538 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 14

Airborne as Potential Sources of Acoustic Re¯ectivity

J. R. RILEY AND A. S. EDWARDS Natural Resources Institute Radar Unit, Great Malvern, United Kingdom 21 July 1995 and 1 April 1996

ABSTRACT Laboratory measurements of the acoustic backscattering cross sections of a variety of insects at 3.4 kHz are reported. The authors argue from these data, and from what is known (from radar studies) about con- centrations in the lower atmosphere, that acoustic returns from insects may seriously contaminate sodar data. They suggest that this possibility should be carefully considered when interpreting backscatter coef®cient data acquired in acoustic sounding experiments.

1. Introduction that the diurnal and seasonal variation of the phenom- enon indicated that the observed dot echoes were more a. Radar ``angels'' probably produced by insects, and the acoustic echoes In the early days of radar, operators occasionally re- certainly appeared to be similar to those previously at- ported anomalous echoes from clear air for which there tributed to ¯ying grasshoppers by Cronenwett et al. was no obvious cause (see references in Crawford (1972). More recently, De et al. (1994) have also sug- 1949). These mysterious echoes that sometimes ap- gested that the dot echoes detected by their sodar were peared as point targets and sometimes as extended struc- of atmospheric origin. The purpose of this paper is to tures and in expanding rings, came to be known as radar demonstrate that airborne insects may on occasion be ``angels'' and formed the subject of much ingenious sources of both acoustic dot echoes and of distributed speculation (Gordon 1949). In particular, it was hy- acoustic re¯ectivity, and to suggest that this possibility pothesized that angels indicated the presence of micro- should be carefully considered when sounder echoes are refractive structures in the atmosphere, associated with being interpreted. The paper follows an earlier note on turbulent layers and with convective bubbles (Plank the same topic (Riley 1994). 1958). It eventually became clear, however, that as Crawford (1949) had originally suggested, most angels seen on ordinary radars could be con®dently attributed 2. Insects as acoustic re¯ectors to either airborne insects (Campistron 1975) or to birds (Pollen 1972; Eastwood 1967). Specialist, highly sen- a. Bat studies sitive radars do indeed produce spectacular displays of atmospheric refractive-index gradients (e.g., Browning It is well known that many species of insectivorous and Watkins 1970; Jordan and McLaughlin 1992), but bat locate their prey by acoustic sounding, and that air- these do not appear as point targets. borne insects must therefore be effective re¯ectors of sound, at least at the frequencies used by bats (typically 30±80 kHz). Laboratory backscattering studies designed b. Sodar angels to investigate the sonar mechanisms of hunting bats During the sixth ISARS meeting in Athens in 1992, have focused primarily on the temporal variation of Pahwa et al. (1992, unpublished manuscript) described 80-kHz echoes from ¯ying insects (Roeder 1963; Kober some sodar observations of dot echo structures and pos- and Schnitzler 1990), and there appear to be no mea- tulated that the dots might indicate the presence of mi- surements reported in the literature of insect acoustic crostructures in the lower atmosphere. It seemed to us backscattering cross sections at sodar frequencies (1.5± 4 kHz). We have therefore made some measurements in this frequency range of the acoustic cross sections of several species of and of a locust. Cross sections Corresponding author address: Dr. J. R. Riley, NRI Radar Unit, were measured with the insects presenting their under- North Site, Leigh Sinton Rd., Great Malvern, Worcestershire WR14 ILL, United Kingdom. side aspect, that is, the attitude from which they would E-mail: [email protected] be viewed when over¯ying a sodar.

᭧1997 American Meteorological Society

Unauthenticated | Downloaded 09/25/21 06:23 PM UTC JUNE 1997 NOTES AND CORRESPONDENCE 539

