Airglow imager observations of atmospheric gravity waves at Alice Springs and Adelaide, Australia during the Darwin Area Wave Experiment (DAWEX)

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 109,, doi: /2004jd004697, 2004 Airglow imager observations of atmospheric gravity waves at Alice Springs and Adelaide, Australia during the Darwin Area Wave Experiment (DAWEX) J. H. Hecht, 1 S. Kovalam, 3 P. T. May, 2 G. Mills, 2 R. A. Vincent, 3 R. L. Walterscheid, 1 and J. Woithe 3 Received 25 February 2004; revised 23 August 2004; accepted 31 August 2004; published 30 October [1] The Darwin Area Wave Experiment occurred in Australia from October to December An objective was to characterize the atmospheric gravity wave field produced from intense convective activity that is routinely observed around Darwin during November and December. Two airglow imagers were sited at Adelaide and at Alice Springs, each located over 1000 km south of Darwin. Waves were observed at the mesopause region propagating predominantly toward the southeast, with some going to the northwest but with none observed going from east to west. The lack of waves propagating toward the west suggests some wind filtering mechanism below 80 km altitude. Waves observed over Alice Springs were analyzed in detail on three nights. On 16 November they were seen propagating toward the northwest. It is proposed that they were generated by dynamical events associated with a cutoff low-pressure system present over southwest Australia. On 17 and 19 November the observations are consistent with wave generation by convective activity present in the Darwin area. Thus as proposed by Walterscheid et al. [1999] and Hecht et al. [2001a], the ducting of waves from distant sources is shown to be a viable explanation for the quasimonochromatic waves frequently observed in airglow observations. Walterscheid et al. [1999] suggested that ducting of waves from the extensive region of deep cumulus convection over northern Australia explained the strong poleward directionality seen in the summer months. The present study suggests that propagation from northern Australia is selective, and ducted waves from this region may not be the primary source of waves over Adelaide when convection is occurring over central Australia. INDEX TERMS: 0310 Atmospheric Composition and Structure: Airglow and aurora; 3332 Meteorology and Atmospheric Dynamics: Mesospheric dynamics; 3334 Meteorology and Atmospheric Dynamics: Middle atmosphere dynamics (0341, 0342); KEYWORDS: airglow and aurora, waves and tides, convective processes, mesospheric dynamics Citation: Hecht, J. H., S. Kovalam, P. T. May, G. Mills, R. A. Vincent, R. L. Walterscheid, and J. Woithe (2004), Airglow imager observations of atmospheric gravity waves at Alice Springs and Adelaide, Australia during the Darwin Area Wave Experiment (DAWEX), J. Geophys. Res., 109,, doi: /2004jd Introduction [2] Since the first mesopause region airglow imaging observations 30 years ago [Peterson and Kieffaber, 1973], many of the observed structures have been attributed to the passage of atmospheric gravity waves (AGWs) that are generated in the troposphere and propagate upward to the upper mesosphere and lower thermosphere [Moreels and 1 Space Science Applications Laboratory, The Aerospace Corporation, Los Angeles, California, USA. 2 Bureau of Meteorology Research Centre, Melbourne, Victoria, Australia. 3 Department of Physics and Mathematical Physics, University of Adelaide, Adelaide, South Australia, Australia. Copyright 2004 by the American Geophysical Union /04/2004JD Herse, 1977; Hecht et al., 1995; Swenson et al., 1995; Taylor et al., 1995; Nakamura et al., 1999; Smith et al., 2000]. Passage of these waves through the airglow emission layer in the mesopause perturbs the neutral density and temperature, which subsequently perturbs the airglow emission. Images of the airglow when AGWs are present show bright and dark regions that when seen in a time sequence, appear as a series of wave-like structures moving across the sky. [3] It is not clear exactly what are the dominant sources of the AGWs observed in the airglow images. A general picture of AGWs generated in the troposphere includes primarily orographic production (flow of air over large mountains) or dynamical production associated with intense convective activity [Wang and Geller, 2003]. Many of the airglow observations to date are at sites that could be seeing AGWs generated from either source. However, many of the observed AGWs have short horizontal wavelengths, 1of15

2 suggesting they should originate only a few hundred kilometers from the observing site. Thus in many cases there is no obvious source for the observed AGWs. [4] Several recent studies have examined this last point and concluded that many of the AGWs observed by airglow imagers are ducted over distances of km [Walterscheid et al., 1999; Hecht et al., 2001a]. The Walterscheid et al. [1999] study is particularly interesting, as they reported on 9 months of airglow observations at Adelaide, Australia from April 1995 to January 1996, a period encompassing the summer monsoon period in northern Australia. AGWs were observed to come mainly from the north to northwest, a direction devoid of any large mountains and a region that is often quite dry in the spring and summer. They hypothesized that the AGWs were being generated by the intense convective activity that occurs near Darwin in November and December every year. This, however, means that these AGWs traveled nearly 3000 km, an event which is only likely if they were ducted. Walterscheid et al. [2001] showed theoretically that deep tropical convection, such as is found in the Darwin area, can indeed populate the lower thermosphere with fast short wavelength AGWs. [5] A subsequent study further investigated the ducting hypothesis using data from a site in Illinois also far distant from large mountains [Hecht et al., 2001a]. The primary directionality around (Northern Hemisphere) summer solstice was of AGWs being generated from the south and southwest, away from larger mountain ranges but consistent with generation by convective sources. These waves were hypothesized to have been ducted, although simple modeling suggested that ducting may only be effective over somewhat shorter distances of 1000 to 2000 km. If accurate this would cast some doubt that the waves