, bats, and insects with of as of were insects and were birds on 2003, and of were insects and nine were birds on 2003. nights September 2000 and nearly were birds and we detected few foraging bats of or At night, migrating birds were m above (Fig. with a altitude of m. insects detected were m. in the radar beam and not the TI and these may have been insects high to be in the but to an echo in the radar beam. of foraging bats were m and the was at m (Fig. of birds and bats detected in the (A) of birds detected in samples on the evening of 26–27 September 2000 and in samples on nine nights from to 3 2003. (B) of birds detected in daytime samples on 1 May 2003. The data are for numbers of bird and not numbers of of bats detected in samples on the evening of 26–27 September 2000 and in samples on nine nights from to 3 2003. 1 for The during our nocturnal samples was on 26–27 September 2000 when that birds a 1609.34 m front per during the sample on and were birds in our samples flying above m. evening bird was detected in the radar beam during samples and it was m. The single bird at altitude a of birds per 1609.34 m of front per The flight of birds and insects detected in the TI and vertical radar beam on the evening of 26–27 September 2000 are in The of the for migrating birds were high and to around the during the and of sampling from to to classified as insects toward a direction of and during the and then toward during the when the The flight of foraging bats were and are not in spring 2003 Figure is a time from the VPS showing moving north, with m above the radar and the m above radar The tracks of individual birds in the are visible and birds were clearly in the Because the was the of the field of the number of birds in the was unknown. of the individual birds in the higher are to because of the altitude of the (Fig. but were also in the The tracks in the higher are more evident when the VPS is being generated from the video and when in and through image (Fig. tracks to left) are also visible in each of these was moving from to and the from to were to be detected from the echo in the vertical radar beam. Video peak store showing the arrival of of trans-Gulf migrants on the Texas on 1 May 2003 at (A) VPS The a track of an individual bird in a of birds at m The tracks of in a m are more (B) VPS showing detection of individual bird tracks in the higher tracks to left) are also visible in each We detected more birds than insects during 3 of daytime on 1 May 2003 During the of were to the peak arrival time of the the proportion of birds was were m and trans-Gulf migrants were above m. Figure the altitudinal of birds detected in the TI and in the vertical radar beam. Although many birds were m the of trans-Gulf migrants were m individual birds of or but were of or more birds size ± birds, range MTRs for arriving trans-Gulf migration on 1 May 2003 from birds per 1609.34 m of front for the to for the and for the of sampling The number of m was for the for the and for the m were and birds foraging in the The flight of the insects were toward the and as were during the observations from and at The flight of the trans-Gulf migrants were oriented between and and the circular standard was than that for insects the of arriving trans-Gulf migration the circular standard The at the m above were and, at the m were from the at The VERTRAD/TI a valuable tool for during day or The the ceilometer beam technique (Gauthreaux 1969), no for target possible of on the behavior of the birds, bats, and The system best cloudless conditions because heat from can heat from The altitude at individual birds in a were in our TI was m Liechti et al. (1995) found that nocturnal migrants 3 km were detected crossing the field of a passive IR This in detection range can be to the of the parabolic antenna and the power of the radar we used for the the TI detected targets we were high flying birds not detected by the vertical radar. We that more radar (e.g., 25 or will this The of the system not be the radar and thermal camera be at between and Although Liechti et (1995) found that the of the thermal from birds in the field of a TI were to the distance of the birds measured with tracking radar, we that the of the size of the thermal in the field of view of the TI to estimate the heights of migrants (Zehnder et al. 2001) can The of a vertically pointing radar and a vertically pointing TI errors and more accurate measures of altitude needed to of radar in a radar display is difficult and visual The TI and VPS because the VPS a time image of the thermal radiance of a target. of the track of the target can be used to the of have relatively thermal radiance and appear at relatively altitudes. a produces than the heat of a bird (Zehnder et al. 2001). targets moving of a through the TI and not detected by the radar can be as Although birds and bats high thermal we that migrating bats be with migrating birds that have a bats did not VPS tracks that modulation from because they were flying bats were on the of their high radiance and flight of flying birds clearly in the VPS tracks generated from video of the TI and of in the radar beam. the these may be in the of bird The of the TI also investigators to if an echo in the radar beam is produced by a single bird or a of The VERTRAD/TI detected the arrival of trans-Gulf migrants during daylight and, because many in during the day (Gauthreaux the TI was in individual birds in of TI cameras is their thermal cameras are more than The of a TI from to The thermal camera used in our about in note the The radar display should have the and so that in the radar beam are clearly visible when the radar screen is being video This can also be by a few or per we an echo in the vertical radar but in the possible is that targets the radar beam and the visual field of the TI a radar echo because the width of the radar beam is by the power in and a target not the to a is that the radar targets, but the TI because of their size and altitude. This is be for The vertically pointing, of the data about the altitude of targets detected in the The beam width of the was and when the beam was to the TI and in the of the visual field of the it the vertical of the viewing but not the the TI detected more targets than the radar beam at a altitude. The altitudes of birds at night were typically than during the day, as also noted by Gauthreaux detect flying birds 25 m so it is not possible to estimate the to birds in this altitudinal is possible to compute a using TI data 25 but because the sampling is more samples or samples of are of the number of birds moving the is 25 each bird is with a determination of nocturnal (and migration different altitudinal strata. data be used to estimate the potential risk of migrating birds colliding with man-made of various heights (e.g., communication and broadcast towers and wind turbines)—a subject of increasing importance to conservation biologists. We the of the of for this the National Wildlife for the spring was for the fall at the and through the at The field of and during spring was The was greatly by the of and