The impact of Santa Ana winds on wildland fire regimes

Document Type

Conference Paper/Presentation

Publication Date



Santa Ana winds have been implicated as major drivers of extensive wildfires that occur annually in southern California. While numerous anecdotal reports dictate an obvious relationship, there is little quantitative analysis in current literature on how this loosely-defined weather phenomenon impacts fire progression regimes. A new satellite-derived fire progression dataset developed using MODIS and Landsat was linked to a network of 82 Remote Automated Weather Stations (RAWS) to evaluate three differing a priori classifications of Santa Ana events defined using three weather parameters: wind speed, wind direction, and relative humidity. The fire progression dataset comprised 528 burn area polygons representing 151 distinct fire events from 2001-2009 in southern California. Each burn area polygon was assigned weather data from one of the RAWS units by shortest straight-line distance. These data and methods show quantitatively that burn area is dramatically larger under Santa Ana conditions than under non-Santa Ana conditions (see table). Outliers of both types (large burn areas under non-SA conditions; small burn areas under SA conditions) were identified and closely examined to identify potentially confounding variables. Time-lag effects (particularly with respect to cumulative precipitation preceding day of burn) and effects with regard to local vs. regional measurements were examined as well. We tested 17 variables (3 relative humidity variables, 5 wind speed variables, 3 temperature variables, 3 moisture variables, previous day burn area, census-derived population density, and the number of hours meeting one Santa Ana classification) individually and in combination for correlation with both burn area and burn area change from the previous day to test their predictive power. Mean daily relative humidity was found to have the strongest correlation (Pearson's r = -0.451) with burn area. All variables except moisture variables were more strongly correlated with the change in burn area from the previous day than with day-of burn area.

Publisher's Statement

© 2011 American Geophysical Union.

Publication Title

AGU Fall Meeting 2011