seen at Adelaide routinely originated near Darwin. [6] The Darwin Area Wave Experiment (DAWEX) was conceived as an effort to study AGW production in the troposphere and stratosphere by (1) the nearly daily convective event known as Hector that occurs over the Tiwi Islands just north of Darwin during the November premonsoon buildup, and (2) the regular monsoon activity that occurs over Darwin area in December [Hamilton et al., 2004]. As part of this experiment, it was desired to also understand whether the AGWs produced by such events might be the source of the AGWs seen in airglow imagers. Airglow imagers were located close to Darwin, at Katherine, and at Wyndham [Hamilton et al., 2004] in order to observe AGWs locally generated by these events. [7] Two other observation sites were located at Adelaide and at Alice Springs at distances of approximately 2600 and 1300 km from Darwin, respectively. On the basis of the Hecht et al. [2001a] study AGWs could be ducted to the Alice Springs site. The results from those observation sites were analyzed in order to determine if any of the AGWs seen in those instruments can be attributed to any specific weather activity in general and to the convective activity which was the focus of DAWEX in particular. 2. Experimental Instrumentation and Technique [8] The unique observational data reported on in this work are the combined airglow imager data obtained at Alice Springs and Adelaide by the instruments described below. Ground-based weather data obtained by the Australian Bureau of Meteorology Research Center (BMRC) instruments and remote sensing weather data from the Japanese Meteorological Satellite are used in the interpretation of the imager data. While the sources for these latter data are more fully described by Hamilton et al. [2004], a brief summary of the instrumentation is given below. Wind and temperature data are obtained from a variety of sources including the two Medium Frequency (MF) radars deployed for DAWEX by the University of Adelaide. The details of the wind and temperature databases used in this paper are also described more fully in the overview paper of Hamilton et al. [2004] but will also be summarized below. The locations of the important sites are shown in the work of Hamilton et al. [2004] but are called out with respect to distance below and on the figures in this paper where appropriate. A knowledge of the distances involved is important for tracing the path from AGW generation in the troposphere to an observation in the lower thermosphere. [9] The data described in this paper originate from two nearly identical airglow imagers. One was deployed at Buckland Park, which is 40 km north of Adelaide (3455S, 13836E), and a second at Alice Springs (2342S, 13353E) Australia, where for example, 3455 represents 34 degrees and 55 min. Buckland Park, also the site of an MF radar, is approximately 1270 km due south from Alice Springs and 2580 km southsoutheast from Darwin (1228S 13051E). Alice Springs is approximately 1290 km southsoutheast of Darwin. Katherine (1427S, 13216E), the site of a second MF radar and an airglow imager which is discussed by Hamilton et al. [2004], is 270 km southeast of Darwin but is almost 1043 km due north of Alice Springs. The Tiwi Islands (Bathurst and Melville Islands at approximately 1127S and 13020E) are just offshore Darwin at a distance of approximately 126 km. Thus several of these locations (Tiwi Islands, Katherine, Alice Springs, and Adelaide) form a chain that runs almost north to south across the central portion of Australia. As discussed by Hamilton et al. [2004], another imager was deployed at Wyndham (1547S, 12810E), which is 460 km west of Katherine. However, in this paper only the results from the Alice Springs and Buckland Park imagers will be discussed Airglow Imagers [10] The airglow instruments at Buckland Park (BP) and at Alice Springs (AS) are modified versions of the Aerospace CCD nightglow camera which was originally described by Hecht et al. [1994]. The modified version was described by Hecht et al. [2001b]. This instrument obtains images of the OH Meinel (hereinafter OHM) and O2 Atmospheric (hereinafter O2A) band emissions. A sequence of five images is obtained, each at 1 min integration, through separate narrow passband filters. Two of the filters cover two different rotational lines of OHM (6, 2) band, two filters cover different portions of the O2A (0, 1) band, and one filter covers the background and has almost no airglow emission in its passband. The latter is used to correct the airglow images for background skylight. Thus one can obtain, besides images of the airglow, the intensity 2of15

3 operating, and owing to weather AGW image data from BP were obtained on only a few nights. However, the previous study of Walterscheid et al. [1999] provides some information on the transition from premonsoon to monsoon activity Wind and Temperature Datasets [12] Hamilton et al. [2004] describe how the datasets were combined to form average wind and temperature data sets appropriate for IOP2. Briefly, they determined from a comparison with in situ measurements during DAWEX that the United Kingdom Meteorological Office (UKMO) analysis was sufficient to provide wind and temperature data up to the stratopause. These data were extended to 100 km by using the wind climatology from the Upper Atmosphere Research Satellite (UARS) atmospheric reference project (URAP) [Swinbank and Ortland, 2003]. Atmospheric tides which are important above 70 km were incorporated by using the Global Scale Wave Model (GSWM00) [Hagen et al., 1999]. This model was tuned so as to match the measured MF radar winds, during DAWEX, at Katherine and at BP. The resultant average wind and temperature datasets for IOP2 are shown by Hamilton et al. [2004] for Figure 1. Plots of the climatology of the zonal wind during IOP2 at three times. and temperature of the OH Meinel and O2 Atmospheric emissions. Examples of this technique are found in previous studies [Hecht et al., 1997a, 1997b, 2000]. The imager now uses a 1536 by 1024 Kodak CCD chip. The pixels are binned 8 8 resulting in images that have pixels. The resultant field of view is now 46 by 69, giving a field of view of approximately km at 90 km altitude. In this paper the main data discussed will be the AGW wavelengths and horizontal velocity. In past studies it has been found that the O2A band AGWs greatly resemble those seen in the OHM images. [11] The DAWEX campaign consisted of three separate intensive observation periods (IOPs) during IOP1 occurred during the new Moon period in October and was supposed to measure pre-hector and premonsoon convective activity. IOP2 occurred during the new Moon period in November and was designed to measure the Hector convective activity, while IOP3, which occurred during the December new Moon period, was designed to measure monsoon convective activity when the convection is widespread but relatively weaker in terms of vertical motion. Unfortunately, data from the BP and AS imagers were only available during IOP2. During IOP3 the AS imager was not Figure 2. Plots of the climatology of the meridional wind during IOP2 at three times. 3of15

