A REVIEW OF IMAGING LOW-LATITUDE IONOSPHERIC IRREGULARITY PROCESSES

Transcription

1 A REVIEW OF IMAGING LOW-LATITUDE IONOSPHERIC IRREGULARITY PROCESSES J. J. Makela E.O. Hulburt Center for Space Research Code 7607 Naval Research Laboratory Washington, District of Columbia, USA A review of the past 30 years of imaging low-latitude irregularity processes is presented. The physical framework of the generalized Rayleigh-Taylor instability, which is understood to be responsible for the generation of these irregularities, is briefly summarized. The signature in optical data of this process is the development of a region of low emission that typically shows eastwest dimensions of 50 to several hundred kilometers. In the meridional direction they can at times extend over two-thousand kilometers poleward from the genesis region at the magnetic equator. A fairly consistent view of these depletions has arisen based on observations around the world. However, several properties show considerable seasonal, longitudinal, and solar cycle dependence. Although considerable work and progress has been made in understanding this phenomenon, several questions relating to the day-to-day variability of its occurrence, the latitudinal dependence of its drift velocity, and conjugate nature are still unanswered. The answers should be attainable in the near future as more coordinated observations are carried out. 1. A BRIEF HISTORY Optical observations of the Earth s ionosphere have a long history of productivity. The aurora occurring at high latitudes has been observed by humans for centuries, as they are bright enough to be viewed by the naked eye. A multitude of instruments and techniques have been developed over the years to more quantitatively study the various optical emissions such as the aurora that occur naturally in the nighttime atmosphere. The two optical instruments that have historically provided extensive datasets for our understanding of the ionosphere are the photometer and Fabry-Perot Interferometer. These instruments provide excellent, quantitative information along a single look direction, essentially providing a point measurement. Two-dimensional information was occasionally obtained by scanning these instruments, although this technique tends to produce some ambiguities regarding spatial versus temporal variations. Due to the large intensity of the emissions at high latitudes, two-dimensional images of the aurora could be obtained using conventional photographic techniques. However, at lower latitudes, the naturally occurring emissions are so dim, that images were difficult to obtain. Beginning in the 1970s, new highly sensitive film was coming onto the market allowing for researchers to take images of the dim emissions that occur outside of the polar regions. The initial studies were of mesospheric dynamics [Peterson and Kieffaber, 1973], specifically atmospheric gravity waves and they still suffered from the problems faced by other one-dimensional optical techniques. The necessary integration time for these early observations was on the order of 10 minutes, giving results that were still somewhat ambiguous in terms of spatial versus temporal variations. Better temporal resolution was gained by employing lowlight television cameras [Mende and Eather, 1976; Crawford et al., 1978], but research was still relegated to the brighter emissions, such as those produced in the aurora and the broadband OH emission in the mesosphere. As instruments steadily improved, researchers were able to apply two-dimensional imaging techniques to a variety of new problems, including that of low-latitude irregularity processes [e.g., Weber et al., 1978; Mendillo and Baumgardner, 1982; Mendillo and Tyler, 1983]. These studies used the OI nm emission that arises due to the dissociative recombination of O 2 + [e.g., Link and Cogger, 1988]. Other studies included the dimmer OI emission at nm due to radiative recombination of O + [e.g., Tinsley et al., 1973] to provide additional information on these phenomena [e.g., Moore and Weber, 1981; Weber et al., 1982]. More recent studies employ the nm emission [e.g.,

2 2 MAKELA: LOW-LATITUDE IMAGING PASSIVE OPTICS SELF-ASSESSMENT Fagundes et al., 1995; Mendillo et al., 1997; Takahashi et al., 2001; Kelley et al., 2002] which is typically dominated by mesospheric sources. Occasionally, however, the F-region portion can become dominant. In these cases, the nm emission is preferable to the nm emission (which originates from the same dissociative recombination of O 2 + ) due to its prompt emission (compared with the 110-second lifetime of the nm emission) resulting in a sharper image. The studies performed in the late 1970s and early 1980s were our first view of the twodimensional structure of depleted regions in electron density in the equatorial and low-latitude ionosphere. These were thought to be associated with the equatorial irregularity process caused by the Rayleigh-Taylor instability (RTI), commonly referred to as equatorial spread-f. A better term is perhaps Convective Ionospheric Storms (CIS) which is a physics-based, rather than an instrument-based, description of this phenomenon. These storms are among the most explosive events that naturally occur in our ionosphere and are an important component of ongoing Space Weather research. Comparisons to other datasets, namely radar [Mendillo et al., 1992] and in-situ satellite data [Weber et al., 1982], confirmed that the depleted regions seen in images of the low-latitude nightglow were indeed associated with this process and opened a new era in the study of the CIS. The next step in the evolution of imaging science came with the development of the chargecoupled device (CCD), which replaced traditional film for recording photons. The advantages of the CCD over traditional film-based techniques became quickly evident. The CCD offers an increase in sensitivity allowing for observations of dimmer emissions using shorter integration times. This results in a higher temporal resolution increasing the utility of imaging to study the dynamics of the ionosphere. CCDs are also linear devices, allowing for a much easier quantitative analysis of the collected data in comparison to 35-mm film. With the advent of CCDs and their integration into allsky imaging systems [Baumgardner et al., 1993], the community was poised to make a new push towards studying the dynamics of the ionosphere. The first CEDAR campaign to make use of this new technology for a long-term study was the Multi-Instrumented Studies of Equatorial Thermospheric Aeronomy (MISETA), which commenced in October of 1993 [Mendillo et al., 1997]. An imaging system was installed at Arequipa, Perú, and obtained