Two case studies of the relationships between the multiparameter radar-inferred microphysics of convective storms and associated lightning have been presented. Both case studies involved thunderstorms which underwent major transitions in structure and intensity sometime during their lifetimes. Emphasis was placed on examining these major transitions to better understand the relationships between microphysics and lightning, especially to examine hypotheses to explain low cloud-to-ground (CG) flash rates in certain intense convective storms. The case studies utilized data from the CSU-CHILL multiparameter Doppler radar, the ONERA VHF lightning interferometer (ITF), and the National Lightning Detection Network (NLDN). The CHILL radar was used to characterize storm structure and to infer both rain and hail rates, and also to differentiate between regions of different bulk precipitation types. The ITF was used to infer intra- cloud (IC) flash rates, while the NLDN provided CG flash rates and ground strike locations.

The storm of 10 July 1996 underwent a major transition approximately halfway through the observation period, evolving from a multicellular line to an intense unicellular storm which featured some characteristics common to supercells, such as mid-level rotation and a Weak Echo Region (WER). Throughout this storm’s lifetime CG flash rates were extremely low, averaging 0-2 CG flashes per 5-minute period. During the peak CG flash rate period flashes typically were associated with less intense (in terms of vertical radar reflectivity structure) cells along the line, while the most intense cells were associated with the bulk of the IC flashes. During the storm’s intense unicellular stage, CG production was zero, while IC flash rates peaked near 50 per minute. In general, IC flash rates trended the low-level precipitation fluxes well, although occasionally these fluxes lagged the IC flash rates. Small (< 2 cm) hail was produced throughout the storm’s lifetime, though overall precipitation fluxes were significantly lower compared to other intense storms (Carey and Rutledge, 1997).

The storm of 12 July 1996 underwent a major transition from an intense, hail-producing multicellular storm to a weaker multicellular rainstorm with little or no hail production. Before this transition, CG flash rates were low (for both positive and negative polarity flashes), averaging less than 5 per 5-minute period. Afterward, negative CG flash rates rose significantly, to 25 per 5-minute period at their peak. IC flash rate led the fallout of significant precipitation early in the storm’s lifetime, but afterward it lagged the low-level precipitation fluxes. This may be due to the superposition of the contributions from multiple cells to these parameters, which could obscure the usual single-cell trend of IC flash rates leading the fallout of significant precipitation (e.g., Williams et al., 1989a,b). When hail was produced by this storm, it was predominantly small hail. Overall, precipitation fluxes seemed comparable to the fluxes seen by Carey and Rutledge (1997). However, Carey and Rutledge (1997) studied a storm that produced significant quantities of radar-detected large hail, unlike 10 and 12 July 1996.

There are some interesting features common to both of these storms. One is that when hail was produced by either storm, even if such production underwent significant pulsing, CG flash rates were low. Another common feature is that CGs were most often associated with weaker cells, in terms of the vertical structure of radar reflectivity and the lack of hail production. The most intense cells typically were associated with the bulk of the IC production. A final common aspect is that, especially when CG flash rates were low, these storms did not seem to follow the common pattern, established by other researchers (Williams et al., 1989a,b; Carey and Rutledge, 1996; Changnon, 1992), of CG flash rates peaking when significant precipitation reached the ground. In fact, CG production was so low during these times that it was difficult to resolve any trends at all. CG flash rates remained low regardless of whether individual cells were developing or collapsing. Only when hail production was largely terminated, on 12 July, did CG rates increase. On 10 July, when hail production was terminated the storm itself collapsed, and no final burst of CGs occurred.

Based on this evidence, the elevated dipole hypothesis (MacGorman and Nielsen, 1991) seems to be a reasonable explanation for the low production of CGs. In cells producing the most CGs, the available proxy data imply weaker updrafts than in cells producing little or no CGs. In the CG-producing cells, vertical reflectivity structure was weaker, in that high radar reflectivity contours either did not exist or extended to lower altitudes than the most intense, low-CG cells. The existence of hail is another proxy for updraft strength, in that intense updrafts are needed for hail production to occur. Thus, if there is little or no hail being produced - as during the latter half of the 12 July storm’s lifetime - this implies weaker updrafts, on average, than a hail-producing storm.

Recall that the elevated dipole mechanism involves intense updrafts shifting the negative charge region of the storm (e.g., Williams, 1989) to higher altitudes than average updrafts would suspend this charge region (MacGorman and Nielsen, 1991). This higher altitude should then disfavor CGs, due to the reduced (on average) electric field between the negative charge region and ground. However, based on either convective or precipitation-based charging theories, substantial charging should still be occurring, owing to the intense updraft. Thus, IC flash rates should be higher than an ordinary (i.e., single-cell or non-severe multicell) storm, especially since the charge neutralization process provided by CG flashing is reduced.

