Intense or severe convective storms are sometimes characterized by low production of cloud-to-ground (CG) lightning flashes during all or a portion of their lifetimes (e.g., MacGorman and Nielsen, 1991; Billingsley and Biggerstaff, 1994; Maddox et al., 1997). This anomalous lightning signature can be associated with very high intra-cloud (IC) lightning flash rates, as well as a higher-than-normal percentage of positive CG lightning (Carey and Rutledge, 1997). Thus, it is possible for a storm to produce large hail, strong winds and tornadoes, but have a CG lightning signature similar to that of a non-severe convective cell. Hence, as CG lightning data become more available for the "nowcasting" of severe weather, it is important to document and develop physically based explanations for low CG rates in certain intense storms.

Several hypotheses have been offered to explain this anomalously low production of CGs. One hypothesis invokes the non-inductive charging theory of cloud electrification (NIC). According to NIC, charge is separated through collisions between graupel and ice crystals in the presence of supercooled water. The sign of the charge transferred to the graupel particle depends on the environmental temperature, the liquid water content (LWC), and hydrometeor terminal velocities. For typical values of cloud LWC, however, the sign of the charge transferred to the graupel particle changes at a given temperature, referred to as the charge-reversal temperature. Takahashi (1978) found this temperature to be approximately -10 °C. Other researchers (e.g., Saunders et al., 1991) have found somewhat different, but still comparable, charge reversal temperatures. At temperatures lower than this, in the upper regions of the thunderstorm, the graupel charge negatively and the ice crystals charge positively. At temperature higher than this, near the base of the cloud, the graupel charge positively and the ice crystals charge negatively. This lower region of net positive charge has been discussed in Williams (1989), and is thought to provoke CG lightning (Clarence and Malan, 1957; Williams et al., 1985; Williams et al., 1989a). However, based on the results of Saunders et al. (1991) and Saunders and Brooks (1992), significant charging should not occur during wet growth regimes, as the rebounding efficiency of collisions goes to zero in such situations. If substantial wet growth is not occurring in these lower regions of the cloud, above the charge-reversal temperature, then the lower positive charge region should not develop as strongly. Hence, CG lightning might be suppressed. Note, however, that the results of Saunders et al. (1991) and Saunders and Brooks (1992) are in contradiction to the results of Takahashi (1978), especially as interpreted by Williams et al. (1991), who found significant charging in regimes of probable wet growth. This conflict has not been resolved yet.

A second hypothesis is that for intense storms with high precipitation rates, the electrical current associated with charged precipitation (precipitation current, J_P) may be large enough to offset charge carried to ground by the CG lightning current (J_L). Results vary widely, but the current state of research suggests that precipitation current (Rust and Moore, 1974; Moore, 1976; Moore and Vonnegut, 1977; Gaskell et al., 1978, Marshall and Winn, 1982; Soula and Chauzy, 1996; Baranski, 1996) may be comparable to CG lightning current (Livingston and Krider, 1978; Krehbiel, 1981), at least in an average sense. Hence, it may be possible that in certain intense convective storms, the instantaneous precipitation rates become high enough that J_P is large compared to J_L. In this case the precipitation current may remove separated charge quickly enough that CG lightning is less favored.

A final hypothesis is known as the elevated dipole hypothesis (MacGorman and Nielsen, 1991). In this hypothesis, intense updrafts of severe storms loft the negative charge center and the upper positive charge region to greater heights. (See Williams, 1989, for a discussion of the tripole structure of thunderstorms.) Because of the increased distance between the negative charge center and ground, IC lightning should be preferred over CG lightning. In addition, such strong updrafts could enhance charging and hence flash rates. Thus, enhanced IC lightning could neutralize enough charge so that when the cores do descend as the cells decay, CG lightning still is not favored because of lower electric fields due to less space charge.