4 two times: at 1800 LT (830 UT), which is near sunset, and at 0600 LT (2030 UT), which is near sunrise. For the purpose of the discussion in this paper, Figures 1 and 2 show the IOP2 zonal and meridional wind components at 1800 LT, 0000 LT (1430 UT), and at 0600 LT, the middle time being close to the observations of AGW activity on several nights. Note that the layered structure of meridional wind which is antisymmetric between the Southern and Northern Hemispheres is due to tidal structure. For the purposes of the ray-tracing discussion the wind and temperature fields were linearly interpolated to provide climatologies for times between the three listed above Other Meteorological Datasets [13] Hamilton et al. [2004] discuss some of the datasets available for this study. In addition, BMRC produces daily maps of ground level rainfall over the Australian landmass obtained from rain gauges. Data used to produce those maps were overlaid on some of the AGW activity maps produced below in order to show potential regions of convective activity. Rainfall is shown when it exceeds an arbitrary level here taken as 2.5 mm over 24 hours. [14] BMRC performs an operational analysis of the winds across Australia at different pressure levels. These can be used to determine the presence of ageostrophic winds, a possible source of AGWs in the troposphere. [15] A third data source is the IR brightness temperature measured from the Japanese Geostationary Meteorological Satellite (GMS). These are calculated with a pixel size of 4 km by 4 km from the two IR channels on the satellite and three hourly imagery has been used here. The brightness temperature gives a good measure of the height of the cloud tops for optically thick clouds Model Analysis [16] Since this work is concerned with possible sources of the AGWs seen in airglow images, it is necessary to incorporate ray-tracing techniques into the analysis. Raytracing techniques are used to computationally investigate the effects of background wind and temperature variation on gravity wave propagation. These techniques as applied to AGW propagation are well summarized in the works of Jones [1969], Lighthill [2001], Marks and Eckermann [1995], and Eckermann and Marks [1996]. [17] For waves with a dispersion relationship G (w, k, x, t), where w, x, k, and t are the frequency, position vector, wave number vector, and time, respectively, then the following equations describe the ray path and the refraction of the wave vector along the ray. dx=dt dk=dt dw=dt Equations (1) (3) show how the ground-based group velocity, the wave vector, and the ground-based wave frequency are modified in the presence of winds and wind and temperature gradients. ð1þ ð2þ ð3þ [18] Following Jones [1969] and Marks and Eckermann [1995], the nonhydrostatic dispersion relation appropriate for gravity waves on a slowly varying background flow is expressed as w 2 i ¼ ðw uk vlþ 2 ¼ N 2 ðk 2 þ l 2 Þþf 2 ðm 2 þ 1=4H 2 Þ k 2 þ l 2 þ m 2 þ 1=4H 2 ; ð4þ where w is the ground-based wave frequency, w i is the intrinsic wave frequency, k, l, and m are the wave numbers in the x, y, and z directions, respectively, f is the inertial frequency, N is the Brunt Vaisala frequency, and H is the density scale height. From equation (4) an expression for m, the vertical wave number, follows as m 2 ð ¼ k2 þ l 2 Þ N 2 w 2 i w 2 1=4H 2 : ð5þ i f 2 [19] Equations (4) and (5) neglect a term w i 2 /c 2, where c is the speed of sound, which is found in the more complete dispersion relation given by Gossard and Hooke [1975]. This is done for computational purposes, but for the wave frequencies considered here this term is negligible. Furthermore, terms including f, the inertial frequency, are also negligible for the wave frequencies considered in this work. Equations (4) and (5) can then be used to derive, via equations (1) (3), the motion of the wave packet through the atmosphere. For this work ray tracing was performed in two different modes. [20] First, the ray-tracing equations appropriate to nonhydrostatic gravity waves, as given in Appendix A of Marks and Eckermann [1995], were used to trace wave packets from their source taken near 10 km to the airglow emission layer near 85 km or vice versa. In this case, equation (3) above was not used as the time variations of the atmosphere were small. For each ray there was specified an initial longitude, latitude, altitude, a wave azimuth, and a groundbased horizontal phase speed. Group velocities were evaluated and were then used to calculate the ray paths. The horizontal group velocity is in the direction of the horizontal wave vector, while the vertical wave number, m, was chosen to be appropriate to w i so that the vertical group velocity is positive ensuring upward propagation of the wave energy. In order to cover the variety of typically observed wave parameters, 1000 such rays were simulated with phase speeds of m/s, horizontal wavelengths of km, and wave periods in the range 8 30 min. These rays were launched at 0000 LT, 0600 LT, and 1800 LT from a fixed point. This was done both in the forward direction (rays launched from Katherine at 10 km altitude and followed to 85 km altitude) and backward direction (rays followed backward in time from an observation at Alice Springs at 85 km altitude to a source region at 10 km altitude). From Katherine the rays were launched at azimuths between 135 and 165 degrees in a coordinate system where 0 degrees is due north of Katherine and positive is eastward. From Alice Springs the rays were followed backward at degrees in a coordinate system where 0 degrees is due south of Alice Springs and positive is westward. [21] Second, a simplified case was also considered where once the ray reached 85 km the ray path was followed, 4of15

