DETECTION OF BIRDS AND INSECTS USING POLARIMETRIC RADAR OBSERVATION

Transcription

1 P5.3 DETECTION OF BIRDS AND INSECTS USING POLARIMETRIC RADAR OBSERVATION Pengfei Zhang *, Alexander Ryzhkov, and Dusan Zrnic 2 CIMMS, University of Oklahoma, Norman, Oklahoma 2 NOAA/National Severe Storms Laoratory, Norman, Oklahoma. INTRODUCTION It has een known that polarimetric radar can discriminate iological and meteorological scatterers. Furthermore, Zrnic and Ryzhkov (998) have shown that the polarimetric properties of small Rayleigh scatterers like insects and large non-rayleigh scatterers like irds are significantly different. Vaughn (985) has stated that prolate spherical water drops are a etter model of irds and insects than spherical water drops (Riley, 985). Usually the ratios of widthto-length are etween :2 and :3 for irds and etween :3 and :0 for insects (Vaughn, 985). Due to the difference in shape and size, the polarimetric signatures of insects and irds are different. Birds have higher differential phase and lower differential reflectivity than insects. Thus, it is possile to develop an algorithm to distinguish etween irds and insects using polarimetric variales. These advantages of polarimetric radar can significantly improve the current radar applications. First enefit is for quantitative precipitation estimation; non-meteorological echoes caused y irds and insects can e identified and removed to prevent contamination of rainfall estimates. Second, irds can contaminate Doppler radar velocity measurements (Jungluth et al. 996; Gauthreaux et al. 998a,). In the presence of migrating irds (mostly nighttime during the migrating seasons), radar measured velocities can e very different from the air velocities (projected along the radar eams) and the differences are typically in the order of 0 m/s (Gauthreaux et al. 998; Collins 200; Bi et al. 2002). On the other hand, insects as passive tracers of air motions are usefull for wind measurements in most of cases. Third, for flight safety purpose, it might e possile to issue ird strike advisories (ird strikes to aircraft are a growing prolem in world). A classification algorithm that enales identification of different types of meteorological and non-meteorological echo has een developed at NSSL using oservations with the polarimetric prototype of the WSR-88D radar. Discrimination etween insects and irds will e part of this algorithm. In this paper, several examples of irds and insects oserved y the polarimatric KOUN radar are presented in sections 2 and 3. The discussions and conclusions are in section 4. *Corresponding author address: Pengfei Zhang, CIMMS, University of Oklahoma, Norman, OK 7309; pzhang@ou.edu 2. OBSERVATIONS OF TRANSITION BETWEEN INSECTS AND BIRDS Continuous oservations of polarimetric KOUN radar from 00:00UTC 9 May 2004 to3:00utc 0 May 2004 provide unique opportunity to study the transition etween insects and migrating irds during sunrise and sunset in spring over Oklahoma. The sunset time on 9 May 2004 and sunrise time on 0 May 2004 are 0:22UTC and :30UTC respectively. During the oservation period (from 00:00UTC to 0:5UTC on 9 May 2004) some scattered storms appeared aout 00 km away from the KOUN radar. Insects, irds, and ground clutter were present near the radar. Using the classification algorithm we have separated and removed storm echoes and ground clutter from iological scatterers. Measurements of differential reflectivity Z dr were corrected at low signalto-noise ratio (SNR). Also, the oservations with low cross-correlation coefficient (less than 0.3) were excluded from analysis. Fig. shows the reflectivity, differential reflectivity, and differential phase of iological echoes oserved with the KOUN radar at 0.5 o elevation angle at 0:9 UTC and 02:35 UTC on 0 May It is clearly evident that the clear air echoes in the upper panels (efore sunset, Fig. ) are significantly different from the echoes in the lower panels (after sunset). The extent of radar echoes increases from 50 km (first range ring in Fig.) efore sunset to almost 00 km (second range ring) after sunset. The disk-like and central symmetric shapes of reflectivity fields Z (Fig. a and d) indicate horizontal homogeneity of the scattereres. The azimuthal dependencies of differential reflectivity Z dr and differential phase are clearly pronounced in Fig., c, e, and f). In order to investigate the evolution of the polarimetric characters of these two different kinds of echoes, the average values of the radar oserved polarimetric variales are estimated. The average is made over all range gates with valid oservations at 0.5 o elevation angle. It is worth noting that the differential phase is a sum of ackscatter differential phase δ and a propagation component. Zrnic and Ryzhkov (998) found that δ dominates in the oservations of iological scatterers and the propagation component can e neglected. Henceforth, is assumed to equal δ in our oservations of irds and insects.

