New indices for the spatial validation of plume forecasts with observations of smoke plumes from grassfires

Abstract

The purpose of this work is to propose new indices for the spatial validation of hazardous plumes forecast, and apply and test them with data of a case study. One, the Plume-Overlap-Area Hit index, is a modification of a widely used index that considers the overlap area between observed and forecast plumes. The other one, the Plume-Mean-Orientation Hit index, introduces a new concept in plume forecast validation, i.e., the mean direction of plume propagation. These two indices are combined in a new two-dimensional Combined-Direction-Area Hit index. The new indices are applied to the spatial validation of smoke plume forecast for a case study of uncontrolled grassfires that took place during April and May 2008 in the La Plata River region in South America. Operational models at the Argentine National Meteorological Service (SMN) are employed to produce the plume forecast. The HIRHYLTAD dispersion model is used to forecast the smoke plumes, employing the Eta/SMN meteorological forecast model outputs. The forecast plumes are compared to the observed plumes in high-resolution MODIS imagery from AQUA and TERRA satellites, from which a total of 59 smoke plumes are identified. The study concludes that the presented methodology that employs operational meteorological models and simplified dispersion models can be used to produce reasonably accurate forecasts of the areas affected by the smoke plumes that originate in forest and grassland fires, particularly in cases when limited information is available about the fires. Although the present study is specifically applied to smoke plumes, the validation technique with the proposed indices can be of utility to study pollutant plumes of diverse nature.