5 Table 1. Convective Events in the Katherine/Darwin During IOP2 Date Time Comments 16 Nov UT Major Hector (Tiwi) 17 Nov UT Major Hector (Tiwi) 17 Nov UT Squall Line (Darwin) 18 Nov UT Convection (Katherine/Darwin) 19 Nov UT Squall Line (Katherine/Wyndham) assuming the wave was trapped between layers of evanescence and only propagated horizontally. In the trapped region the wave packets are assumed to be freely propagating, bouncing back and forth between layers of evanescence. This simplified the analysis, following Hecht et al. [2001a], and allowed a consideration of whether the winds significantly rotated the wave vector or significantly moved the wave packet closer to or farther away from Alice Springs. While rigorously this ignores the effects of winds at the boundaries where the waves are evanescent, these effects should be small since the packet spends most of the time in the free propagation region. 3. Results and Discussion 3.1. Airglow Observations and Background Meteorology [22] Table 1 lists the convective events that occurred during IOP2 in the Katherine/Darwin area. These could potentially lead to AGW production in the upper troposphere which then propagate to the 85 km airglow observation region. However, these events are only part of the extensive tropospheric disturbances that occurred during IOP2 and IOP3. Figures 3 through 8 place these into perspective by showing rainfall maps over the main Australian landmass during the six nights of airglow observations discussed in this paper Event of 16 November 2001 [23] This day was characterized by a strong Hector over the Tiwi Islands from 0400 to 0900 UT. The rainfall map shows extensive rainfall in the Katherine/Darwin area. However, there was also major rainfall on the northeastern coast as well as a storm in the waters off the southwest coast. This latter storm will be discussed below. The airglow observations were only at AS on this night. AGWs came over AS from the direction of Hector to the NNW and from the SW. AGWs from the NNW only appeared for a brief period at the end of the observing period, while those from the SW were present from 1230 to 1700 UT Event of 17 November 2001 [24] Rainfall surrounded BP, but AGWs were only observed originating from the NNW during 1130 to 1300 UT. Another major Hector occurred from 0240 to 0710 UT, and squall line rainfall occurred over Darwin from 0640 to 0900 UT. At AS, AGWs were observed originating from the NNW, the direction toward the Hector event, during the period of 1100 to 1400 UT. Although rainfall occurred south of AS, AGWs were not observed from that direction. Rain was also reported along the east and southwest coasts of Australia Event of 18 November 2001 [25] Rain occurred over three distinct regions of Australia, in the Katherine/Darwin area and in the SW and SE corners of the continent. The rainfall over Katherine/Darwin Figure 3. Map of Australia showing regions of significant rainfall as colored contour regions for the 24 hour period on 16 November 2001 UT. The locations of Buckland Park, Alice Springs, Katherine, and Bathhurst Island are marked. Arrows at those sites show the direction of propagation for AGWs. The period when those AGWs were seen and the horizontal wavelength are shown. The length of the arrow is proportional to the horizontal wavelength. A color bar is shown to indicate the color assigned to the rainfall in millimeters over the 24 hours. consisted of convective activity from 0815 to 1500 UT. AGWs were only reported from AS, and they came from the N between 1040 and 1100 UT and from the NNW from 1100 to 1700 UT Event of 19 November 2001 [26] The rainfall activity was similar to the previous night. Over Katherine there was squall line activity from 0800 to 1100 UT. At AS, AGWs were observed originating from the NNW from approximately 1600 to 1900 UT. Prior to that, AGWs originated from the SE from 1345 to 1500 UT. At BP, AGWs came from the W from 1630 to Figure 4. Same as Figure 3 but for 17 November 2001 UT. 5of15

6 Figure 5. Same as Figure 3 but for 18 November 2001 UT. Figure 7. Same as Figure 3 but for 15 December 2001 UT UT and from the SW from 1800 to 1820 UT. Rain also occurred on the southwest coast Event of 15 December 2001 [27] This was during IOP3 which was generally characterized by more monsoon-like rainfall [Hamilton et al., 2004] than the isolated convective activities and squall lines seen during IOP2. On this day there was extensive rainfall over almost the entire northeast and north central (Katherine/ Darwin) portion of Australia. Yet at BP, waves were still seen to come from the NNW at 1650 to 1750 UT Event of 16 December 2001 [28] While not as extensive as on the previous day, rainfall occurred over the northeast and north central part of Australia. AGWs were seen at BP between 1600 and 1800 from the NNW and between 1130 and 1420 from the SW. [29] The wave directionality is quite striking, although consistent with previous studies [Walterscheid et al., 1999; Hecht et al., 2001a]. AGWs are not seen to come from the east, despite there being places in that direction where there was storm activity. AGWs are seen to originate predominantly from the NNW although occasionally they are seen from the SW. One explanation, following Hecht et al. [2001a], for the predominant NNW origination is that the AGWs are trapped in a lossy duct. The process of getting into the lossy duct with enough energy so that they can propagate a significant distance greatly restricts the range of available directions. This explanation, however, depends on the the presence of strong westward winds below 80 km. Such winds would also explain, through critical level absorption, the absence of westward propagating waves. From Figure 1, at sunrise and sunset such strong westward winds do seem to be present. Alexander et al. [2004], however, specifically modeled AGW production on 17 November 2001 over the Darwin area. They found that there were predominant NE and SE propagation directions which depended both on the tropospheric winds at the altitude of the wave forcing and the filtering by the tropospheric winds above that altitude. Thus part of the directionality observed may also be influenced by the winds present in the troposphere. [30] Typically, airglow imager observations occurred between 1030 and 1900 UT. However, AGWs originating Figure 6. Same as Figure 3 but for 19 November 2001 UT. Figure 8. Same as Figure 3 but for 16 December 2001 UT. 6of15

