Reports of damaging weather phenomena are collected monthly by National Weather Service offices. Individual station reports are sent to NOAA's National Climatic Data Center in Asheville, NC where Storm Data is assembled. It has been compiled with essentially the same procedures since 1959.
For deaths and injuries, additional information besides these standard data were extracted manually for 1991 to 1994. Activities and locations of victims were identified and categorized in as much detail as provided by the verbal narratives in Storm Data.
Damage costs are reported in Storm Data by numbers ranging from 1 to 9. Category 1 has a range of $0 to $50, category 2 is $50 to $500, and so on. Expanded information on the objects impacted by lightning was sometimes available in the verbal narratives in Storm Data. In a few cases, an exact amount was mentioned in Storm Data beyond the standard categories.
If a single structure was described in Storm Data as catching fire, this was classified as such. But if no mention of fire was included, the entry was identified as "damage-no fire." If Storm Data combined more than one structure into a single entry, and nothing is mentioned about fire, it was listed as "fire unknown."
Newspapers across the United States were scanned for lightning entries and summarized in Kretzer (1895). The goals of the book were to:
•Determine the potential threat from lightning to people and property in
the US
•Document lightning incidents and protection.
•Encourage the public to use lightning rods.
The efforts of Kretzer led to the first collection of such information by federal agencies starting in 1890 (López and Holle, 1998).
The newspapers scanned by Kretzer (1895) in Table 2 are biased toward the region around his St. Louis home. But no publication contained as much detail as this book for decades before and after it was published. In addition, Kretzer lists "Special reports of correspondents'' as a source for some cases, and solicits additional cases for a future volume that does not appear to have been published. There were 894 accounts from 470 cities and towns in 42 states. In addition, 16 events were entered from 8 foreign countries.
Both standard and expanded information on lightning casualties and damages were classified from Kretzer's book from 1891 to 1894. The same system was used as described in section 2A for analyzing Storm Data from 1991 to 1994.
Damages in Kretzer's book were sometimes given in dollar amounts. They were matched with the same categories as in Storm Data. If the Storm Data category was 3 or less ($500 to $5000), then it was considered "minor damage." If it was category 4 or above, it was "heavy damage." If there was more than one house in the same entry, and no damage was directly given for each house, no estimate was attempted and the amount was reported as unknown.
| Illinois | Chicago | 4 |
|---|---|---|
| Indiana | Michigan City | 1 |
| Iowa | Burlington | >1 |
| Davenport | 1 | |
| Des Moines | 1 | |
| Prairie City | 1 | |
| Missouri | Marshall | 1 |
| St. Louis | 5 | |
| New York |
New York City |
1 |
| Nebraska |
Bartlett |
1 |
| Ohio | Cincinnati | 1 |
| Akron | 1 | |
| Alliance | 1 | |
| Wisconsin |
Madison |
1 |
Absolute values of lightning-caused casualties and damages must be considered with caution. Most of these events are less spectacular and more widely dispersed in time and space than tornadoes and hurricanes, for example. Lightning deaths, injuries, and damages have been found to be underreported in the following studies:
Factors contributing to underreporting are:
López et al. (1998) documented a significant decrease in lightning deaths over the last century. The decrease was almost parallel to the decrease in the percentage of the US population living in rural situations, as found by the US Census Bureau. Similar results have been found in other countries.
Therefore, an overall impression was developed from each entry in Kretzer (1895) and Storm Data as to whether the situation was rural or urban. This impression was made only when the situation was clearly described in the verbal narrative. It was not possible to make the rural-urban determination in about a third of the cases. For example, an unknown case frequently occurred when someone was killed or injured by lightning inside a house. If nothing else was mentioned about the setting external to the house, neither a rural or urban classification could be assigned to the casualty.
| Three settings are used in this study:
*Rural |
Storm Data provides an entry classifying the location of victims. But the location is unknown in 40% of the cases in Storm Data, and another 27% are in "Open field, ballparks, playgrounds, etc." (Curran et al., 1997, 2000).
For this study, activity and location were first identified separately, directly from the narratives for each casualty entry. For example, entry number 73 reporting a lightning death might have a location of "under a tree" and an activity of "seeking shelter from rain." This casualty would have code 9 for location and code 192 for activity. In many cases it was impossible to determine the location, activity, or both from Kretzer's book or Storm Data. In those cases, the datum was coded as 99 ("Unknown, Not Reported"). Another death entry might have "under a tree" as the location but "camping" as activity. In this case, the codes would be 9 for location (same as before) but 7 for activity.
Coded location and activity were, in turn, assigned separately to one of seven classes: Agriculture, Indoors, Outdoors, Recreation, Sports, Small Structures, and Unknown. In the two examples mentioned above, location 9 ("under a tree") would be assigned to Outdoors, activity 192 ("seeking shelter") to Outdoors, while activity 7 ("camping") would be Recreation.
A lightning casualty was then assigned an incident category according to how its location and activity were classified. These types were the same as the seven classes above. Thus, the first case would be of overall type Outdoors, while the second would be Recreation.
The typing was done with fixed rules to combine location and activity. When the two classes were the same, typing was obvious. But for a large number of entries, location or activity was missing, and the type corresponded to the non-missing class. In a few instances, the two classes differed and preference was given to activity, as in the second example above. However, location was dominant when the activity was Indoors but location was otherwise. Location was also dominant for an Outdoors activity with a location of Agriculture, Recreation, Small Structure, or Sports. These rules assured that the most specific information was given the most weight in classifying the incident.
The seven types of activities, locations, and types of incidents, in alphabetical
order, are:
The incident took place on a farm or ranch. The location was more substantial than a garden in the backyard of a home in the city. It typically was where the livelihood of the lightning casualty and related family involved agriculture. If there was no specific mention of an agricultural setting for the casualty, it was not included here.
| Seven activities, locations, and types of incidents
are used in this study: *Agriculture |
Next: Chapter 3: Lightning Deaths, Injuries, and Property Damages in the 1890s