Before a discussion on how these hypotheses will be examined, some background material on convective storm classification will be reviewed. Convective storms can be classified in terms of three major types: ordinary single cell, multicell, and supercell (e.g., Weisman and Klemp, 1984). Ordinary single-cell storms are characterized by a single updraft in its developing stages, which is destroyed by evaporative cooling due to entrainment of dry air as well as by precipitation loading. The development of a downdraft in response to these processes signals the decay phase of the cell. When the downdraft strikes the ground it spreads out horizontally as a gust front. Should this gust front trigger the development of new cells, then a multicell storm can result. Such a storm is characterized by the continued decay of old cells and the subsequent development of new cells, typically on a preferred flank. Supercells consist of a single long-lived cell which has a rotating updraft-downdraft couplet. The separation of the updraft and downdraft allows the storm to exist for a considerable period of time, as the updraft and downdraft mutually support one another. The major factors determining whether single cell, multicell, or supercell storms will develop in a given area are the Convective Available Potential Energy (CAPE) and the ambient wind shear (Weisman and Klemp, 1982, 1984).

Additionally, multicell storms may be further subdivided into two types: non-severe and severe, the latter defined according to United States National Weather Service (NWS) criteria as producing large (> 2 cm in diameter) hail or winds with speeds of 50 knots or more. Severe multicells typically are characterized by strong updrafts which tilt over the downdraft, so that precipitation falls out of the updraft into the downdraft, and thus loading is not as effective at weakening the updraft. The strong updraft allows for the development of large hail.

Up to this point the discussion has focused on storms under the assumption that they remain the same type throughout their lifetimes. However, this need not be the case. During Phase A of the Stratosphere-Troposphere Experiments: Radiation, Aerosols, Ozone (STERAO-A) field project, which will be discussed in the next section, many storms could not be classified as a single type throughout their lifetimes. There was a case where a storm began as a multicellular line, but later became unicellular in character. There were many cases of multicell non-severe storms which became severe at some point in their life cycles. These transitions often were accompanied by major shifts in the storms' lightning patterns. These included changes in the IC-CG lightning ratio, changes in IC flash rate, changes in CG flash rate, and changes in dominant CG lightning polarity (i.e., positive vs. negative CGs).

Such storms provide an excellent opportunity to examine the hypotheses for the significant reduction of CGs in some intense convective storms. By examining a storm's lightning and microphysical structure before, during, and after a major transition, any changes in these features can be clearly documented. By noting what has changed and what hasn't as the storm undergoes a major evolution, a better understanding of the relationships between storm microphysics and lightning can be obtained. Knowledge of these relationships then can be used to test the aforementioned hypotheses.

In addition, the study of these transitions provides an additional benefit for the "nowcasting" of severe weather. Documentation of any notable changes in lightning patterns during these transitions can assist forecasters in attempts to determine whether a given storm is transitioning from non-severe to severe, or from severe to non-severe.

1.2 Overview of the STERAO-A field project and the case studies

Phase A of the Stratosphere-Troposphere Experiments: Radiation, Aerosols, Ozone (STERAO-A) field project took place during the summer of 1996 over the plains of northeastern Colorado. The purpose of the project was to study trace gas production and transport by deep convection, especially nitrogen fixation by lightning. The project was centered around the CSU-CHILL multiparameter Doppler radar near Greeley. The radar was used to characterize the microphysical and kinematic structure of sampled storms. Also available was the Office National d'Etudes de Recherches Aerospatiales (ONERA) VHF lightning interferometer (ITF), which maps VHF emissions from lightning in three dimensions. The National Oceanic and Atmospheric Administration (NOAA) WP-3D aircraft had several on-board instruments to make low- to medium-altitude concentration measurements of various trace gases, especially NO_x (NO and NO₂). The WP-3D also had a tail-scanning X-band Doppler radar to characterize the storm-relative wind field via dual-Doppler observations. Instruments on board the University of North Dakota (UND) Citation aircraft made high-altitude trace gas concentration and in situ microphysical measurements of several storms, and often worked in concert with the WP-3D. There was a network of three field change meters (FCM; also know as flat plate antennas) - one mobile, two fixed - to make measurements of total flash rate for storms within the nominal sensor range of 35-40 km. The fixed sites were the CSU-CHILL radar and the Fort Morgan Airport, approximately 75 km ESE of CHILL. The mobile field change meter was operated from a chase van that was deployed from CHILL. Additionally, an electric field mill was stationed at the Fort Morgan Airport. Finally, project participants had access to CG data from the National Lightning Detection Network (NLDN).