7 Figure 9. Operational surface analysis of pressure prepared by BMRC at 0000 UT on (top) 16 November and (bottom) 17 November from the NNW appeared generally toward the end of that period, consistent possibly, as is shown below, with generation in the troposphere in the late afternoon over Katherine/ Darwin. AGWs from the SW generally appeared earlier in the observation period. These presumably could originate from sources that occurred at more random periods throughout the day. [31] To investigate this further, we provide a more detailed look at three events: a source for the observation of AGWs originating to the SW of AS on 16 November 2001 and the source of AGWs originating from the NNW of AS on 17 and 19 November These events were chosen not only because they represent the two predominant directionalities observed but because there are known isolated sources which may be the cause of the observations Wave Observations 16 November 2001 UT and the Presence of Ageostrophic Winds [32] On 16 November 2001, AGWs at AS were seen from 1230 to 1700 UT propagating from the SW. This was the most extensive set of observations of AGWs coming from this direction during DAWEX and is even more interesting in that strong tropospheric weather occurred in that direction during this period. [33] Figure 9 shows the BMRC operational surface analysis on the morning before the wave event. A complex 7of15

8 cutoff low system was dominating southwestern Australia, just west of the direction from which the AGWs were observed. These types of systems often produce significant rainfall and have vigorous rainbands. The rainfall map (Figure 3) does indicate that a significant amount of rainfall occurred in the direction from which the AGWs are seen to propagate from when observed at AS, and in fact over the next 24 hours the whole system had moved eastward over this region. The location of the cutoff low at this region suggests that this system may have been the origin of the observed AGWs. The nominal observed AGW periods are 15 to 20 min, and thus neglecting background winds, the group and phase velocities are nearly equal. Therefore since the phase velocity of the AGWs are 200 km/h, it would take approximately 6 to 8 hours to travel from this system to Alice Springs, again neglecting the effects of background winds. Thus AGWs launched at 0900 UT might reach Alice Springs during the 1230 to 1700 UT period where waves were observed that night. [34] It is interesting to note that while rainfall occurred over the southwest portion of the Australian landmass on all four observation nights of IOP2, only on two nights were AGWs observed from this direction at Alice Springs and only on this night did observations occur for more than 1 1/2 hours. Figure 10 shows a satellite picture of the clouds associated with this low at 0800 UT shortly before the observed AGWs may have been launched. The coldest cloud tops in the low pressure region are only 40 to 50C, and the tropopause temperatures from the radar soundings, located at Eucla (3141S, 12853E) near the coast on the border between the states of South and Western Australian (SA/WA), are 65 to 70C. These temperatures are therefore inconsistent with vigorous convective activity. The radar data confirm this as there appears to be relatively little intense rainfall associated with this low pressure system in the period around 0900 UT. Thus something other than intense rainfall may have been the origin of the AGWs observed on 16 November [35] The cloud complex near 35S, 130E is associated with the cutoff low. There is a strong upper level jet overlying this system, as is shown in the Bureau of Meteorology analyses. The top panel of Figure 11 shows the 250 hpa isotach analysis at 0500 UTC, 16 November The main feature is a strong northwesterly jet stream over southern WA (above the surface low), with an anticyclonically curved exit near the SA/WA border. The bottom panel of Figure 11 shows the ageostrophic component of the 250 hpa flow, and these ageostrophic winds are a maximum near the jet exit with a significant component in the direction of the flow, indicating that the analyzed winds are stronger than the geostrophic wind speed indicative of supergeostrophic flow. The 40 knot ageostrophic wind near the SA/WA border is nearly half the magnitude of the total wind speed at that point, showing the flow is highly unbalanced and that geostrophic adjustment processes, largely achieved by gravity wave generation, are likely to be occurring. The anticyclonically curved flow near the jet exit and the anticyclonic shear on the poleward side of the jet axis lead to absolute vorticity values in this region which approach zero (not shown) and even become slightly positive. Such flows are inertially unstable and again are frequently associated with rapid flow adjustments. The Figure 10. Japanese Meteorological Satellite infrared color temperature image over Australia at 0800 UT on 16 November The grey scale to the right shows the temperatures associated with each color. The coldest cloud tops appear as white. situation here is very like the cases described by Uccellini and Koch [1987], who showed that a consistent feature in a number of gravity wave events over the US was the presence of a jet streak propagating toward a downstream ridge. [36] There have been a number of studies that reported on AGWs generated by ageostrophic motions [Plougonven et al., 2003; O Sullivan and Dunkerton, 1995; Fritts and Luo, 1992]. Typically, the periods for these AGWs were much longer than those reported in this study. Data, however, suggest that dynamical events associated with fronts and/or the jet stream are the source of short horizontal wavelength AGWs at least in the upper troposphere and lower stratosphere [Fritts and Nastrom, 1992]. A study of AGWs generated by large wind shears did show that the AGWs generated by such events would only reach the mesosphere if their intrinsic frequency was high [Buehler et al., 1999; Buehler and McIntyre, 1999]. [37] While the data strongly suggest that these observed AGWs are indeed associated with this cutoff low system, it is not clear as to their origin. The rainfall on 16 November 2001 did not appear to be associated with any significant convective activity, and other nights where there was significant rainfall (e.g., 18 November 2001) did not produce observed AGWs at AS for any extended temporal period. However, the observed directionality of the AGWs on 16 November 2001 is directly to the center of the rainfall maximum. While ageostrophic motions are a candidate source, the modeling to date indicates this would produce larger-scale waves than observed here. Whether the current models could resolve such small-scale waves is an open question Wave Observations at AS on 17 and 19 November 2001 UT and Their Relation to Convective Activity Near Katherine [38] Most of the AGW activity seen at AS consisted of waves propagating from the NNW over the imager site. A 8of15

