Abstract

National maps of cloud-to-ground lightning flash density (in flashes per square kilometer per year) for one or more years have been produced since the National Lightning Detection Network (NLDN) was first deployed across the contiguous United States in 1989. However, no single publication includes maps of cloud-to-ground flash density across the domain and adjacent areas during the entire diurnal cycle. Cloud-to-ground lightning has strong and variable diurnal changes across the United States that should be taken into account for outdoor lightning-vulnerable activities, particularly those involving human safety. For this study, NLDN cloud-to-ground flash data were compiled in 20 km by 20 km grid squares from 2005 to 2012 for the lower 48 states. A unique feature of this study is that maps were prepared to coincide with local time, not time zones. NLDN flashes were assigned to 2-h time periods in 5° longitude bands. Composite maps of the 2-h periods with the most lightning in each grid square were also prepared. The afternoon from 1200 to 1800 local mean time provides two-thirds of the day’s lightning. However, lightning activity starts before noon over western mountains and onshore along the Atlantic and Gulf of Mexico coasts. These areas are where recurring lightning-vulnerable recreation and workplace activities should expect the threat at these times, rather than view them as an anomaly. An additional result of the study is the midday beginning of lightning over the higher terrain of the western states, then the maximum activity moves steadily eastward. These storms pose a threat to late-afternoon and evening recreation. In some Midwest and plains locations, lightning is most frequent after midnight.

Abstract

There is a major difference in population-weighted lightning fatality rates between the lower fatality rates in developed countries and the higher fatality rates in developing countries. The large decrease in annual rates of population-weighted lightning fatalities in the United States is described over the last century. A similar large reduction in lightning fatality rates has occurred during recent years in Australia, Canada, Japan, and western Europe, where there has also been a change from a mainly rural agricultural society to a primarily urban society. An important accompanying aspect of the lower casualty rates has been the widespread availability of lightning-safe large buildings and fully enclosed metal-topped vehicles, as well as much greater awareness of the lightning threat, better medical treatment, and availability of real-time lightning information. However, lightning exposure for many people in lesser-developed countries is similar to that of a century ago in developed countries. The number of people living in these areas may be increasing in number, so the number of people killed by lightning may be increasing globally due to these socioeconomic factors. It can be difficult to locate national lightning fatality data because of their mainly obscure publication sources. The present paper synthesizes lightning fatality data from 23 published national-scale studies during periods ending in 1979 and later, and maps these fatality rates per million by continent.

Abstract

Cloud condensation nuclei (CCN) at cloud base strongly affect the droplet concentration at cloud base, which in turn influences the life history of a cloud. There are usually more nuclei over land than over water because of surface sources of nuclei, and vegetation fires produce large numbers of nuclei which may keep much of a cloud's water in small droplets that fail to reach raindrop size. Smoke particles from drought fires are thought to result from the burning of four major vegetation types in South Florida. Samples of these were burned in the laboratory and produced (at 0.75% supersaturation) between 10⁹ and 10¹⁰ nuclei per gram burned. A simple calculation, assuming reasonable burn rates for these materials, resulted in 4600 CCN cm⁻² when mixed uniformly to cloud base over a large area.

The drought between 1 April and 15 May of 1967 over Florida was found to be related predominantly to synoptic-scale dryness and subsiding northerly winds aloft. There was no significant large-scale lag in rainfall caused by lingering CCN from fires and the dry surface, since dynamic changes explain the onset of normal rainfall. Individual cloud rainfall may have been affected by high CCN counts as indicated by cumulus model calculations. Liquid water fallout from small clouds is affected to a greater degree than from tall clouds.

Abstract

Radar populations of seedable clouds observed between 4.57 km (15,000 ft) and 7.62 km (25,000 ft) each hour were gathered near Miami and Tampa during various months through the year. They show that numbers of tops in this range vary greatly from year to year and one location to another. The maximum was 503 tops in July 1971 over 4396 km² near Miami; the minimum was zero in the identical area from 16 to 30 April 1970. Diurnal variations are also shown.