Figure 1.1 is a schematic map of the area in which the STERAO-A project took place. Included are the positions of the CHILL radar which had a maximum range of 150 km during STERAO-A, the positions of the ITF receiving stations and the ITF's two best-resolution lobes, and the positions of the fixed FCMs along with range rings.

The presence of data from both the CHILL radar and the ITF provided a unique data set for which to study the relationships between convective storm microphysics and associated lightning. This research utilized data from the CHILL radar, the ONERA ITF, the NLDN, and the flat plates; and focused on two case studies of deep convective storms, 10 and 12 July 1996. (All dates and times are in terms of the local time, Mountain Daylight Time.)

The 10 July 1996 storm began as a multicellular line but toward the end of its long lifetime it became unicellular in nature, with characteristics that are often displayed by supercells, such as mid-level rotation and a Weak Echo Region (WER). Throughout the storm's lifetime, CG rates were extremely low, peaking at only 9 CGs in a 5-minute period, but usually varying between 0 and 2 CGs per 5-minute period. The IC flash rate for this storm was relatively high, around 20 ICs per minute throughout much of the storm's lifetime. The radar-inferred precipitation fluxes, especially the hail flux, trended the IC rates well, but sometimes precipitation peaks lagged the IC peaks by up to several minutes. IC flashes peaked near 50 per minute during the storm's quasi-supercellular stage. During this stage there were almost no CGs.

The storm of 12 July 1996 was a multicell storm which became severe before full volume radar scanning was initiated. This storm reverted to non-severe status during the course of detailed observations. Before this transition, radar-inferred hail fluxes were quite high, but afterward, hail fluxes were small. Before the transition, both negative and positive CG rates were relatively low, usually well under 5 per 5-minute period. Positive CGs made up a significant fraction of this small CG population. After the transition, however, positive CG rates remained low but negative CG rates exploded, varying between 10 and 25 per 5-minute period for over an hour.

1.3 Scientific objectives and organization of the thesis

The primary objective of this research is to analyze a unique data set consisting of concurrent measurements from the CSU-CHILL radar and the ONERA VHF interferometer, and during the course of this analysis to examine the validity of various hypotheses offered to explain the low production of CGs in some intense convective storms. To review, these hypotheses are: significant wet growth of graupel and hail that would suppress the development of a lower positive charge region (which normally could act to provoke CG lightning), removal of charge by large precipitation currents relative to lightning currents, and the elevated dipole mechanism acting in concert with enhanced IC flash rates. Offshoots of this objective are the presentation of detailed case studies of major transitions in convective storms and the documentation of additional cases of low-CG intense convective storms. This information could have potential use in the "nowcasting" of severe weather.

A secondary objective is to provide an intercomparison of the data from the various lightning sensors; in particular, to provide verification for the ITF flash rates.

This thesis is organized into six chapters. After this introductory chapter, Chapter 2 discusses the various sensors used in this research, and the methodology used to interpret their data. Chapter 3 presents results from an intercomparison of the various lightning sensors. The results are interpreted in order to determine a strategy for how best to include the ITF data in the research. Chapter 4 presents the first case study, the storm of 10 July 1996. Chapter 5 presents the second case study, the storm of 12 July 1996. Chapter 6 presents conclusions and recommendations for future research.