Wednesday, 6 August 2003: 2:10 PM
Optimizing matching criteria for start time analysis of real-time “spot” forecasts and climatology for long range oil spill trajectory research
An initial study of long range oil spill modeling involved random sampling of historic wind records to determine probable oil trajectories. An approach is presented to isolate those portions of climatology that pertain to the wind events during a given spill incident. Previous research on extended trajectory analysis of oil spills focused on selective matching of "spot" regional forecasts to seasonal climatology. The matching start times are used in sampling historical wind data to generate representative patterns for a long range oil spill model. This paper extends that work, analyzing a set of time blocks following the data matching start times. These time blocks are several days in duration and are correlated (using wind speed and direction) with themselves and with random chosen time blocks in climatology. The results of this correlation analysis determines the validity of the initial matching criteria which can then be optimized. Ideally, the matching criteria are optimized to obtain an array of start times that better sample climatology, resulting in an optimal long range oil spill trajectory. Various matching criteria are analyzed and several different topographical regions are reviewed.