®ed and bandpass ®ltered, and the returns from cali- bration targets and from insects were displayed on an oscilloscope and recorded. The insects were either alive during the experiments and made comatose by cooling, or were freshly killed (by freezing). Signal amplitude was calibrated in terms of cross section by using the expression 25 ␴ϭ K42␲r, b ΂΃9 (where K ϭ 2␲/␭, r is the sphere radius, and ␭ the acoustic wavelength) to estimate the acoustic backscat-

tering cross section ␴b of the calibration spheres (Riley 1994). This expression is accurate only for spheres that are small enough to lie in the Rayleigh scattering region, that is, where 2␲r/␭ K 1. For a frequency of 3.4 kHz FIG. 1. Experimental layout. The transmitting antenna was a 0.6- this corresponds to diameters much less than 3.6 cm. m-diameter paraboloid, ®tted with a 3.4-kHz 0.5-W transducer at the focus and with a 1.2-m-long cylindrical shroud. The receiver was an We used calibration spheres with diameters in the range unshrouded 0.6-m paraboloid with a miniature microphone (sensitiv- 1.91±1.26 cm, which barely meet this criterion, but are ityÐ64 dB at 1 kHz, 0 dB ϭ 1V␮bϪ1) at the focus. Both antennas close enough to it for the equation to produce reasonable pointed horizontally over the smooth, concrete ¯oor to the measure- estimates of cross section. ment point 6 m in front of the them, where calibration spheres and insect targets were placed. c. Results The results of the measurements are summarized in b. Experiments Table 1, together with estimates of backscattering cross sections at the commonly used sodar frequency of 1.5 The experiments were carried out in a large (9.5 m kHz, derived by applying the scaling factor (1.5/3.4)4 ϫ 11.3 m ϫ 4.6 m) anechoic enclosure with a ¯at and to the cross-sectional data measured at 3.4 kHz. very smooth concrete ¯oor. The equipment was ar- Radar backscattering cross sections of insects have ranged so that the ¯oor acted as the mirror in an acoustic been found to be rather similar to those of water spheres analog of the Lloyd's mirror arrangement (Longhurst of equivalent mass (Riley 1985), and so it was of interest 1963). This arrangement had two advantages. First, the to see if there were an analogous relationship for acous- insects and calibration spheres could be simply placed tic cross sections. Our estimates of insect scattering on the ¯oor in front of the acoustic source and receiver cross sections at 1.5 kHz have therefore been plotted as (Fig. 1), obviating the need for elaborate anechoic sup- a function of their mass in Fig. 2, together with the port structures. All that was needed to hold the insects acoustic backscattering cross section calculated for wa- in place was a single, vertical pin, held upright by a ter spheres, also plotted as a function of mass (Riley small piece of Plasticine ®xed to the ¯oor. Second, un- 1994). It can be seen that the insect underside aspect like the electromagnetic case, acoustic power density is cross sections do indeed increase with mass in a similar a maximum at the mirror surface (approximately four way to those of water spheres, although the insect values times the free space value), so targets on it are optimally are some 10 dB higher. It is to be emphasized that insect placed to act as re¯ectors. Calculations of power density acoustic cross sections, like their radar analogs, will be as a function of height above the ¯oor showed that the aspect sensitive, and that the ``water sphere plus 10 dB'' ®rst interference minimum was at 30 cm, and that over relationship is presented here only as a convenient the body length of the largest insects (5 cm) illuminating means of estimating the effect that airborne insects may power varied by less than 10%. Because the acoustic have on vertically looking acoustic sounders. We found that, unlike the case with radar (Riley 1973), the wings echoes from the insects were expected to be very weak, of the insects made a major contribution to their cross we used a close working range (6 m), and this dictated section, a result similar to that found in the 80-kHz the use of separate (bistatic) transducers for transmission backscatter experiments (Roeder 1963; Kober and and reception, so our results do not correspond exactly Schnitzler 1990). Re¯ections from moth wings were to monostatic backscattering cross sections. However, higher than those from locust wings. the angle subtended between the transducers at the target position was small (6Њ), so the difference is unlikely to be signi®cant. Because the transmitter transducer avail- 3. Sodar returns from insects ¯ying in the lower able for the experiments operated most ef®ciently at 3.4 atmosphere kHz, our measurements were made at this frequency (␭ a. Dot echoes from individual insects ഠ 10 cm). The pulse length was set at 3 ms. Cronenwett et al. (1972) deduced that the dot echoes The signal from the receiver microphone was ampli- detected at an altitude of 100±250 m above their 1.5-

Unauthenticated | Downloaded 09/25/21 06:23 PM UTC 540 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 14

TABLE 1. Measurements of the acoustic backscattering cross sections of various insects at 3.4 kHz. The measurements were made using the apparatus shown in Fig. 1 and by comparing the amplitude of the signals returned from the insects with those returned from a series of solid spheres. We also list in the table estimates of insect backscattering cross sections at the frequency of 1.5 kHz commonly used in sodars. These were made by scaling the measurements at 3.4 kHz by the ratio (1.5/3.4)4, on the assumption that the insects would act like Rayleigh scatterers.