The one-dimensional Lagrangian cumulus model EMB 68K was used to calculate potential top growth due to seeding supercooled cumuli with silver iodide by the dynamic method. This computation was made for all Miami and Tampa soundings between 16 April and 31 May, 1969. Given cloud base at 610 m, radii were found that reached heights of 6.10 and 7.62 km unseeded; then the seedability was obtained for these radii with the numerical model. From the numerous statistics shown for these radii, it was determined, for example, that seedability is larger when clouds in the seedable height range are observed within 6 hr of sounding time compared to similar time periods without any seedable clouds.

The relationship between seedability and rainfall change based on a group of experimental clouds over South Florida was explored. Although not sharply definitive, a linear regression was determined. With this relation, it is possible to combine observed clouds and observed seedabilities to compute rainfall changes due to seeding. Calculations are made with a set of specific restrictions for simplification. Suggestions are given regarding improvements and further study, since this method of estimating dynamic seeding results without aircraft awaits refinements in both theoretical and observational understanding of cumuli before reliable quantitative results are obtained.

Abstract

Oceanic cloud populations were photographed from the R. V. Crawford stationed near 13N, 55W during a 22-day period in August–September 1963. A panoramic series of six pictures per hour was taken with a Hasselblad camera around the complete horizon during daylight. These pictures were reduced by photogrammetry to yield oceanic cumulus height distributions related to the synoptic state and time of day, information that previously has not been obtained often for one location in the tropics away from land influences. Such quantitative data are useful for several types of budget studies and should be applicable for adjacent portions of the tropical North Atlantic Ocean and Caribbean Sea. Within 18 n mi of the ship it was found that half the hours had one or more cumulus clouds over 9000 ft and 15% of the hours had clouds over 18,000 ft. Other height percentages were obtained for disturbed and undisturbed hours, and also for hours with 3-cm radar echoes near the ship. Both maximum and average hourly cloud tops were obtained. It was estimated that during as many as 60% of the hours within 18 n mi of the ship there was at least one cloud which precipitated during its lifetime. No cumulus were seen above 40,000 ft during the entire period and tops exceeded 30,000 ft during less than 2% of these daytime hours.

Clouds were measured to be highest at 0600 and 0700 LST and decreased somewhat throughout the day. Similar cumulative percentages for days rather than hours showed that on half the days during daylight at least one cumulus was above 20,000 ft. No quantitative nighttime population could be obtained from the data.

Radiosonde data at the ship were analyzed to find simultaneous relationships between the cloud height population and sounding parameters. Compared to normal conditions during the period, highest cumulus clouds tended to occur during hours with a) lower cloud base than normal, b) moister soundings from 1000–700 and 700–500 mb, c) less stable soundings from 1000–700 and 400–200 mb, and d) more stable soundings from 700–500 and 500–400 mb. A combination of these sounding parameters comprised an index which was fairly well correlated with the cumulus height measured simultaneously. Attempts were made to calculate vertical motion and divergence, but values were unreliable for several reasons and could not be included in the index. From this study it is apparent that temporal and spatial scales of the dynamic and thermodynamic variables smaller than the synoptic scale must be examined to explain many aspects of a tropical cumulus population.

Abstract

No abstract available.

Abstract

Lightning stroke density measured by the Global Lightning Dataset (GLD360) has shown several strong maxima around the globe. Several of these extremes are located over large tropical water bodies surrounded by terrain features. Four prominent maxima are examined and compared in this study: Lake Maracaibo in South America, the Strait of Malacca in equatorial Asia, Lake Victoria in East Africa, and Lake Titicaca in South America. Specifically, the authors observe that all four water bodies exhibit sustained maxima in lightning occurrence all night, the peak lightning frequency occurs very late at night or the following morning at three of the four sites, and the nocturnal maxima are out of phase at the four locations even though the afternoon maxima over the surrounding terrain all occur between 1500 and 1700 local solar time. The meteorological factors affecting the diurnal cycle of lightning occurrence over these four water bodies, which are all adjacent to mountains, are explored in this study.

