Airborne volcanic ash forecast area reliability

 

Stunder, B.J.B., J.L. Heffter,and Roland R. Draxler

 

Weather and Forecasting, 2007,Volume 22, pages 1132-1139.

 

Abstract - In support of aircraft flight safety operations, daily comparisons between modeled, hypothetical, volcanic ash plumes calculated with meteorological forecasts and analyses were made over a one-and-half year period. The Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model simulated the ash transport and dispersion. Ash forecasts and analyses from seven volcanoes were studied. The volcanoes were chosen because of recent eruptions or because their airborne ash could impinge on well-traveled commercial aircraft flight paths. For each forecast-analysis pair, a statistic representing the degree of overlap, the threat score (TS), was calculated. A forecast was classified as acceptable if the TS was greater than 0.25. Each forecast was also categorized by two parameters: the forecast area quadrant with respect to the volcano and a factor related to the complexity of the meteorology. The forecast complexity factor was based on the degree of spread using NCEP ensemble output or using a HYSPLIT offset configuration. In general, the larger the spread of the ensemble or offset forecasts, the greater the complexity. The forecasts were sorted by complexity factor, and then classified by the quartile of the complexity. Volcanic ash forecast area reliability (VAFAR) was calculated for each forecast area quadrant and for each quartile of the complexity factor. VAFAR is the ratio of the number of acceptable forecasts to the total number of forecasts. Most VAFAR values were above 70%. VAFAR values for two of the seven volcanoes (Popocatepetl and Tungurahua) tended to be lower than the others. In general, VAFAR decreased with increasing complexity of the meteorology. It should be noted that the VAFAR values reflect the reliability of the meteorological forecasts when compared to the same calculation using analysis data; the dispersion model itself was not evaluated.

 

 

Email me the full report