Severe and extreme weather events in the form of storms, floods and tornados can result in fatalities, injuries and economic losses. It is possible to explore their consequences using the US National Oceanic and Atmospheric Administration - NOAA - detailed weather reports, which are publicly available and cover the period from 1950 to 2011. The data contains a great a deal of detail; so much so that instead of a neat categorisation of weather events - e.g. flood, tornado… - there is a plethora of 985 categories consisting of:
- abbreviations, like tstm instead of thunderstorm,
- clarifications, like gusty wind and rain,
- typos, like torndao instead of tornado
- and vague terminology, like red flag criteria
Which can, nonetheless, be sorted using regular expressions - see how I did it - to obtain 9 major categories:
| Hot weather |
Rain | Wind | ||
| Cold weather |
Flood | Tornado | ||
| Thunderstorm | Fire | Other |
After which we are ready to provide some answers on their consequences.
Which types of events are most harmful with respect to population health?
Tornados have been by far the most dangerous weather event in the US. 60,700 tornados were registered from 1950 until 2011, injuring 91,407 and killing 5,636 people. Tornados were also the major cause of injuries, up to 65% of the total, while the next major cause of injuries, rain, represents only 10%.
Which types of events have the greatest economic consequences?
Floods have been the major cause of economic losses from 1950 to 2011. In that period 86,136 floods were reported and while they caused damages in the crops valued in 7,4 billion dollars, the resulted in property damages reaching 167 billion dollars.
Weather evolution
The record is only complete for all weather events from 1993 onwards. In those years the most significant characteristic is the rapid increase in the number of thunderstorms, although floods also display some increase and interestingly there is an abrupt leap in cold weather in recent years. The large increase of “other” events may be a result of the categorization established in the data cleaning process but it may also reflect the fact that many more minute appreciations have been registered in recent years, such as ““red flag criteria”, “remnants of floyd” or “waterspout funnel cloud”, which do not provide very specific information and fell in the category other.
The original study was carried out for the course Reproducible Research. You may also find the R code on GitHub.