9 Figure 11. Analysis at 250 hpa on 0500 UTC 16 November The wind barbs show the direction and speed of the wind. A long barb represents 10 knots, a short barb represents 5 knots, and a flag represents 50 knots (approx 5, 2.5, and 25 m/s). (top) Wind speed. Contours are at knots. (bottom) Ageostrophic winds. 9of15

10 Figure 12. Japanese Meteorological Satellite infrared color temperature ( K) images over Australia on 19 November 2001 centered over the Tiwi Islands. The locations of Katherine, Wyndham, and Alice Springs are marked with red letters K, W, and A, respectively. The scales are latitude and longitude and the color bar indicates the cloud top temperature in K. (top left) 0200 UT. (bottom left) 0500 UT. (top right) 0800 UT. (bottom right) 1100 UT. straight line extrapolation intersects the Katherine/Darwin area, suggesting that convective activity there could be responsible for the AGWs seen at AS. The night of 19 November 2001 UT is especially noteworthy, as there appears to be a relatively isolated (in time) convective event that occurs during the late afternoon and early evening near Katherine and subsequent postmidnight observations of AGW activity over AS. Figure 12 shows the development of the convective activity from satellite infrared images of cloud cover centered over the Tiwi Islands. The images are shown every 3 hours from 0200 to 1100 UT. At 0200 and 0500 UT, relatively clear skies occur in the Katherine area. However, cold cloud tops suggestive of convective activity appear at 0800 and 1100 UT in the Katherine area. Figure 13 shows a radar reflectivity map, indicating that vigorous rainfall is associated with these clouds around Katherine. [39] To investigate further whether AGWs launched from this event could be responsible for the observations of AGWs over AS beginning near 1600 UT, ray traces were performed to trace back the origin in the troposphere, assuming no ducting, of AGWs seen over AS at 85 km altitude. Figure 14 shows the background wind and temperature field that a ray launched at 85 km passes through to reach 10 km altitude, i.e., a ray traced backward in time. The other panels show the AGW intrinsic periods and the x and y horizontal distances for the wave packet. Figure 15 shows a map of the ray paths. Clearly, this is consistent with Walterscheid et al. [1999]. AGWs that are not ducted with relatively short horizontal wavelengths only travel a few hundred km from their source in the troposphere to reach 85 km altitude. It takes 2 hours to reach 85 km from a 10 km origin. None of the rays reach Katherine. As there is little weather or orographic features in this region around AS, it is difficult to understand what would launch these AGWs. [40] Figure 16 shows two maps of rays which are launched at 10 km from Katherine and followed to 85 km 10 of 15

11 Figure 13. This map shows the radar reflectivity at 0808 UT on 19 November The color scale is in decibels with red indicating the most intense reflection and indicative of intense rainfall. This map shows only a portion of the north central region of Australia imaged in Figure 12. The location of Katherine is marked with a K. Wyndham is located just off the southwest corner. The scales are the distance in kilometers from the center of the map. altitude. The top panel is for AGWs launched at 1430 UT, the same time as the backward ray trace shown in Figure 15. The results are the same; AGW wave packets travel only a few hundred km away from Katherine not reaching AS. Interestingly, although the wave vector is taken as 135 to 160 degree azimuth and thus the packets are initially launched in those directions, the background winds advect the packets so that by the time they reach 85 km they are almost due south of Katherine. The bottom panel shows another set of ray traces, this time launched at 0830 UT, close to the time that the convective activity actually occurred near Katherine. Again, most of the rays only travel a few hundred km from Katherine. However, there are some rays that actually reach AS. The reason for this can be seen in Figure 17, which shows the background winds and temperatures and intrinsic period for the AGWs. Some of the AGWs suffer critical level absorption near 80 km altitude, as there is a strong meridional wind toward the south at this time. However, calculations show that even those wave packets reach 85 km within 3 hours, so again these would not account for the AS observations where AGWs are not seen until 1600 UT. Note that the strong southward wind decreases with time so that waves launched later would not suffer this absorption. [41] However, earlier studies such as Walterscheid et al. [1999] and Hecht et al. [2001a] suggested that once AGWs reached 85 km they were trapped in a leaky duct that allowed them to travel longer distances than implied by the ray traces shown in Figure 16. In the work of Hecht et al. [2001a] the wave packets were considered to be freely propagating throughout most of the trapped region and the packets were assumed to bounce off the upper and lower boundaries where the AGWs become evanescent. So a simple ray trace was performed using equations (1) to (5) but only looking at the horizontal propagation, assuming there was no net vertical propagation for a packet that is bouncing between trapping layers above and below. This allowed a look at how the wave packet would be blown by the meridional and zonal winds and to investigate whether the horizontal wavevector would change in direction and