Thursday, 17 January 2002: 11:15 AM
Probabilistic early warning of cloud-to-ground lightning at an airport
The present study is directed toward quantifying lightning warning at airports and other facilities where there is considerable outdoor activity with risks to people and property. In the past, lightning decision-making has been guided by the general probabilistic assertions that lightning within the area is "possible" or "likely" within the next n minutes, but the conditions established for these levels were determined subjectively rather than with observable quantities. We begin to address this problem using lightning location data alone within an area of interest that encompasses the facility of interest and in larger regions in the vicinity of the facility (warning regions). In the first study of this type, the warning regions were concentric about the facility of interest. In an attempt to reduce false alarm rates, we now design the warning regions by examining regional flash density maps during appropriate intervals prior to the lightning onsets within the area of interest in a large number of storms. Despite this, we find that false alarm rates are sensitive almost exclusively to the sizes of the warning regions and not their positions unless storm motion is explicitly taken into account. Even without the explicit handling of storm motion, a suitable choice of two warning regions produces an outermost region that indicates lightning "possibility" and an intermediate region that indicates lightning "likelihood". We present the results of such an analysis for the Tucson International Airport (TIA) in Arizona. To account explicitly for storm motion, we subdivide the warning regions and produce cumulative probabilities and false alarm rates for each subdivision. Preliminary results from this type of analysis are also presented for the TIA.