6.1 Conclusions

The major results of this work may be grouped as follows.

Indirect measurement of ionization

Determining the degree of ionization in sprites has been a recurring theme and point of debate over the last decade, largely due to the difficulty of measuring optical output from emission-lines of ionized molecular states. However, based on the modern understanding of the streamer breakdown process [Pasko et al., 1998a; Pasko et al., 2000] it seems incontrovertible that intense ionization occurs when filamentary structure is seen in sprites, although this ionization may occur only briefly [Armstrong et al., 1998a]. In this work we have shown three additional ways to infer the existence and degree of ionization in sprites and elves:

• With the help of modeling, upwardly-concave curvature in sprite halos can be interpreted as a sign of intense ionization augmenting conductivity gradients and preferentially expelling the electric field.
• The exponential relaxation of sprite luminosity appears to manifest ionization changes, typically over more than one order of magnitude. The rapid (mostly $<$1 ms) relaxation constant may also give insight into why blue (ionized) emissions are only seen briefly near the beginning of sprite optical signals.
• A comparison between the experimentally determined spatial extents of elves and a model of the lightning electromagnetic pulse interaction with the lower ionosphere suggests that the heating effect is often easily large enough to include $>$200 km regions of enhanced ionization.

Remote sensing of the electric field

In addition, an unexpected new method of remotely sensing the electric field within sprites was found when bright optical pulses were seen to relax exponentially. For a vertically narrow field-of-view for which the optical source altitude can be well constrained, the relaxation time constant may be related directly to the electric field by the curve $\ensuremath{\nu_{\rm i}}-\ensuremath{\nu_{\rm a}}$ in Figure 2.4. This technique could be combined with the broadband two-color photometric method described in Section 4.5, which does not rely on the detection of weak ionized lines but does also require a well constrained viewing elevation angle in order to determine the electric field at the source.

Discrimination of 1 ms flashes

Following distant, strong cloud-to-ground lightning, at least three classes of optical emissions are observed to have characteristic durations of $\sim$1 ms. These are scattered lightning flashes, elves, and the diffuse upper portion of sprites, observed as sprite halos. In addition, further sprite development (with streamer breakdown) may be observed to occur for several to many milliseconds.

Telltale signatures of elves, Rayleigh-scattered lightning flashes, and sprites were determined for the Fly's Eye photometric array and used to discriminate between these phenomena and to correctly associate them with their physical causes. Photometry with the Fly's Eye is a robust and sensitive method for identifying elves and determining their horizontal extent and lowest altitude extent.

In addition, the distinction between video signatures of elves and sprite halos was elucidated for the first time, based on measurements from a high speed image-intensified video system. In these measurements sprite halos are observed as a transient descending glow with lateral extent on the order of 40 to 70 km preceding the development of streamer structures at lower altitudes. These characteristics agree well with recent theoretical analysis of electrical breakdown properties at different altitudes [Pasko et al., 1998a], previous sprite modeling using the quasi-electrostatic (QE) model [Pasko et al., 1997b], and our one-to-one fully electromagnetic modeling of sprite driving fields and optical emissions for the observed events.

The introductory comment of Barrington-Leigh and Inan [1999] that ``video recordings at standard frame rate are an inefficient and sometimes confusing method for identifying elves in comparison with a photometric array'' seems now to be even more compelling given the common misidentification of sprite halos. In addition, high temporal resolution is clearly needed for both horizontal and vertical photometer arrays to discriminate between elves and sprite halos.

Verification of theoretical models

Asymmetries previously observed in the polarity of lightning seen to cause elves and sprites contradicted several of the predictions for the EMP and conventional breakdown QE models.

The bias of elves' occurrence towards an association with cloud-to-ground lighting of positive polarity likely disappears when lightning statistics are taken into account and a fair triggering method is used for data acquisition. Indeed, the correspondence between occurrence and optical intensity of elves is shown in Section 4.2 to be fully consistent with the lightning electromagnetic pulse model.

The observation of at least two ``negative sprites'' shows that conventional, rather than relativistic runaway, breakdown is responsible for the discharge in (at least some) sprites. This observation further suggests that the rarity of negative sprites may result primarily from a rarity of large charge moment changes in association with negative cloud-to-ground lightning. Pasko et al. [2000] has recently quantified the slight asymmetry expected between positive and negative streamers forming sprites.

Ubiquity of sprites and elves

The importance of sprites and elves in chemical, charge, and energy balances of global scale is still under investigation. In this work elves were shown to be a ubiquitous phenomenon, not likely to be restricted to large mesoscale convective systems in the way that sprites may be. In addition, elves may cause significant cumulative modification to the nighttime electron density profile near 90 km altitude over large regions of thunderstorm activity.

While negative sprites are reported as a rare event by Barrington-Leigh et al. [1999a], occurrence of the sprite halos reported here is not unusual in association with $-$CGs, based on Fly's Eye and normal-rate video observations (e.g., Figure 5.1). Due to the exponential reduction with altitude of the fields required for ionization and optical excitation, the diffuse region is easier to excite with lower driving fields (i.e., lower thundercloud charge moment changes). It may be more frequently excited by strong negative lightning even though negative strokes do not often carry long-duration continuing currents. Sprite campaigns which are focused on storms producing strong positive cloud-to-ground strokes may not adequately account for the occurrence of what one may call ``negative sprite halos.''