magnitude. For this analysis the winds were taken either at 85 or 90 km altitude, representative of the altitudes of the OHM and the lower portion of the O2A emission. Horizontal wavelengths of 40 and 50 km and observed periods of 15 and 20 minutes were used in the calculation. Note that on this night, while the median AGW horizontal wavelength was approximately 50 km, a range of wavelengths was observed. From these calculations it was found that the wind and temperature gradients were small enough so that the wave vector remained essentially unchanged during the ray trace. AGWs of 40 km horizontal wavelength, a 15 min period, background winds representative of 85 km and launched at 1000 UT from 1600S latitude would reach 2400S latitude at 1600 UT. This is consistent with the observations at AS at that time. Such a packet would essentially travel due south. However, if the AGW saw background winds representative at 90 km altitude, it would be blown a few hundred km to the east. This is not surprising, as the background winds at 85 km are rather weak and those at 90 km (and above) are stronger. [42] Since the wave packet would spend part of the time in rather weak winds blowing the packet somewhat toward the west and some time above 90 km where it would see stronger winds blowing the packet a few hundred km to the east, the combined ray trace analysis suggests that the Figure 14. Four plots from a ray trace of AGWs seen at 85 km at AS and followed backward in time to their origin near 10 km. The background atmosphere is taken at 1430 UT (0000 LT). (top left) Solid line represents zonal winds in m/s, dotted line represents meridional winds in m/s, dashed-dotted line represents temperature in K. (top right) The AGW intrinsic period. (bottom left (right)) Distance traveled in the x(y) direction, where positive is toward the east(north). 11 of 15

12 Figure 15. A map of Australia showing some results for ray trace paths of AGWs seen at 85 km over Alice Springs and followed backward in time to their origin at 10 km altitude. The ray trace background conditions are taken to be those appropriate for 1430 UT (0000 LT). mesopause on both 17 and 19 November. For Hector-type storms the horizontal wavelengths are mostly below 50 km and the phase speeds are less than 40 m/s. For squall-like events the horizontal wavelengths and phase velocities can be somewhat larger. On 17 November AGWs were observed at AS from 1100 to 1300 UT, suggestive, based on the ray trace for 19 November, of AGW generation in the Darwin area from 0300 to 0500 UT. This is the period of the intense Hector event. The AS observations showed horizontal wavelengths around 35 km, somewhat shorter than those observed on 19 November, consistent with Alexander et al. [2004]. The observed phase speeds on 17 November are close to 60 m/s, somewhat above their model results Observations at BP and Monsoon Activity [46] As at AS, there were many observations of AGWs originating from the NNW at BP. The rainfall map on 15 and 16 December, for example, shows some rainfall occurring close to AS and between AS and BP, and these AGWs seen at AS in the postmidnight to sunrise time frame would have been launched a few hundred km west of Katherine in the period after 0800 UT. This certainly seems consistent with the convective activity shown in Figure 12. [43] There is some evidence for winds that can trap these waves, as seen in Figure 2. In the bottom panel, at 2030 UT (0600 LT) the winds just below 80 km and apparently above 100 km are strongly northward. At 0600 LT the peak IOP2 winds of 60 m/s occur at 100 km near 1800 S and appear to be increasing at higher altitudes. For AGWs of 15 min period traveling southward at a typical observed velocity of 40 m/s, approximately 80 m/s winds are needed to cause m 2 to go negative and trap them. Given the climatological nature of the IOP2 winds and the day to day variability of upper atmospheric tides, such velocities are certainly plausible. [44] This trap, once formed, slowly moved down during the night, and the bottom portion is present between 80 and 85 km at 1430 UT. At 0830 UT (1800 LT) the bottom is centered around 90 km. In the work of Hecht et al. [2001a] it was found that for certain propagation directions the wave packets could tunnel through the bottom of the trap, retaining enough energy to propagate considerable distance. Interestingly for IOP2, this trap is a function of latitude because the northward winds diminish toward the equator. Thus the wave packets can enter the trapped region from below, closer to the equator, where relatively less energy would be lost in penetrating the evanescent region than at more southerly latitudes. As the wave packets move south and the evanescent region becomes thicker, the trapped region would become relatively less lossy than in the Hecht et al. [2001a] simulation. [45] Finally, we note that Alexander et al. [2004] specifically modeled AGW generation for the Hector and squall activity events on 17 November. They found that for both Hector and squall line events the wave field had mainly northeast and southeast directionality in the stratosphere consistent with the directionality observed at AS near the Figure 16. Maps of Australia showing some results for ray trace paths of AGWs launched from Katherine at 10 km altitude and followed to 85 km altitude. (top) Rays launched at 1430 UT (0000 LT). (bottom) Rays launched at 0830 UT (1800 LT). 12 of 15

13 Figure 17. A forward ray trace with AGWs launched at 10 km altitude and followed forward in time until they reach 85 km. (bottom) 1430 UT (0000 LT). (top) 0830 UT (1800 LT). 13 of 15