measurements at various wavelengths, including nm, nm and nm. The results of this campaign showed the improved image quality and temporal resolution obtained by using a CCD-based imaging system in terms of studying the depleted regions associated with CIS. For example, using consecutive images, the zonal drift velocity of the depleted regions were measured and shown to be consistent with the background zonal drift velocity. Up until the mid-1990s, very few imaging investigations had been performed at mid-latitudes, with the conventional wisdom stating that this region was relatively quiescent and contained very little structure that imaging would shed light on. This is not to say that optical studies at mid-latitudes had not been performed, or had not been successful. Arecibo, for example, has a healthy and productive optical program using photometers, Fabry-Perot interferometers, and Lidar. In addition, imaging experiments had been carried out to study mesospheric waves [e.g., Hecht et al., 1994; Taylor and Garcia, 1995]. However, no two-dimensional imaging experiments had been carried out to study the ionosphere, as the traditional one-dimensional techniques were believed to be sufficient to capture the information desired on the dynamics in this region. An allsky instrument was installed at the Arecibo Observatory for the 10-day run at Arecibo in 1993 to uncover the possible scientific yield of fielding an imaging system at midlatitudes. The results of this experiment showed that there were in fact optical signatures of disturbances observable at this latitude range. Additional studies were performed in the subsequent years, and a large dataset of these medium-scale traveling ionospheric disturbances (MSTID), as well as a geomagnetically enhanced version of the MSTID, has been collected [e.g., Garcia et al., 2000a, b; Kelley et al., 2000; Otsuka et al., 2004]. In conjunction with other observing techniques, we have come to a better understanding as to the physical process behind their occurrence [Kelley and Makela, 2001]. The imaging of the Earth s ionosphere has a rich tradition for studying the dynamics at many different latitudes. We will limit the discussion in this paper to a review of the numerous studies of the low-latitude irregularity process, commonly referred to as equatorial spread-f, but perhaps more properly termed Convective Ionospheric Storms. The terminology to describe this phenomenon is varied and at times confusing. Terms such as biteouts and depletions are common in the satellite literature, irregularities and plumes in the radar literature, and depletions and bubbles in the imaging literature. We will attempt to remain consistent, referring to the optical signature of the CIS as a depletion, due to the low light level measured with the imaging instruments. However, it should be noted that all of the above terms, when used in the realm of describing features in the post-

3 PASSIVE OPTICS SELF-ASSESSMENT MAKELA: LOW-LATITUDE IMAGING 3 Figure 1. Diagram of the Rayleigh-Taylor instability in the equatorial plane. After Kelley [1989]. sunset equatorial and low-latitude ionosphere generally refer to the same irregularity process which is described below. 2. IMAGING OF EQUATORIAL INSTABILITY PROCESSES As seen above, the study of equatorial instability processes using two-dimensional imaging has been carried out over the past 30 years. Since the first of these observations by Weber et al. [1978], much has been learned about the dominant instability process in the post-sunset equatorial ionosphere, the Rayleigh-Taylor instability (RTI). There is currently much interest in this phenomenon as it produces irregularities that have an adverse affect on trans-ionospheric radio signals [e.g., Basu et al., 2002, and references therein], on which we are becoming evermore dependent. As will be shown below, imaging can provide a wealth of information, some of which is entirely unique and some that is complimentary or corollary to other techniques. While an in-depth review of the RTI is not appropriate here, we begin with a very brief primer on the RTI while pointing the interested reader to other sources for a more in depth discussion [e.g., Kelley, 1989]. This is followed by a description of recent results produced by CEDAR-related studies that have made a contribution to our current understanding of this important instability process The Rayleigh-Taylor Instability and Equatorial Depletions The Rayleigh-Taylor instability (RTI) process was first suggested as the physical mechanism for creating Convective Ionospheric Storms (CIS), also known as equatorial spread-f, by Dungey [1956]. The conditions necessary for the RTI are typically set up in the post-sunset time frame when the bottomside of the F-layer has recombined and the F- layer has been raised considerably due to the prereversal enhancement, creating a very sharp vertical gradient in electron density. This situation is analogous to one that is perhaps a bit more intuitive, that of a heavy fluid being supported by a lighter fluid, which is dynamically unstable. In the case of the post-sunset ionosphere, the heavy fluid is the F-layer, which does not recombine away as quickly as the bottomside, while the supportive light fluid is the Earth s magnetic field. The situation is unstable to vertical perturbations in the ionosphere, which can release the gravitational energy stored in the height of the ionosphere. This is illustrated in the simplified sketch shown in Figure 1 where a sinusoidal perturbation has been imposed on a simplified model of the post-sunset ionosphere consisting of a step function in electron density. A gravitational current flows in the x direction and causes a perturbation electric field. This perturbation in turn causes an upward drift in the regions of depleted density and a downward drift in the region of enhanced density, causing the depletion to grow vertically. This is indeed a simplification of reality, where a zonally eastward electric field or a neutral wind can also destabilize the layer. What produces the initial seed perturbation is still the topic of much debate, and is especially relevant to being able to predict the occurrence of CIS. We point the reader to the discussion in Kelley [1989] for a more thorough treatment of the physics. Typically, the generation of irregularities by the RTI is confined to the equatorial regions. This is because off of the magnetic equator, B is not perpendicular to gravity, and the plasma can fall along the field lines rather than remaining in the dynamically unstable situation. Indeed, under extreme conditions the RTI can create irregularities off the equator, even as far poleward as