These enhanced IC flash rates may be the reason that CGs still are not produced in even moderate quantities when individual cells reach maturation in the storms of 10 and 12 July. High IC flash rates could have served to neutralize most of the produced charge before it began its descent. With reduced charge on the descending precipitation, CGs could be suppressed.

It may be illustrative to view these two storms as points on a continuum of updraft speeds. On one end there are the cells with very weak updrafts and which produce little or no lightning of any type. Then, further along the updraft scale, there would be the “garden-variety” storms, whose individual cells favor average IC lightning flash rates in their developing stages, but once their cores reach ground the production of CG lightning occurs. These are the type of thunderstorms that Williams et al. (1989a,b) and others studied. They may or may not produce hail, but if they do so, it would be only a very small component of the precipitation flux. Then there would be storms like those of 10 and 12 July, with strong updrafts and high IC flash rates, but relatively low CG lightning production. These storms produce mostly small hail. Then, at the most extreme end, there are the intense severe storms which produce significant quantities of large hail and positive CG lightning, such as those studied by Carey and Rutledge (1997) and others. In these storms, updraft strengths may be so high that large quantities of positive charge are created and then deposited in the upper levels of the storms. This enhanced region of positive charge would provide a source for positive-polarity CGs, despite the high IC flash rates observed in these cases.

This continuum idea is especially consistent with the present cases, since positive CG lightning was present in both storms, though in smaller quantity than the storms studied by Carey and Rutledge (1997) and others. In the storms of 10 and 12 July, positive CGs, when they occurred, were typically associated with the most intense cells, which made the fraction of positive CGs produced by the most intense cells very high (~ 50% or more), even though their overall CG production was low. Perhaps, with somewhat stronger updrafts and greater longevity in the severe phase, these storms could have become significant positive CG producers themselves.

If one accepts the results of Saunders et al. (1991) and Saunders and Brooks (1992), and significant charging does not occur in wet-growth regimes, then the storms of 10 and 12 July might have produced enough wet-growth hail to prevent the production of a lower positive charge region (below the main negative charge region), whose existence is thought to stimulate CG lightning activity (Clarence and Malan, 1957; Williams et al., 1985; Williams et al., 1989a). Hence this is another mechanism that may account for the paucity of CGs in these storms. In even stronger storms such as the one studied by Carey and Rutledge (1997), wet growth may be more prevalent, and this mechanism would act more efficiently. Though the available observations imply that hail undergoing wet growth was certainly a possibility - indeed a likelihood - in these two storms, the observations also imply that the dry growth regime was more prevalent, due to the paucity of extended regions of enhanced LDR. Note that, because the identification of hail above the freezing level with multiparameter radar data is still in an exploratory stage (e.g., Kennedy et al., 1997), and because there are no in situ data available, this mechanism cannot be totally ruled out. If this mechanism were operating, no significant lower positive charge region should have formed, which could account for the low CG production of these storms. It would be consistent with the rise in CGs after hail fall ended in the 12 July storm, since then there should be little or no wet growth occurring, as hail was not present in bulk. Thus, during the latter half of the 12 July storm, a lower positive charge region could have formed and CGs once again would have been favored.

Note that this wet growth mechanism is compatible with the elevated dipole mechanism, and in fact could act in concert with the latter mechanism to suppress CGs. However, the lack of charging during wet growth observed by Saunders et al. (1991) and Saunders and Brooks (1992) is in contradiction with the laboratory results of Takahashi (1978), especially as interpreted by Williams et al. (1991), who found significant charging during probable wet-growth regimes. The debate over which results are correct is still ongoing.

It is difficult to rule out the possibility that the precipitation current could provide a feasible substitute for the CG lightning current, since charge on precipitation was not measured during STERAO-A. However, as noted in Chapter 1, several researchers have made estimates of the magnitudes of both precipitation currents (Rust and Moore, 1974; Moore, 1976; Moore and Vonnegut, 1977; Gaskell et al., 1978, Marshall and Winn, 1982; Soula and Chauzy, 1996; Baranski, 1996) and CG lightning currents (Livingston and Krider, 1978; Krehbiel, 1981). Given the wide range of estimates for both types of current, it seems possible that precipitation current could become high enough (provided precipitation fluxes are high) to transfer significant charge to ground and suppress CG lightning. Precipitation fluxes were relatively high for the storm of 12 July, but were much lower during the 10 July storm. For both of these storms the major contributor to the precipitation flux was rain. The rain probably is not contributing enough to the precipitation current to cause suppression of CGs, as rain rates for 12 July were significant both before and after the major transition in the storm. Also, these rain rates were much higher than those that occurred during 10 July, which also featured low CG flash rates. Thus, any hypothesis that invokes the precipitation current due to rain to account for low CG rates faces these counterpoints.