14 tropospheric weather systems may be responsible for the BP observations on those nights. However, on 19 November there is no rainfall close to BP or even south of Katherine that could plausibly be the origin of the AGWs observed over BP. [47] These data do not resolve the question as to the origin of the AGWs seen over BP. While it cannot be ruled out that these are generated by ageostrophic winds, their horizontal wavelengths, as discussed previously, may be too short based on current models. Previously, Walterscheid et al. [1999] suggested that ducting of waves from the extensive region of deep cumulus convection over northern Australia explained the strong poleward directionality seen in the summer months at BP. The analysis of Hecht et al. [2001a] as well as the ray tracing in the previous section suggests that wave propagation over such large distances is a selective process (for example, strongly trapped waves generated at certain azimuth angles during favorable tidal phases) and sufficiently prolific wave sources closer to BP would be favored. In general, attributing the BP observations to long-range ducting of AGWs generated by convective events over northern Australia in the late afternoon in accordance with the usual diurnal cycle of convection over land seems unlikely since the wave group velocities are too small to allow them to reach BP during the evening observation period. However, in the far north, AGWs may originate over water where the diurnal cycle can be reversed (e.g., nocturnal convection over coastal waters) or fairly weak (especially closer to the equator). Future coordinated observations are probably required to determine their origin. [48] It is also worth noting that even though there was extensive monsoon activity to the east and northeast of BP on several nights, AGWs were not seen from these directions. The westward winds present at or below 80 km at sunrise and sunset may filter out the westward traveling waves. 4. Conclusions [49] The main results of this paper are the following. [50] 1. Observations of airglow emissions at two sites in central Australia, Alice Springs (AS) and Buckland Park (BP), revealed the presence of wave-like perturbations in the airglow intensity attributed to atmospheric gravity waves passing through the emission layer. The waves were found to originate mainly from the NNW of the site propagating toward the SSE. Some waves were seen to originate to the SW of the sites propagating toward the NE. No waves were seen propagating from the east to the west. Waves coming from the NNW were mainly seen close to or after local midnight, while the waves seen coming from the SW were seen earlier. [51] 2. An analysis of the waves seen over AS on 19 November 2001 was consistent with those waves being generated by convective activity in the Katherine area around sunset. However, if that was the origin of the waves seen at AS, then those waves must have been trapped or ducted. [52] 3. An analysis of the waves seen over AS on 16 November 2001 coming from the SW revealed no obvious strong convective source for those waves, although the AGWs are clearly associated with a cutoff low-pressure system present over southwest Australia. While it is possible that they were dynamically generated by ageostropic winds associated with this low, modeling studies to date suggest that such waves would have long horizontal wavelengths and periods inconsistent with our observations. [53] 4. On several nights there was evidence of tropospheric weather considerably south of Katherine and sometimes between AS and BP, which may have been the source of the wave observations at BP. However, waves seen at BP were coming from the NNW on 19 November in spite of the fact that there was no obvious origin such as convective activity south of Katherine that could account for the observations. While it cannot be ruled out that these waves were (1) generated dynamically by ageostrophic winds or (2) convectively generated by activity over northern Australia or even closer to the equator, their origin is still unknown. [54] 5. The lack of waves propagating westward suggests some wind filtering mechanism either in the troposphere or the mesosphere. [55] Acknowledgments. Thanks to Peter Strickland for the considerable help at Alice Springs. We also thank Elizabeth E. Ebert of BMRC for providing us with the rainfall maps. The Aerospace results could not have been obtained without the invaluable help given by Kirk Crawford in all aspects of this project. JHH and RLW were supported by NSF grant ATM and by The Aerospace Corporation MOIE program. References Alexander, M. J., P. T. May, and J. H. Beres (2004), Gravity waves generated by convection in the Darwin area during the Darwin Area Wave Experiment, J. Geophys. Res., 109, D20S04, doi: / 2004JD004729, in press. Buehler, O., and M. E. McIntyre (1999), On shear-generated gravity waves that reach the mesosphere. Part II: Wave propagation, J. Atmos. Sci., 56, Buehler, O., M. E. McIntyre, and J. F. Scinocca (1999), On shear-generated gravity waves that reach the mesosphere. Part I: Wave generation, J. Atmos. Sci., 56, Eckermann, S. D., and C. J. Marks (1996), An idealized ray model of gravity wave-tidal interactions, J. Geophys. Res., 101, 21,195 21,212. Fritts, D. C., and Z. Luo (1992), Gravity wave excitation by geostrophic adjustment of the jet streams, part I: Two-dimensional forcing, J. Atmos. Sci., 49, Fritts, D. C., and G. D. Nastrom (1992), Sources of mesoscale variability of gravity waves. Part II: Frontal, convective, and jet stream excitation, J. Atmos. Sci., 49, Gossard, E. E., and W. H. Hooke (1975), Waves in the Atmosphere, Atmospheric Infrasound and Gravity Waves Their Generation and Propagation, 456 pp., Elsevier Sci., New York. Hagen, M. E., M. D. Burrage, J. M. Forbes, J. Hackney, W. J. Randel, and X. Zhang (1999), GSWM-98: Results for migrating solar tides, J. Geophys. Res., 104, Hamilton, K., R. A. Vincent, and P. T. May (2004), Darwin Area Wave Experiment field campaign to study gravity wave generation and propagation, J. Geophys. Res., 109, D20S01, doi: /2003jd004393, in press. Hecht, J. H., R. L. Walterscheid, and M. N. Ross (1994), First measurements of the two-dimensional horizontal wavenumber spectrum from CCD images of the nightglow, J. Geophys. Res., 99, 11,449 11,460. Hecht, J. H., S. K. Ramsay Howat, R. L. Walterscheid, and J. R. Isler (1995), Observations of spectra of brightness fluctuations of the OH Meinel nightglow during ALOHA 93, Geophys. Res. Lett., 22, Hecht, J. H., R. L. Walterscheid, D. C. Fritts, J. R. Isler, D. C. Senft, C. S. Gardner, and S. J. Franke (1997a), Wave breaking signatures in OH airglow and sodium densities and temperatures: 1. Airglow imaging, Na lidar, and MF radar observations, J. Geophys. Res., 102, Hecht, J. H., R. L. Walterscheid, J. Woithe, L. Campbell, R. A. Vincent, and I. M. Reid (1997b), Trends of airglow imager observations near Adelaide, Australia, Geophys. Res. Lett., 24, Hecht, J. H., C. Fricke-Begemann, R. L. Walterscheid, and J. Höffner (2000), Observations of the breakdown of an atmospheric gravity wave 14 of 15