Abstract

Temporal and spatial distributions of the North American monsoon have been studied previously with rainfall and satellite data. In the current study, the monsoon is examined with lightning data from Vaisala’s Global Lightning Dataset (GLD360). GLD360 has been operating for over three years and provides sufficient data to develop an exploratory climatology with minimal spatial variation in detection efficiency and location accuracy across the North American monsoon region. About 80% of strokes detected by GLD360 are cloud to ground. This paper focuses on seasonal, monthly, and diurnal features of lightning occurrence during the monsoon season from Mazatlán north-northwest to northern Arizona and New Mexico. The goal is to describe thunderstorm frequency with a dataset that provides uniform spatial coverage at a resolution of 2–5 km and uniform temporal coverage with individual lightning events resolved to the millisecond, compared with prior studies that used hourly point rainfall or satellite data with a resolution of several kilometers. The monthly lightning stroke density over northwestern Mexico increases between May and June, as thunderstorms begin over the high terrain east of the Gulf of California. The monthly lightning stroke density over the entire region increases dramatically to a maximum in July and August. The highest stroke densities observed in Mexico approach those observed by GLD360 in subtropical and tropical regions in Africa, Central and South America, and Southeast Asia. The diurnal cycle of lightning exhibits a maximum over the highest terrain near noon, associated with daytime solar heating, a maximum near midnight along the southern coast of the Gulf, and a gradual decay toward sunrise.

Abstract

Eight flash flood events occurred in the Tucson area of southeastern Arizona during the 1990 summer when a high-resolution lightning detection network was operated in the region. A total of 3479 cloud-to-ground lightning flashes was composited with respect to times and locations of these flash floods. The analysis region was a square of 40 km on a side that nearly coincided with a small hydrologic region that drains runoff from high mountains around Tucson and results in streamflow near and through the city.

Most lightning in the 40-km-square area occurred between 10 h before the flood and the reported time of the flood. Flashes were most frequent around 2 h prior to the flood, but advance timing was not consistent. The most important factor in determining whether a flash flood report followed lightning was the number of consecutive 5-min periods with two or more flashes in the 40-km-square area. Intensity of the maximum flash rate was not systematically related to the amount of flooding, except that the 2 days with highest lightning frequencies were associated with the most widespread flood effects of the summer in the Tucson area. No precursor was found in positive flashes. While lightning data identified many of the flash flood events and avoided most false detections, the sample size was very small and there were no other cases for an independent test.

Streamflow increased abruptly after the occurrence of lightning for two flood periods on 19–20 July and another flood on 24 July. About 2 h after the maximum flash rate, streamflow gauges at three locations in the hydrologic region measured rapid rises in runoff that were indicative of the arrival of flash floods. Additional studies with flashes from an operational lightning detection network need to take into account additional factors, such as different locations, seasons, storm types, and precipitable water in low to midlevels, in order to more fully explore the possibility of lightning strikes providing useful precursors of flash flood events, particularly in rugged semiarid terrain.

Abstract

Network-detected cloud-to-ground lightning coincident with mainly frozen precipitation (freezing rain, sleet, snow) was studied over the central United States during two outbreaks of arctic air in January 1994. During the first event, the ratio of positive to total flashes was 59%, flashes were few and disorganized in area, and no surface observer reported thunder. For the other event the ratio was 52% during the first few hours in subfreezing surface air, then decreased when flashes formed in the nearby region above freezing. Also, flashes in this case were linearly aligned and coincided with conditional symmetric instability; thunder was heard infrequently by surface observers. On radar, reflectivity cores grew from weak to moderate intensity within a few hours of the lightning during both cases. Echo area increased greatly before flashes in one case, while the area increase coincided with flashes in the other. Some base-scan reflectivities were strong in both thunderstorm regions due to the radar beam intersecting the melting level. Regions with lightning often could be identified better by high echo tops than reflectivity. Analyses on the scale of one or two states diagnosed the strength of low-level warming that contributed to formation of thunderstorms and significant frozen precipitation. Quasigeostrophic analyses showed that 850-mb temperature advection and 850–500-mb differential vorticity advection were similar in magnitude in the lightning area during both events. Once convection formed, lightning and echo-top information identified downstream regions with a potential for subsequent frozen precipitation.