2 2 ␴b (cm ) ␴b (cm ) scaled Symbol in Species Mass (g) measured at 3.4 kHz to 1.5 kHz Fig. 2 Locusta migratoria female 1.34 7.69 ϫ 10Ϫ1 3.2 ϫ 10Ϫ2 ✪ Locusta migratoria male 1.00 9.61 ϫ 10Ϫ1 3.6 ϫ 10Ϫ2 ✪ Locusta migratoria female 1.83 9.7 ϫ 10Ϫ1 3.7 ϫ 10Ϫ2 ✪ tiliae (lime hawk) 0.83 9.7 ϫ 10Ϫ1 3.7 ϫ 10Ϫ2 ⅜ (lime hawk) 0.32 1.6 ϫ 10Ϫ1 6.1 ϫ 10Ϫ3 ⅜ Mimas tiliae (lime hawk) 0.94 5.2 ϫ 10Ϫ1 2.0 ϫ 10Ϫ3 ⅜ Mimas tiliae (lime hawk) 0.35 1.2 ϫ 10Ϫ1 4.5 ϫ 10Ϫ3 ⅜ Sphinx ligustri (privet hawk) 2.60 7.2 2.7 ϫ 10Ϫ1 ᭪ Noctua pronuba (large yellow underwing) 0.24 2.1 ϫ 10Ϫ1 7.8 ϫ 10Ϫ3 ⅷ kHz sodar were caused by discrete targets, small com- weighs about 2 g (N. Jago 1993, personal communi- pared to the sounder wavelength (23 cm), and moving cation), and it would appear from our data on morpho- horizontally at speeds of 0.8±7.5 m sϪ1. Flying long- logically similar locusts (Fig. 2) that a grasshopper of horned grasshoppers (Neoconocephalus robustus) were this mass would have an underside scattering cross sec- numerous in the area at the time, and it was concluded tion at least four times higher. However, in Cronenwett that these insects were the source of the echoes. It was et al.'s observations, the grasshoppers were actively ¯y- noted that the amplitude of the echoes corresponded to ing, so their wings would have been broadside to the an estimated acoustic backscattering cross section of sodar beam for only part of the time, and the sodar about 10Ϫ2 cm2. Neoconocephalus robustus typically would be expected to register a substantially lower av- erage cross section than we found in the laboratory. In any event, given the uncertainty of the position of the point targets in Cronenwett et al.'s sodar beam, and the corresponding uncertainty in the cross-sectional esti- mates, their identi®cation of their 10Ϫ2-cm2 targets as grasshoppers seems entirely plausible. It seems clear from the above that the larger grass- hopper and moth species (and, of course, birds and bats) will be individually detectable on many acoustic sound- ers, especially those with large, high gain antennas. These targets should therefore be at least considered as candidates to explain the occurrence of the dot echoes sometimes observed on acoustic sounders.

b. Distributed echo from insect concentrations associated with atmospheric features The dot echoes produced by large insects, once iden- ti®ed, are unlikely to prove a problem in acoustic sound- ing measurements. A potentially much more serious ef- fect will occur if smaller insects accumulate in meteo- rological features being studied by sodar and contam- inate the atmospheric acoustic returns with their distributed collective echo. This seems to be a very real possibility, because it has been found in entomological FIG. 2. Estimated acoustic backscatter at 1.5 kHz from insects, plotted as a function of their mass: underside viewing, wings spread. radar studies (Reynolds 1988; Riley 1989) that noctur- ✪ Locusta migratoria (migratory locust), ⅙ Mimas tiliae (lime hawk nally ¯ying insects frequently accumulate in dense lay- moth), ᭪ Sphinx ligustri (privet hawk moth), and ⅷ Noctua pronuba ers (sometimes multiple layers) (Drake and Farrow (large yellow underwing). Error bars are omitted for clarity; the re- 1988), often near temperature inversion levels (Cam- peatability between measurements was on the order of Ϯ50%. The solid line shows the acoustic backscatter from water spheres, cal- pistron 1975; Schaefer 1976; Riley and Reynolds 1979; 3 2 2 4 Drake 1984), and that spectacular concentration also culated from the expression ␴b ϭ (25␲r m )/␳ ␭ (Riley 1994), where ␳, the density of water, is taken to be 1 g cmϪ3. occurs in zones of atmospheric convergence associated