Arecibo [Kelley and Nicolls, submitted], but this is the very rare exception and in general it can be assumed that the RTI is confined to within a few degrees of the magnetic equator. However, this does not mean that the effects of these storms are constrained to the equator. In fact, the effects can be seen as far poleward as Hawaii at 20 N magnetic longitude [e.g., Rohrbaugh et al., 1989; Kelley et al., 2002; Makela and Kelley, 2003] and even at Arecibo, Puerto Rico at 30 N magnetic longitude [C. Martinis, personal communication]. These far reaching effects are due to the large conductivity and high diffusion velocity along the magnetic field lines. The perturbation electric fields generated at the equator are easily transmitted along the magnetic field lines where they affect the local ionosphere. Thus, as the irregularities grow in altitude at the equator, they effect field lines that have footprints in the ionosphere at increasingly high latitudes. The development of a CIS in the equatorial ionosphere is not a purely random process. The physical conditions must be conducive to the develop-

4 4 MAKELA: LOW-LATITUDE IMAGING PASSIVE OPTICS SELF-ASSESSMENT ment of the instability. A large body of work has been constructed to study this matter using a variety of techniques and a general understanding of the occurrence of these storms in different longitude sectors and seasons has been developed. Perhaps the most notable of these are the frameworks presented by Tsunoda [1985] and Maruyama and Matuura [1984]. In the Tsunoda [1985] framework, the pattern of irregularity occurrence is related to the alignment of the terminator and geomagnetic flux tubes. These so called sunset nodes are the times when the conjugate E-regions enter into darkness at the same time creating a large gradient in the E-region Pedersen conductivity. This large longitudinal conductivity gradient can enhance the eastward electric field [Heelis et al., 1974] which, as we have seen above, can be destabilizing to the RTI. Thus, Tsunoda [1985] concluded that near simultaneous sunset in the conjugate E-regions was a necessary condition for the development of equatorial irregularities associated with the RTI. The Maruyama and Matuura [1984] study looked at a different parameter for control over the occurrence of equatorial irregularities - that of the transequatorial neutral wind. Basing their study on two years of topside sounding measurements, they found that asymmetry in the distribution of electron density around the magnetic equator acted as an inhibitor to irregularity development. Conversely, irregularities were prevalent during times of symmetrical distribution leading them to conclude that a necessary condition for the growth of the RTI was an absence of a transequatorial thermospheric wind in the magnetic meridian. They argued that the effect of a transequatorial neutral wind was to lower the local ionosphere off of the equator in one hemisphere and raise it in the other. This has a drastic effect on the fieldline integrated Pedersen conductivity that would suppress the development of the RTI. Thus, the seasonality of the equatorial irregularities could be explained using knowledge of the seasonal properties of the transequatorial wind and the declination of the magnetic field. The overall control of whether the conditions are favorable for the RTI is most likely a combination of these two mechanisms. Mendillo et al. [1992] suggested that the requirements for growth are: 1. The alignment of the geomagnetic flux tubes and the terminator as suggested by Tsunoda [1985]. 2. The absence of a strong transequatorial thermospheric wind that would suppress the growth rate as suggested by Maruyama and Matuura [1984]. 3. The rapid post-sunset rise of the F-layer. 4. The availability of a seed perturbation. More recent results, however, have called in to question the importance of the meridional neutral wind as a suppression mechanism [Mendillo et al., 2001; Otsuka et al., 2002]. Although this general understanding is sufficient to explain the overall longitudinal/seasonal characteristics of equatorial irregularities, they do not address what has become a very important topic: day-to-day variability. Within the spread-f season for a given longitude, irregularities are not observed every night. Similarly, outside of the longitude s season, irregularities sometimes develop. A proper understanding of this variability is essential to the community s ability to predict the occurrence of irregularities and will be based upon a better understanding of the physical processes at play and a more robust observational database Recent Imaging Observations of Equatorial Depletions The observations reviewed here stem from the efforts of a variety of research groups operating at many locations around the globe. A summary of these locations is shown in Figure 2. As is seen in this Figure, most of the sites used thus far have been constrained to being at or equatorward of the equatorial anomaly region (the dashed lines at ±15 magnetic latitude) and a majority have been in the southern hemisphere of South America. These studies employ allsky cameras observing towards the zenith and record data as the depletions pass overhead. The individual circles show the 75 zenith angle projected to a typical F-layer height of 350 km. An alternate observing technique was proposed by Tinsley [1982] who proposed observing at low elevation angles from a site poleward of the anomaly. By doing so, the look directions from the observing site are tangent to the magnetic field lines at the F-layer height, as shown in Figure 3. This allows for the best possible spatial resolution because the observation primarily comes from a single magnetic flux tube, rather than crossing multiple tubes. Studies that have employed this observing technique have produced fantastically detailed images of the longitudinal and vertical structures of the depletions. However, this observing scheme has only been employed on the Haleakala Volcano on Maui, Hawaii. This is not the only location where such observations are possible, and hopefully future experiments of this type will be carried out elsewhere. For the most part, studies have been performed at individual sites, and thus the results are specific to the longitude sectors from which the data were collected. Some results are generalizable (such as the spatial characteristics of the depletions) and