2 a c Z Z dr d e f Z Z dr Fig.. The upper three panels (a), (), and (c) are reflectivity Z, differential reflectivity Z dr, and differential phase f dp oserved y KOUN at 0.5 o elevation angle at 0:9UTC on 9 May The lower three panels (d), (e), and (f) are the same as (a), (), and (c), except at 02:35 UTC on 9 May The radar is at the center of each image. The white rings indicate range 50 km and 00 km from the radar. The time series of average reflectivity Z, average differential reflectivity Z dr, and average differential phase φdp over a twenty-four hour period are displayed in Fig. 2. These three variales exhiit dramatic changes around sunset and sunrise time. Between 02:30 UTC and :00 UTC, Z dr and are at almost constant levels 2.2 db and 70 o respectively, as irds were primarily responsile for the echoes. The values of Z dr and agree with the results of previous polarimetric radar oservations of irds (Zrnic and Ryzhkov, 998). In the daytime periods (00:00UTC to 0:UTC and 2:00UTC to 24:UTC), is aout 30 o that is evidently smaller than 70 o of ird echoes. But Z dr gradually increases from 0 db at aout 4:20UTC to its peak 3.5 db at 0:00UTC. For 0 cm wavelength polarimetric radar, 3 db and 30 o are typical values of Z dr and for insects (Zrnic and Ryzhkov, 998). The reason for the increase of Z dr from the morning to afternoon (4:00UTC to 24:00UTC) is unknown. This result indicates that migrating irds cause clear air echoes in the night and insects are major contriutors to the clear echoes in the late afternoon. Z, Z dr, and reach their local minimums at aout 0:40UTC right after sunset at 0:22UTC and then sharply reach their plateau value in aout an hour. At aout :00UTC (half hour efore the sunrise) Z, Z dr, and experience a significant drop from 2.5 dbz, 2.2 db, and 70 o to 2 dbz, 0.dB, and 40 o respectively in aout three hours. Oviously, the transition from irds to insects at sunrise is consideraly slower than the transition from insects to irds at sunset. In less than aout.5 hour, insects are replaced y irds in the oundary layer. Similar phenomena are oserved y the KTLX radar for a whole migrating season of the spring 2003 (Zhang et al., 2004). The fact that the morning transition starts efore the sunrise time (:30UTC) on the ground is likely due to the altitude of the migrating irds; at 2 km aloft the migrating irds see the sun efore it can e oserved on the ground and start their descent. Usually polarimetric signatures of insects and irds exhiit a well-pronounced azimuthal dependence that is determined y their fight direction. Zrnic and Ryzhkov (998) have plotted the azimuthal dependencies of Z dr and for reflections from irds and compared these to model results for olate spheroids. The azimuthal dependencies of Z dr from insects were oserved y Mueller and Larkin (985) and Achtemeier (99). Fig. 3 shows the azimuthal distriutions of Z dr, and Doppler velocity V r for insects and irds in our oservations. Z dr, and V r in Fig.3 are averaged over radial distances from 30 km to

3 a two areas from 9:00 to 2:00UTC (4:00 to 6:00CDT) (see Fig.5). The area (flock ) is ounded y 24 o and 248 o in azimuthal directions, and from 43 km to 47 km in range. The area 2 (flock 2) is ounded y 286 o and 290 o in azimuthal directions, and from 44 a c Fig. 2. The evolution of (a) Z, () Z dr, and (c) φdp oserved y KOUN radar at 0.5 o elevation angle in the period from 00:00UTC to 24:00UTC on 9 May km (60 range gates). Rawinsonde oservations on 9 May 2004 indicate that the wind is from the south that is consistent with V r otained from insects in Fig. 3a. The magnitude of V r otained from irds in Fig.3 is larger ecause the migrating irds fly with their favorite tail-wind. There are no ovious systematic azimuthal dependencies in Z dr and for insects (Fig.3a). For migrating irds (Fig.3), exhiits a symmetric pattern with respect to wind direction. It is consistent with the theoretical and oservation results of Zrnic and Ryzhkov (998). However, only to the west of the radar (i.e. azimuth 0 o 80 o, see Fig. 3), Z dr shows azimuthal dependency. 3. OBSERVATIONS OF LOCALIZED FLOCKS OF BIRDS By carefully examining the radar oservations, we found persistent flocks of nesting irds aove certain areas near rivers. We have selected two areas of localized irds echoes (see Fig. 4) and examined the evolutions of average reflectivity Z, average differential reflectivity Z dr and average differential phase corresponding to these Fig.3. Azimuthal dependencies of Z dr (lue), (red), and Doppler velocity V r (green) at 0.5 o elevation angle for reflection from (a) insects at 0:8UTC and () irds at 02:57UTC. km to 48 km in range. Fig. 5 displays the Z, Z dr and φdp for flock as function of time. The evolutions of these variales for flock 2 (not shown) are similar to flock. Z and fluctuate around 0dBZ and 70 o respectively during the two hours. These values are well aove the ackground value of Z and (4 dbz and 40 o correspondingly) during the same period of time. Z dr of aout 0 db is well elow the.5 db of ackground Z dr. In general, irds are larger than insects and have larger ackscattering cross sections than insects. High Z and are consistent with returns from irds whereas low Z dr implies that these irds might e more randomly oriented in the region compared to migrating irds. We lael the irds in the region as nesting irds ecause the corresponding Doppler velocities scattered around zero (Fig. 6) indicating that the irds do not fly in the same direction as migrating irds do. Further this also confirms that