Unauthenticated | Downloaded 09/25/21 06:23 PM UTC JUNE 1997 NOTES AND CORRESPONDENCE 541

TABLE 2. Examples of airborne insect concentrations associated with meteorological phenomena, together with estimates of the acoustic backscatter coef®cient of the concentrations. The insect scattering cross sections have been calculated on the assumption that they are ten times greater than the cross section of a water drop of the same mass. Estimated Alti- cross section Density Backscatter Meteorological tude Estimated (m2) (No. per coef®cient Reference conditions Species (m) mass (g) (␭ ϭ 23 cm) 106 m3) (mϪ1) Drake 1982 Sea breeze Grasshoppers (Acridi- 350 1±2 6 ϫ 10Ϫ6 7 4.2 ϫ 10Ϫ11 dae) Drake and Farrow Stable large-scale Australian plague lo- 160 0.5±1 1.6 ϫ 10Ϫ6 40 6.4 ϫ 10Ϫ11 1983 air¯ow cust Greenbank et al. Storm front con- Spruce budworm moth 150 0.1±0.2 6 ϫ 10Ϫ7 1100 6.6 ϫ 10Ϫ10 1980 vergence zone Farrow 1981 Stable large-scale Australian plague lo- 150 0.5±1 1.6 ϫ 10Ϫ6 300 4.8 ϫ 10Ϫ10 air¯ow cust Pedgley et al. 1982 Storm out¯ow African armyworm 250 0.1±0.3 1.2 ϫ 10Ϫ7 108 1.3 ϫ 10Ϫ11 moth Riley and Reynolds Wind shift line Grasshoppers 300 1±2 6 ϫ 10Ϫ7 1000±1500 7.5 ϫ 10Ϫ10 1983 Schaefer 1976 Intertropical front Grasshoppers 450 1±2 6 ϫ 10Ϫ7 1000 6.0 ϫ 10Ϫ10