5 PASSIVE OPTICS SELF-ASSESSMENT MAKELA: LOW-LATITUDE IMAGING 5 Location of Equatorial Imagers Figure 2. Locations of imagers used to study equatorial plasma depletions. The circles show the fieldsof-view of each imager at an zenith angle of 75 projected to 350 km. The arch shape south of Hawaii is for the field-aligned viewing geometry. Note that not all imaging sites are currently operating. seem to show similar characteristics in all of the studies. Others, however clearly demonstrate longitudinal differences (such as the drift velocities and occurrence statistics). This presents the need for a study to take the collected data and put it into a global framework. When the results appear generalizable, we refrain from calling out the specific location from where the data was obtained. Otherwise, specific sites are mentioned. It is important to recognize that the CEDAR community has been involved in these studies from the beginning. The collaboration between CEDAR scientists and those from the international community is clearly evident from perusing the author lists or the acknowledgment sections in the published literature. A large majority have some mention of CEDAR, whether it be through the direct participation of a CEDAR researcher or through the broader collaboration with the CEDAR community in regards to instrument development and procurement. It is safe to say that we would not be at the level of maturity we are in regards to understanding these equatorial depletions without the participation of the CEDAR community Depletion Characteristics Over the past 30 years of imaging experiments observing equatorial depletions, we have developed a general understanding of depletion characteristics. Observing from an equatorial station they appear as striations of dark bands in the nightglow, oftentimes giving the impression of looking at an orange peel or beachball as seen in Figure 4a. At an observing site under the equatorial anomaly, some structuring on the walls begins to become evident as demonstrated in Figure 4b. Seen from a site poleward of the anomaly, the vertical structuring (in the magnetic equator s frame) can be imaged in incredible detail as seen in Figures 4c, especially using a narrowfield camera viewing along the magnetic field lines as in Figure 4d. The depletions have east-west dimensions that can range from 50 to several hundred kilometers [e.g., Mendillo and Tyler, 1983; Rohrbaugh et al., 1989; Kelley et al., 2002; Otsuka et al., 2002; Mukherjee, 2003], are typically depleted by at least 30-50% [e.g., Sinha and Raizada, 2000; Mukherjee, 2003], and can cover up to 40% of the sky at any given time [Mendillo and Baumgardner, 1982]. By using the nm emission or the method of observing along the field lines as described in Tinsley Apex Height [km] Narrowfield Vieweing Geometry Haleakala, HI Figure 3. Field-aligned viewing geometry for the imager on Haleakala, Hawaii. The field-of-view (blue lines) maps along the magnetic field lines to apex altitudes between about 350 and 1000 km for an emission that occurs at 300 km (red line).

6 6 MAKELA: LOW-LATITUDE IMAGING PASSIVE OPTICS SELF-ASSESSMENT a) b) Chistmas Island, Sept. 28, 1995 Brazapolis, Brazil, Nov. 8, 2002 c) Haleakala, Oct. 11, 2002 d) Haleakala, Oct. 11, 2002 Figure 4. Examples of equatorial depletions observed in the nm emission obtained from different latitudes. In each image, north is to the top and east is to the right. (a) Christmas Island (2.00 N, W; magnetic latitude 3.1 N) near the magnetic equator. Image courtesy of M. J. Taylor at Utah State University. (b) Brazapolis, Brazil (22.53 S, W; magnetic latitude 16.0 S) under the equatorial anomaly. Image courtesy of Y. Sahai at Universidade do Vale do Paraíba. (c) Haleakala, Hawaii (20.71 N, W; magnetic latitude 21.3 N) poleward of the anomaly. (d) Haleakala, Hawaii obtained simultaneous to (c) using a narrowfield camera observing at a low-elevation along the magnetic field lines. [1982], a higher spatial resolution can be obtained, showing structuring in the depletions below 10 km [Tinsley et al., 1997; Abalde et al., 2001; Makela et al., in press]. Pimenta et al. [2003a] has shown that the east-west dimensions of the depleted regions have different characteristics during the typical spread-f season and outside of it. During December, January, and February (the season for the Brazilian sector) they quoted the dimensions as ranging from km, while during October, November, and March (during the transition into and out of the spread-f season) the dimensions were typically km. They suggested that this difference is due to the effectiveness of the polar-

7 PASSIVE OPTICS SELF-ASSESSMENT MAKELA: LOW-LATITUDE IMAGING 7 ization electric fields in controlling the generation of irregularities, which have a seasonal and solar cycle dependence. In the north-south direction, the dimension of the depletions varies significantly. The poleward distance that they reach is directly related to the apex altitude to which the CIS grows at the equator, and can have a strong dependence on the solar cycle [Sahai et al., 1994, 2000]. However, in general, the apex height of the bubbles varies between 500 km and 1500 km [e.g., Mendillo and Baumgardner, 1982; Rohrbaugh et al., 1989; Mendillo et al., 1997; Sahai et al., 2000], although in some extreme cases, depletions associated with CIS have been observed to reach poleward of Hawaii [Kelley et al., 2002] and even to Arecibo [C. Martinis, personal communication]. In order to confirm that the depletions show similar characteristics in conjugate hemispheres, Otsuka et al. [2002] presented simultaneous images from Darwin, Australia (12.4 S, E; magnetic latitude 24 S) and Sata, Japan (31.0 N, E; magnetic latitude 24 N). Figure 5 shows an example from their study in which images obtained within two minutes of one another are presented from the two sites. The image from Darwin has been mapped along the magnetic field lines to the northern hemisphere for direct comparisons to the Sata image. It is clear that, although the background intensities show different characteristics, with the Darwin image exhibiting higher intensities in the nm emission, the depleted regions themselves show similar characteristics down to scale sizes of around 40 km. This confirms the notion that the polarization electric fields generated by the RTI at the equator at scale sizes above 40 km are efficiently mapped along the magnetic field lines. Using satellite data, Saito et al. [1995] Alt. (km) 36 N Lat. (deg.) Sata Long. (deg.) E Intensity (R) 36 N Lat. (deg.) Darwin Long. (deg.) Intensity (R) Figure 5. Near-simultaneous images taken of the nm emission at (left) Sata, Japan (right) and Darwin, Australia. The Darwin image has been mapped along the magnetic field to the northern hemisphere. The emission layer altitude is assumed to be 250 km. The scale to the left shows the equivalent apex altitude of the images. After Otsuka et al. [2002]. E showed that