NWC SEMINAR SERIES

A Climatology and Experimental Automated Probabilistic Lightning Forecasts for Alaska

David Buckley
School of Meteorology
The University of Oklahoma, Norman, OK

08 May 2009, 2:00 PM
National Weather Center, Room 5600
120 David L. Boren Blvd.
University of Oklahoma
Norman, OK
Directions to the NWC (.pdf, 60 kb)

Though located much farther north than the lower 48 states, Alaska still experiences a lightning season running from mid-May through August, with the vast majority of lightning activity taking place between during June and July. Though largely unpopulated, lightning is a concern as CG lightning strikes have the potential to start forest fires which can grow large and threaten communities and the ecosystem of a sizeable portion of the state. To combat this problem, a climatology of lightning activity was derived from the Bureau of Land Management Lightning Detection Network data during the years 2000-2007 to identify the extent and traditional areas and times of lightning activity.

To fully tackle the forecast problem, an objective forecast system was needed. To this end, a statistical model was derived using the Perfect Prognosis approach. The training data for the model was June-August for the years 2000-2007 using the data from the North American Regional Reanalysis and the lightning climatology. Using Principle Component Analysis (PCA), logistic regression model equations yielding the probability of observing one or more and ten or more CG flashes in a 45km x 45km grid box were derived. Using input parameters from the GFS model, the forecasts were produced four times a day during the summer of 2008 with output at 3 hour intervals out to 180 hours. Major lightning/wildfire events that occurred during the testing period will be presented as well as results of the verification done on the 2008 forecasts. For 2009, the model equations will be re-derived on a 10km grid.


Back to NWC Seminar Series and Colloquia