with storm front out¯ows (Schaefer 1976; Riley and Crawford, A. B., 1949: Radar re¯ections in the lower atmosphere. Reynolds 1990), with sea breezes (Greenbank et al. Proc. Inst. Radio Eng., 37, 404±405. Cronenwett, W. T., G. B. Walker, and R. L. Inman, 1972: Acoustic 1980; Drake 1984), and sometimes with solitary waves sounding of meteorological phenomena in the planetary bound- in the nocturnal boundary layer (Drake 1985). Linear ary layer. J. Appl. Meteor., 11, 1351±1358. concentrations of airborne insects have also been ob- De, A. K., S. Tripathy, and J. Das, 1994: On ®ne structure of dot served in the daytime, apparently associated with the echoes as observed by acoustic sounder. Int. J. Remote Sens., cellular pattern of convection currents (Schaefer 1976). 15, 2157±2165. Drake, V. A., 1982: Insects in the sea-breeze front at Canberra: A Table 2 gives some examples of insect concentrations radar study. Weather, 37, 134±143. found to be associated with atmospheric features, to- , 1984: The vertical distribution of macro-insects migrating in gether with estimates of the resulting acoustic back- the nocturnal boundary layer: A radar study. Bound.-Layer Me- scatter coef®cient, based on our measurements of in- teor., 28, 353±374. , 1985: Solitary wave disturbances of the nocturnal boundary dividual cross sections. layer revealed by radar observations of migrating insects. The values for backscatter coef®cient from the insect Bound.-Layer Meteor., 31, 269±286. concentrations are comparable to those to be expected , and R. A. Farrow, 1983: The nocturnal migration of the Aus- from both the convective daytime and the stable night- tralian plague locust, Chortoicetes terminifera (Walker) (Or- time boundary layer (10Ϫ11±10Ϫ10 mϪ1) (A. Grant 1995, thoptera: Acrididae): Quantitative radar observations of a series of northward ¯ights. Bull. Entomol. Res., 73, 567±585. personal communication), so it appears that insect con- , and , 1988: The in¯uence of atmospheric structure and centrations must be considered as potentially misleading motions on insect migration. Annu. Rev. Entomol., 33, 183±210. sources of acoustic backscatter in sodar studies. It is to Eastwood, E., 1967: Radar Ornithology. Methuen, 278 pp. be noted too, that larger species have signi®cant air- Farrow, R. A., 1981: Aerial dispersal of Scelio fulgidus [Hym.: Sce- speeds (3±5 m sϪ1) and so may introduce errors in acous- lionidae], parasite of eggs of locusts and grasshoppers [Ort.: Acrididae]. Entomophaga, 26, 349±355. tic Doppler wind sounders. In cases where it is suspected Gordon, W. E., 1949: A theory on radar re¯ections from the lower that insect contributions to atmospheric backscatter co- atmosphere. Proc. Inst. Radio Eng., 37, 41±43. ef®cient might critically affect the outcome of a sodar Greenbank, D. O., G. W. Schaefer, and R. C. Rainey, 1980: Spruce experiment, a new type of inexpensive, rotating polar- budworm (: Tortricidae) moth ¯ight and dispersal: ization, vertical-looking radar (Smith et al. 1993) may New understanding from canopy observations, radar and aircraft. Mem. Entomol. Soc. Canada, 110, 49 pp. be used to unambiguously detect the presence, size, 2 Jordan, J. R., and S. McLaughlin, 1992: Ultrahigh resolutionCn pro- speed, altitudinal distribution, and aerial density of ov- ®les derived from FM-CW radar. Proc. Soc. Photo-Opt. Instrum. er¯ying insects. Eng., 1688, 287±293. Kober, R., and H. U. Schnitzler, 1990: Information in sonar echoes of ¯uttering insects available for echolocating bats. J. Acoust. REFERENCES Soc. Amer., 87, 882±896. Longhurst, R. S., 1963: Geometrical and Physical Optics. Longmans, Browning, K. A., and C. D. Watkins, 1970: Observations of clear air 551 pp. turbulence by high power radar. Nature, 227, 260±263. Pedgley, D. E., D. R. Reynolds, J. R. Riley, and M. R. Tucker, 1982: Campistron, B., 1975: Characteristic distributions of angel echoes in Flying insects reveal small-scale wind systems. Weather, 37, the lower atmosphere and their meteorological implications. 295±306. Bound.-Layer Meteor., 9, 411±426. Plank, V. G., 1958: Atmospheric angels. Electronics, 31, 140±144.

Unauthenticated | Downloaded 09/25/21 06:23 PM UTC 542 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 14

Pollen G. E., 1972: Distribution of radar angels. Inst. Electr. Electron. ,and , 1983: A long-range migration of grasshoppers ob- Eng. Trans. Aerosp. Electron. Syst., AES-8, 721±727. served in the Sahelian zone of Mali by two radars. J. Anim. Reynolds, D. R., 1988: Twenty years of radar entomology. Antenna, Ecol., 52, 167±183. 12, 44±49. ,and , 1990: Nocturnal grasshopper migration in West Af- Riley, J. R., 1973: Angular and temporal variations in the radar cross- rica: Transport and concentration by the wind, and the impli- sections of insects. Proc. Inst. Electr. Eng., 120, 1229±1232. cations for air-to-air control. Philos. Trans. Roy. Soc. London, , 1985: Radar cross-section of insects. Proc. Inst. Electr. Elec- Ser. B, 328, 655±672. tron. Eng., 73, 228±232. Roeder, K. D., 1963: Echoes of ultrasonic pulses from ¯ying . , 1989: Remote sensing in entomology. Annu. Rev. Entomol., Biol. Bull., 124, 200±210. 34, 247±271. Schaefer, G. W., 1976: Radar observations of insect ¯ight. Symp. Roy. , 1994: Acoustic sounding, atmospheric structure and insects. Entomol. Soc., 7, 157±197. Int. J. Remote Sens., 15, 293±297. Smith, A. D., J. R. Riley, and R. D. Gregory, 1993: A method for , and D. R. Reynolds, 1979: Radar-based studies of the migratory routine monitoring of the aerial migration of insects by using a ¯ight of grasshoppers in the middle Niger area of Mali. Proc. vertical-looking radar. Philos. Trans. Roy. Soc. London, Ser. B, Roy. Soc. London, Ser. B, 204, 67±82. 340, 393±404.

Unauthenticated | Downloaded 09/25/21 06:23 PM UTC