the electric field with a scale size of 30 km was only attenuated by 10% as it was transmitted along the magnetic field lines from one hemisphere to the other. Electric fields with larger scale sizes were even more efficiently transmitted Westward Tilts The depletions are generally aligned from north to south along the magnetic meridian. However, this alignment is not perfect. A westward tilt with increasing poleward latitude is generally observed [e.g., Mendillo and Tyler, 1983; Rohrbaugh et al., 1989; Abalde et al., 2001; Mukherjee, 2003]. This westward tilt is also readily evident in radar measurements [e.g., Woodman and LaHoz, 1976], appearing as a C shape in the radar backscatter. This is thought to be due to the fieldline integrated conductivity falling off both below and above the F peak [Zalesak et al., 1982]. In fact, when viewed from space, Kelley et al. [2003] pointed out that a backwards C shape is visible in the optical emission as well. However, this is different than the C shape seen in the radar data which comes from both above and below the F peak. The backwards C shape seen in the optical data is from the mapping of the depletion to both hemispheres as discussed above, and is only from the portion of the depletion above the F peak. The westward tilt observed in images of equatorial depletions develops over time and is thought to be due to the integrated effect of the shear in zonal plasma drift setup by the ratio of equatorial F-region and off-equatorial E-region Pedersen conductivities [Zalesak et al., 1982]. Mendillo and Tyler [1983] presented a simple equation based on optical observations for the tilt over time. Although the tilt is generally westward with increasing latitude, this is not always the case. Sinha and Raizada [2000] present data from the Indian sector that show an eastward tilt. They interpreted this result in terms of a variation of plasma drifts with altitude, relating it to the mechanism proposed by Anderson and Mendillo [1983] in which a latitudinal decrease in zonal wind drives the typical decrease in eastward drift velocity with altitude, rather than a decrease in the Pedersen conductivity. Thus, an anomalous wind field could be the cause of the eastward tilt observed. Using the field-aligned observing technique and combining the images together to form a composite image, Makela and Kelley [2003] present a striking example of this tilt over time, as reproduced in Figure 6. Here, an entire night of data has been combined into a single image. The depletions observed from Hawaii early in the evening are to the left of the image and appear as vertically elongated dark regions (when mapped to apex altitudes). As the night progresses, the depleted regions are observed

8 8 MAKELA: LOW-LATITUDE IMAGING PASSIVE OPTICS SELF-ASSESSMENT Apex Altitude (km) Observation LT (hrs) 19:38 20:45 21:52 22:59 00:06 01:13 02:20 03:27 04:34 19:37 19:55 20:07 20:20 20:31 20:42 20:53 21:04 21:13 Formation LT (hrs) Formation geomagnetic longitude (E) Figure 6. Example of a composite image formed from images of the nm emission. The top axis shows the time at which each portion of the image was taken. The bottom axis shows an estimate of the formation time and location of each portion of the image, assuming each depletion formed 82 minutes after local sunset. After Makela and Kelley [2003]. from Hawaii to have increasingly westward tilts. An estimate of the velocity shear using a method similar to that of Mendillo and Tyler [1983] was performed from the tilt angles and ranged from m/s/km to m/s/km for the individual depletions. An independent estimate of the velocity shear was obtained by estimating the zonal drift velocity at three different altitudes in temporally adjacent images resulting in a slightly higher shear of 0.1 m/s/km. Makela and Kelley [2003] suggested that this could be due to the time dependence of the velocity shear since the estimate from the composite image is from the effect of the time-integrated shear, while the technique using the individual images gives a better estimate of the instantaneous value Drift Velocity of Depletions Imaging provides a fairly dense set of data, both temporally and spatially, from which excellent estimates of the drift velocity of the depletions can be made. However, a few caveats are necessary when interpreting the drifts obtained from images in a physical sense. First, to relate the drift velocity of the depletions to the zonal plasma drift velocity, it is assumed that the depletions and ambient plasma are drifting at the same speed. This is not always the case, especially during the explosive development phase of the depletions, when this assumption is invalid [Martinis et al., 2003]. Second, as pointed out by Pimenta et al. [2003b], the assumed height of the emission layer can have an effect on the estimated drift velocities. They showed that if the drifts are calculated assuming the OI nm emission occurs at 300 km, rather than 250 km as is typically done, the velocities increase by 20%. Thus, discrepancies in velocities derived from images and other techniques may be attributed to an incorrect height being assumed for the emission layer. Finally, it is at times somewhat unclear exactly to what the velocity estimated refers. As Mendillo and Baumgardner [1982] state, this difficulty arises from trying to identify small displacements in nebulous features. Granted, in this case they were referring to estimating meridional velocities from the images, which can be much more difficult to obtain. Yet, the same can be said for the zonal direction considering the bifurcations, secondary instabilities, and blurring of the emission that are all prevalent in the images. Pimenta et al. [2003b] suggest minimizing problems by using the drift velocity obtained from the western wall of the depletion, which as we will see below, tends to have a sharper and more stable gradient than the eastern wall. Taking all of this into consideration, an amazingly consistent view of the zonal drift velocity of the depleted regions arises. The general picture is that of a gradual decrease in the zonal drift velocity from between 100 and 200 m/s at around 2200 LT to below 50 m/s after local midnight [Mendillo and Baumgardner, 1982; Mendillo et al., 1997; Taylor et al., 1997; Sinha and Raizada, 2000; Pimenta et al., 2003b]. These results are in good agreement with climatological estimates of the drifts based on measurements made by the Jicamarca incoherent scatter radar [Fejer et al., 1991]. However, other researchers have seen a slight increase in the drift velocity near local midnight [Tinsley et al., 1997; Santana et al., 2001; de Paula et al., 2002; Makela and Kelley, 2003; Mukherjee, 2003]. Still others have at times seen zonal drift velocities as fast as 400 m/s [Fagundes et al., 1997]. The generally good agreement between the drifts estimated from imaging studies and the climatology derived from radar techniques suggests that accurate velocities can be derived from imaging studies. The differences seen in some of the studies simply highlight the fact that significant variability in drifts