4 2 a 2 2 c Z Z dr Fig. 4. Z (a), Z dr (), and f dp (c) oserved at y KOUN at 0.5 o elevation angle at 9:39 UTC on 9 May Radar is on the cross point of white lines. Yellow oxes mark the sections where nesting irds oserved. White range ring indicates 50 km. Fig. 5. Evolution of Z (red), Z dr (lue), and (green) for the flock (marked in Fig.4) from 9:00UTC to 2:00UTC. Fig. 6. Scattergrams of Vr in area at 20:UTC. the echoes are not ground clutters ecause in general a the Doppler velocities of ground clutters are zero. The magnitudes of Z and are consistent with Z and for nocturnal migrating irds (from 03:00 to :00UTC in Fig. 2), ut Z dr is clearly elow Z dr. We elieve the nesting irds are different species with different sizes from the migrating irds. The relatively simple model shows that prolate spheroids with different sizes can have similar values of differential phase ut different differential reflectivity (see Fig. 6 and 7 in the paper of Zrnic and Ryzhkov, 998). The oservations of these two nesting ird flocks demonstrate the capaility of polarimetric radar to provide accurate information aout the location of relatively small flock of irds. In this case the area where irds reside is aout 6 km 2. Such information might e useful to the aviation community ecause encounters with irds can damage the windshield or extinguish jet engines. 4. DISCUSSION AND CONCLUSION The results of case study presented herein are consistent with the previous studies on the polarization-dependent radar characteristics of insects and irds (Hajovsky et al., 966; Riley, 975; Zrnic and Ryzhkov, 998). The polarimetric radar is a powerful tool not only to discriminate etween iological and meteorological scatterers, ut also to discriminate etween irds and insects. Continuous oservations with the KOUN radar have shown the rapid changes of polarimetric variales like Z dr and, at sunset and sunrise time as the occupancy of the planetary oundary layer shifts etween insects and irds. It has also een demonstrated that polarimetric radar can detect localized flocks of irds in a relative small area. But there are still some unanswered questions such as gradual increase of Z dr with time during the day, asymmetry of the Z dr azimuthal profiles

5 with respect to the direction of ird migration, etc. Thus, more polarimetric radar oservations and investigations of insects and irds are required to develop a reliale automatic classification procedure to distinguish etween irds and insects. ACKNOWLEDGEMENTS This research is partially response to requirements and funding provided y the Federal Aviation Administration (FAA). The views expressed are those of the authors and do not necessarily represent the official policy or position of the FAA. REFERENCES Achtemeier,G. L. 99a Oservations of turulent oundary-layer interaction with a thunderstorm outflow - a possile wake region energy source. Boundary-Layer Meteorology, 55, Achtemeier,G. L. 99 The use of insects as tracers for "clear-air" oundary-layer studies y Doppler radar. J. Atmos. Ocean. Tech., 8, Bi, L., A. Shapiro, P. Zhang, and Q. Xu, 2002: Quality control prolems for VAD winds and NEXRAD Level-II winds in the presence of migrating irds. Preprints, 9th Conference on Weather Analysis and Forecasting and 5th Conference on Numerical Weather Prediction, 2-6 August, 2002, San Antonio, TX, Amer. Meteor. Soc., Collins, W.G., 200: The quality control of Velocity Azimuth Display (VAD) winds at the National Centers for Environmental Prediction. Preprints, th Symposium Meteorological Oservations and instrumentations, Aluquerque, Ameri. Meteor. Soc Collins, W.G., 200: The quality control of Velocity Azimuth Display (VAD) winds at the National Centers for Environmental Prediction. Preprints, th Symposium Meteorological Oservations and instrumentations, Aluquerque, Ameri. Meteor. Soc Gauthreaux, S. A., and C. G. Belser, 998a: Displays of ird movements on the WSR-88D: patterns and quantification. Wea. Forecasting, 3, Gauthreaux, S. A., D. S. Mizarhi and C. G. Belser, 998: Bird migration and ias of WSR-88D wind estimates. Wea. Forecasting, 3, Hajovsky, R. G., A. P. Deam and A. H. Lagrone, 966: Radar reflections from insects in the lower atmosphere. IEEE Trans. Antannas Propag., 4, Jungluth, K., J. Belles, and M. Schumacher, 995: Velocity contamination of WSR-88D and wind profiler data due to migrating irds. Preprints, 27th Conference on radar meteorology, 9-3 August 995, Vail, Colorado, Amer. Meteor. Soc., Mueller, E. A., and R. P. Larkin, 985: Insects oserved using dual-polarization radar. J. Atmos. Oceanic Technol., 2, Riley, J. R., 985: Radar cross sections of insects. Proc. IEEE, 73, Vaughn, C. R., 985: Birds and insects as radar targets: A review. Proc. IEEE, 73, Zhang, P., S. Liu, and Q. Xu, 2004: Identifying Doppler Velocity Contamination Caused y Migrating Birds, Part I: Feature Extraction and Quantification. (Sumitted to J. Atmos. Oceanic Technol., June 2004). Zrnic, D. S., and A. V. Ryzhkov, 998: Oservations of insects and irds with a polarimetric radar. IEEE Transactions on Geoscience and Remote Sensing, 36,