Expected Annual Lightning Strikes to a Structure

How often will this building be struck? Multiply the number of flashes that reach ground per square kilometre per year in your area by the building's collection area, adjust for how exposed its surroundings are, and you have the expected number of direct strikes per year. In IEC 62305 it is written $N_D = N_G A_D C_D \times 10^{-6}$, where the $10^{-6}$ is nothing but the bridge from square metres to square kilometres.

The answer is an expectation, not a schedule, and the difference is the most important thing on this page. Lightning arrives, to a very good approximation, as a Poisson process. An expected 0.032 strikes per year does not mean "a strike every thirty-one years, next one due in 2049". It means that in any given year the probability is about 3.2 %, that two strikes in one summer is entirely possible and not evidence that the calculation was wrong, and that a century of nothing is not evidence either. A building "overdue" for a strike is a phrase with no physical content. If you need the probability of at least one strike in $n$ years, it is $1 - e^{-N_D n}$, which for small $N_D$ is very nearly $N_D n$.

Where $N_G$ comes from. Ground flash density is measured by national lightning detection networks — arrays of sensors that locate flashes by time of arrival and direction finding — and published as maps by meteorological services and standards bodies. It varies by more than an order of magnitude within a single country: from a fraction of a flash per square kilometre per year on a cold coast to well over ten in a continental interior with strong summer convection. It is not a number to estimate. The map's own uncertainty is not negligible either, since detection efficiency, the size of the grid cell, and the number of years averaged all move it, and older sources sometimes quote thunderstorm-days instead, which is a different quantity requiring conversion.

$C_D$ is a four-value table, and it is a big lever. IEC 62305-2 gives 0.25 for a structure surrounded by taller objects or trees, 0.5 for one among objects of the same height, 1 for an isolated structure with nothing within three times its height, and 2 for an isolated structure on a hilltop or promontory. That is a factor of eight from one end to the other — more than most of the geometry — and a building sitting on the boundary between two cases is genuinely uncertain by a factor of two. It is a table, not a continuum: a computed $C_D$ of 0.7 is not a location factor you may use.

And this is only the first line of the risk assessment. IEC 62305-2 goes on to count strikes near the structure, strikes to the connected services — power, telecoms, water — and strikes near those services, because a surge arriving down a buried cable does not care that the building was never hit. Each of those is weighted by probabilities of damage and by loss factors covering loss of human life, loss of service to the public, loss of cultural heritage and economic loss, and the totals are compared against tolerable risk. A structure with a low strike count can still need protection because of what is inside it; a structure with a high count can need none. A strike frequency on its own decides nothing, and quoting it as though it did is the most common way this arithmetic gets misused.