9 PASSIVE OPTICS SELF-ASSESSMENT MAKELA: LOW-LATITUDE IMAGING 9 Figure 7. (top) Smoothed average zonal plasma drifts derived from imaging data obtained at Arequipa, Perú (solid line) and Tucumán, Argentina (dashed line). 26 nights of data are used for Arequipa while 17 are used for Tucumán. Data from the September/October and March/April periods are used. (bottom) Common night measurements from the two sites on October 26, 1997 showing the same basic trend portrayed in the averaged data. After Martinis et al. [2003]. occur day to day, and between different seasons, latitudes, and longitudes. A few studies have begun to address the latitudinal variation of the zonal drift velocity derived from images by utilizing multiple imagers in the South American sector. Pimenta et al. [2003a] used allsky images from São João do Cariri, Brazil (7.4 S, 36.5 W; magnetic latitude 1.19 S) and Cachoeira Paulista, Brazil (22.7 S, 45.0 W; magnetic latitude 15.8 S) that cover the region from the about 5 S to 25 S (geographic) to study the latitudinal dependence of the zonal drift velocity. They found that the drifts with respect to latitude and local time very closely followed the zonal winds of the HWM-90 climatological model [Hedin et al., 1991] suggesting that the F-region zonal winds are the source of the electric fields which in turn drive the ion drifts at tropical latitudes. Two peaks in the zonal drift velocity were seen in the data. The first was located at about 19 S between 2000 and 2200 LT, while the second occurred a bit later at the magnetic equator between 2100 and 2200 LT. They also found a valley in the zonal drift velocity at approximately 10 S that they attributed to the increase in ion drag in the equatorial anomaly region that reduces the zonal neutral wind. A similar study was performed by Martinis et al. [2003] using allsky data from Tucumán, Argentina (26.9 S, 65 W; magnetic latitude 13.2 S) and Arequipa, Perú (16.5 S, 71.5 W; magnetic latitude 2.7 S). Their results show that in the postsunset period, the eastward plasma drifts at the magnetic equator (Arequipa) are smaller than near the anomaly region (Tucumán). However, this relationship reverses near 2300 LT, as shown in Figure 7. Using the coupled ionosphere-electric field model of Eccles [1998a, b] this pattern was interpreted to show that the latitudinal dependence of drifts was due the combined influence of E- and F-region Pedersen conductivities and neutral wind shears in both altitude and latitude. The smaller drifts close to the magnetic equator seem to be due to the fact that the E-region dynamo is more important in this region early in the evening. This trend is reminiscent of the evening vortex that has been observed in radar measurements of the nearsunset drift velocities which could play a role in the seeding of these equatorial irregularities [Kudeki and Bhattacharyya, 1999] Bifurcations and Secondary Instabilities From the optical images, it is also evident that as the depleted region develops in altitude it often bifurcates, creating two connected depleted regions [e.g., Mendillo and Tyler, 1983; Rohrbaugh et al., 1989; Kelley et al., 2002]. The occurrence of bifurcations seems to have a dependence on the solar cycle. Pimenta et al. [2001] showed that bifurcations occur at a much higher frequency during periods of high solar activity. They attributed this to the fact that during these times, the width of the depletions were larger. As shown in the numerical simulations of Huang and Kelley [1996], bifurcations occur when the east-west dimension of the depletion is large and do not when it is small. Mendillo and Tyler [1983] suggested two mechanisms behind this formation. The first was that the bifurcation developed symmetrically about the vertical direction in the magnetic equatorial plane. As we have seen above, the depletions tend to tilt westward as they age. This would create the appearance of one vertical bifurcation and one tilted to the west as is typically observed. Their second hypothesis was that as a single depletion tilted further westward with time, the gradient on the eastern wall would become unstable and a secondary depletion would form. This second hypothesis is

10 10 MAKELA: LOW-LATITUDE IMAGING PASSIVE OPTICS SELF-ASSESSMENT belied by the fact, as seen below, that it is the western wall that tends to be unstable to secondary structuring, not the eastern. Recently, Makela and Kelley [submitted] have presented experimental results obtained of the growth phase of a plasma depletion and compared them to the nonlinear model of Zalesak et al. [1982]. These results support the notion of the bifurcation occurring symmetrically about the vertical axis. Another aspect of the geometry of equatorial depletions that has garnered much attention is the difference in gradients on the east versus west wall of the depleted region. Multiple researchers observed that the gradient on the west wall is typically much steeper than that of the east wall [e.g., Mendillo and Baumgardner, 1982; Pimenta et al., 2003b]. Pimenta et al. [2003b] noted that the gradient of the western wall did not change with time, while that of the eastern wall became shallower. They suggested this observation indicated that the eastern wall undergoes a stronger interaction with the typically eastward neutral wind, increasing the momentum transfer from the wind to the plasma and widening the bubble on the eastern wall. It is interesting to point out that Tsunoda [1983] reported on structuring of the western wall of equatorial plumes, as seen in ALTAIR radar data. He suggested that this structuring was driven by an eastward neutral wind that was enhanced by the reduced ion-neutral drag creating a wind-driven gradient drift instability. It was noted that these secondary plumes developed in an analogous fashion to the main plume. These characteristics are confirmed in the imaging data collected from Haleakala, Hawaii while looking along the magnetic field lines [Rohrbaugh et al., 1989; Kelley et al., 2002]. In fact, using the data that is currently being collected from this site, the development of the secondary depletions can be studied in great detail, both spatially and temporally. Although these secondary plumes tend to develop on the western wall, as suggested