However, what if the hailstones were charged? As noted before, negative CG production is negatively correlated, in a bulk sense, with the production of hail. If the hailstones themselves were negatively charged, then the observed hail fluxes may be large enough that the precipitation current due to hail would help suppress CG lightning. The hail fluxes from either storm are typically within a factor of two of one another, so given the possible error involved in these flux calculations there may not be so much difference between the two days. However, note that this hypothesis requires that the hailstones be charged, in contradiction to the hypothesis based on the results of Saunders et al. (1991) and Saunders and Brooks (1992) which contend that hail which underwent significant wet growth would not be charged. Thus, resolving this contradiction would require that the bulk of hail growth for these storms occur in the dry regime. Based on the available data this probably occurred. However, note that hail probably is not a significant contributor to the overall precipitation current because of its typically low number concentrations (Cheng and English, 1983). Graupel should be the dominant charge carrier because it usually is in higher concentration. Note that if most of the graupel melted before reaching the ground, this charge would be carried by the subsequent rain. This perhaps is an argument for rain being the principle charge carrier in these types of storms (i.e., storms like 10 and 12 July wherein most of the precipitation appears to be in the form of rain, regardless of whether hail is produced or not).

Based on this research, it appears that the exact mechanism(s) responsible for low production of CG flashes in certain intense convective storms is still unclear. The elevated dipole mechanism, in conjunction with increased IC flash rates, is perhaps the simplest explanation available, in that it does not require precipitation to carry any specific charge or lack thereof. It also is entirely consistent with the available evidence. However, other mechanisms may be active, either instead of or in concert with the elevated dipole mechanism.

Nevertheless, this study provides additional documentation of the phenomenon of low CG production in intense storms. It also establishes that the production of negative CGs can be inversely correlated - in a bulk sense - with the production of significant quantities of hail and probably also with the potential for severe weather, in these types of storms.

6.2 Recommendations for future research

The main limitations of the present study include the small number of case studies, the lack of direct data on updraft speeds, the lack of measurements of net charge on hydrometeors, and uncertainties in the interferometer data.

To better establish (or perhaps to refute) the results of the present study, more combined radar and lightning case studies of major transitions in convective storm structure are needed. Alternatively, case studies involving a spectrum of different convective storm types and strengths could be examined and contrasted, to provide more insight into what is causing any observed differences in lightning patters. Critical to such studies is the existence of IC flash data, since these data help to establish the extent to which a storm is electrified, and also provide evidence as to what portion of its life cycle (e.g., initial development, maturation, decay) a given cell is occupying. It is especially useful to have either interferometric (as in the present study) or time-of-arrival (e.g., Proctor, 1981) VHF lightning mapping systems, as these sensors locate lightning in three dimensions This allows for discerning the individual contributions of existing cells to the total storm flash rate.

Future studies also would benefit from direct data on updraft speeds, such as those provided by dual-Doppler radar syntheses. The NOAA WP-3D aircraft which flew during STERAO-A gathered dual-Doppler data for a number of storms, including 10 and 12 July. Currently, however, analyses of these data have not been completed so they were not included in the present study. Such data provide a more direct method of ascertaining updraft speeds, without relying on proxy data like vertical reflectivity structure and hail production.

A way to determine the relative magnitudes of the precipitation currents would be to make in situ observations of the net electrical charge on individual precipitation particles, to gain a sense of the average charge carried per hydrometeor. Using precipitation fluxes and size distribution information, estimates of precipitation currents could be made based on these individual charge measurements. Alternatively, a method to measure precipitation current in a bulk sense - perhaps along the lines of Soula and Chauzy (1996), Baranski (1996), or others - could be used. Data on precipitation currents could be used in conjunction with parameters such as lightning currents, displacement currents, and others to develop complete electrical budgets for a selection of low CG production storms. These budgets could be used to ascertain the relative magnitudes of the lightning and precipitation currents, to better understand if precipitation current could act to suppress CG lightning.

There are still several unknowns regarding the observations of the ONERA VHF lightning interferometer during STERAO-A. The most important of these is how the reduced vertical resolution during STERAO-A impacted the flash identification and classification algorithm, since this algorithm reconstructs individual flashes in three dimensions. The reduced resolution conceivably could impact the estimated VHF burst velocities and separation distances, which could cause the algorithm to produce corrupted flash statistics. This issue will be examined in the future. Also, the nature of the short-duration (< 1 ms) discharges is unknown. They may be a misclassification of individual lightning strokes, or a sub-class of IC discharges, or a combination of both. If they are at least in part a subset of IC discharges, it is not clear what role they would play in either the electrical or chemical budgets of thunderstorms.