by Tsunoda [1983], this is not always the case. During magnetic disturbances, when the background wind field can be drastically modified, secondary plumes are seen to develop on the eastern wall. This is illustrated in Figure 8, which shows a series of images obtained from Haleakala, Hawaii in February of The development of the secondary instabilities on the eastern wall occurs at a time when the eastward drift typical of these depletions has reversed, suggesting that a disturbance dynamo has been set up [Blanc and Richmond, 1980]. The bottom portion of Figure 8 shows that this is indeed a plausible explanation by showing the vertical drifts predicted by the storm-time electric field model of Fejer and Scherliess [1997]. Note that the upward drift perturbations are associated with westward drift per- Fejer-Scheleiss Disturbance Dynamo Drifts Feb 04, 2003 [UT] Vz [m/s] 0700 UT 0805 UT 0958 UT 1103 UT 0901 UT 1200 UT Figure 8. Images of the nm emission capture on February 4, 2003 from Haleakala, Hawaii. A depletion drifts eastward as shown in the images in the top row. Between 0900 UT and 1000 the depletions reverse directions, drifting westward. Also at this time, secondary structuring on the eastern wall (denoted by the arrows) develop. This is the same time at which the disturbance dynamo electric field predicted by the Fejer and Scherliess [1997] model predict an increase in the vertical drifts, as shown in the bottom panel. turbations [Abdu et al., 1998]. During the time of the primary plume s reversal in direction and the development of the secondary plumes on the eastern wall, the disturbance dynamo is predicted to be creating drifts in a direction opposite to those typical. Thus, the images can be used to study the effects of storm-time processes on the low-latitude ionosphere Seasonal and Solar Cycle Effects The imaging of low-latitude depletions has now been carried out over a long enough period that a large database has been constructed from which can be deduced their seasonal and solar cycle trends. The largest database exists in South America where the imager at Cachoeira Paulista, Brazil operated from March 1987 until October 1991 and was reinstated in September 1994 continuing to make observations into the new millennium [Sahai et al., 2000; Sobral et al., 2002]. The seasonal trends of the occurrence of equatorial depletions for this site follows the framework suggested by Tsunoda [1985] with the majority of depletions being observed during October through March. Corre-

11 PASSIVE OPTICS SELF-ASSESSMENT MAKELA: LOW-LATITUDE IMAGING 11 spondingly, a minimum is seen between May and August. The more interesting results from the Sahai et al. [2000] study are those pertaining to the solar cycle effects. Having taken over 11,000 images during the decade of data collection, they show that more depletions occur at high solar activity (55% occurrence rate) than at low solar activity (33%). As mentioned above, the depletions also reached higher heights (extended further poleward) during high solar activity (66% reaching 1500 km apex altitude) than during low solar activity (34%). Furthermore, Pimenta et al. [2001] showed that during high solar activity, more bifurcations were observed (53% of depletions) than during low solar activity (37%). As pointed out above, they interpreted this to indicate that during high solar activity, the depletions tend to have a larger east-west dimension in accordance with the results of the modelling study by Huang and Kelley [1996]. A large database is also being collected in the Pacific sector, beginning in Makela et al. [in press] used the first year-and-a-half of this data to analyze the seasonal trends of the plasma depletions. Similar to the studies in the Brazilian sector, they found that the Tsunoda [1985] framework explained the general trends of occurrence, with a maxima seen in April and September and an overall broad maximum from June through October. They also showed that one advantage of imaging studies of these depletions is the ability to make a determination of actively developing depletions versus fossilized depletions. Thus, they concluded that many of the depletions seen from June through August developed to the west of Hawaii and drifted into the field-of-view of the cameras at least two hours after local sunset fully developed and fossilized. This was evidence that the growth region for these depletions was approximately 900 km to the west of Hawaii, suggesting a much finer longitudinal structuring in the spread- F seasons than previously expected. Their results also support those given above in that both the overall number of depletions and the number that reached high altitudes decreased from 2002 to 2003, tracking the decrease in solar activity. Even out of the typical spread-f season, equatorial plasma depletions can be produced, again highlighting the day-to-day variability of this phenomenon. These anomalous nights are generally associated with an increase in magnetic activity [Bhattacharyya et al., 2002; Hysell and Burcham, 2002]. This is clearly seen in imaging data when depletions are observed out-of-season during periods of magnetic activity [Bittencourt et al., 1997; Sahai et al., 1998; Makela et al., in press]. In general, these out-of-season depletions are thought to be initiated by the penetration of high-latitude electric fields into the equatorial ionosphere. Magnetic activity has other effects on the evolution of depletions, which are just now being investigated in greater depth. Such effects include the reversal of the drift velocity [e.g., Taylor et al., 1997; Abdu et al., 2003], as well as the post-midnight generation of depletions Comparisons to Other Techniques As with any observing technique, the imaging of equatorial depletions benefits from coordination with other instruments. As far back as 1979, airglow observations of these irregularities were performed in conjunction with satellite measurements of the in-situ ion density from the AE-E satellite [Weber et al., 1982]. These combined observations confirmed that the depletions seen in the airglow images were indeed caused by the equatorial RTI Latitude [N] September 22, :40 UT Longitude [E] Figure 9. A comparison of imaging data collected from Haleakala, Hawaii and in-situ density measurements taken at 600 km by ROCSAT-1 (red line). The composite image in Figure 6 has been used and mapped along the magnetic field lines to 600 km altitude. The blue arch is the field-of-view of the imager used to take the images. An excellent agreement is seen between the location of the depletions estimated by the composite imaging technique and the ROCSAT-1 data, even in the conjugate hemisphere.

12 12 MAKELA: LOW-LATITUDE IMAGING PASSIVE OPTICS SELF-ASSESSMENT process. They also confirmed the conjugate nature of the phenomenon. Tinsley et al. [1997] compared data collected from Haleakala, Hawaii during August 1988 to insitu data collected by DMSP F8 and F9, as well as the San Marco satellite. They pointed out the advantage of having continuous imaging information in allowing for the identification of the same depletion in consecutive orbits. Another of the advantages of imaging was shown in that the images provide a two-dimensional context for the onedimensional in-situ data. This is echoed in recent results from the observations currently being made from Haleakala, Hawaii as shown in Figure 9. The composite image shown in Figure 6 has been projected onto a map and shifted based on the calculated zonal drift velocity to estimate the location of each individual depletion at 0740 UT. The insitu ion density measured by ROCSAT-1 [Su et al., 2001] at 600 km has also been plotted. To better facilitate the comparison, the composite image has been mapped along the magnetic field lines to the altitude of the satellite. It is clear from this Figure that the images collected from a narrowfield imager at Haleakala, Hawaii (the field-of-view of which is shown by the arch) can be used to estimate the location of the depletions to the east (in realtime) and west (after the fact). Another useful comparison has been performed by Kelley et al. [2003] in which ground-based images from Haleakala, Hawaii are compared to space-based observations obtained by the Global Ultraviolet Imager (GUVI) on board of NASA s TIMED satellite. They presented simultaneous images taken from GUVI of the nm emission and from Haleakala, Hawaii of the nm emission (both of which are due to radiative recombination of O + ) showing the same depletion. This study showed that a space-based optical instrument can indeed be used to study the occurrence of depletions and, in the case of an instrument in a low-earth orbit such as GUVI, can provide a global context. However, the ground-based instrumentation is still required to provide the high spatial and temporal resolution necessary to study the dynamics of the depletions. Of course, comparison to other ground-based instruments are also of value. Much work is currently underway to compare the occurrence of airglow depletions and the scintillation of trans-ionospheric radio signals, which is one of the major effects of the Convective Ionospheric Storm. Kelley et al. [2002] have shown that there is an excellent correspondence between strong scintillations on the GPS-L1 frequency ( GHz) and the location of equatorial depletions observed from Hawaii. Makela et al. [in press] pointed out that depletions are a necessary condition for GPS scintillations, but scintillations are not always present when there are depletions. They explained this by noting that scintillations, especially at GPS frequencies, require a significant background electron density. Thus if the density is too low, scintillations may not occur even though the underlying depletion structure is still present. The small-scale structures responsible for the scintillations at GPS frequencies also decay faster than the overall largescale features observed in the images. Thus, as the structures age, they are still visible in the images (typically until sunrise) even though they no longer create scintillations. Mukherjee et al. [1998] showed a direct correspondence between scintillations on a 244 MHz satellite link and the passage of an airglow depletion. de Paula et al. [2002] also used GPS measurements to show that the smallscale irregularities inside of the depletions can drift at a larger zonal velocity than the large-scale depletion structure. 3. FUTURE DIRECTIONS Over the past 30 years, we have come a long way in imaging low-latitude irregularities. As the technology has matured from film- to CCD-based systems, we continue to make new insights. This is especially true when optical instruments are combined with other observing techniques (e.g., radio and satellite data). However, there is still much to be learned about the processes at play in the post-sunset equatorial ionosphere. Below is a partial list of some of outstanding questions, that in this author s mind are particularly interesting and solvable in the near future. 1. Perhaps the largest unanswered question is regarding the seeding mechanism for the RTI process. Although imaging provides more of a monitoring tool for developing and fully developed plumes, it will still be useful in addressing some of the seeding questions. For example, by studying the distance between individual depletions, several studies have concluded that gravity wave seeding is a plausible mechanism [Fagundes et al., 1999; Sinha and Raizada, 2000; Makela and Kelley, 2003]. Similarly, the results of Makela et al. [in press] suggested that there was a strong seeding region 900 km to the west of Hawaii that produced more depletions from June through August than were produced at the Hawaiian longitude. Thus, although imaging may not be able to actually observe the seeding mechanism, observations of the developing and developed bubbles does provide useful information. By creating an east-west chain of imagers, the longitudinal dependence of depletion development could be studied and contribute to unravelling the seeding question.