²ÊÃñ±¦µä

Moody , John A ¹

¹ U.S. Geological Survey

Extreme floods often follow wildfire in mountainous watersheds. The peak discharge for these post-wildfire floods has been shown to correlate with rainfall intensity. The relation between rainfall intensity and peak discharge per unit basin area is based on data from several areas burned by wildfire. For some burned basins on the order of 0.1-10 km2, peak discharge is a linear function of the maximum 30-minute rainfall intensity, I30. This relation also has a rainfall intensity threshold, below which no runoff was generated, and this suggests infiltration-excess overland flow as a dominant process after wildfire. A linear relation permits the definition of a post-wildfire runoff coefficient, C, defined as the peak discharge per unit area per unit rainfall intensity.

Magnitudes of post-wildfire floods also depend on the burn severities. An empirical relation between burn severity and the post-wildfire runoff coefficient was developed for data collected from basins burned by the 2000 Cerro Grande Fire near Los Alamos, New Mexico. Soil burn severity was quantified by the difference (pre-wildfire minus post-wildfire) of the normalized burn ratio, ?NBR. The normalized burn ratio is the relative difference derived from remote-sensing measurements of two spectral bands sensitive to vegetation (infrared) and bare soil (short-wave infrared). The post-wildfire runoff coefficient was found to be a linear function of the basin average ?NBR (R2=0.45). The basin average ?NBR does not include any information about the connectivity of burn severity patches. So, a new burn severity variable, the average hydraulic functional connectivity, which includes information about the spatial sequence of the soil burn severity along hillslope flow paths, was defined: this variable provided an improved relation with the post-wildfire runoff coefficient (R2=0.65).

To use these relations, it is important to have some idea of the spatial and temporal character of the rainfall intensities. Analysis of about 20 years of pre-wildfire Doppler radar rainfall over the area burned by the 2002 Hayman Fire showed that the areal extent above a given I30 decreased as the magnitude of I30 increased. For example, I30 = 5 mm h-1 intensity might cover a total area of about 650 km2, whereas an I30 = 50 mm h-1 might cover only a total area of 240 km2. These total areas, within an individual rainstorm, are composed of several separate cells, which may or may not correspond to the size of burned basins within a burned area.

An interesting effort would be to apply this method within the area burned by 2010 Fourmile Canyon Fire. Some data on burn severity already have been collected, but they need to be processed. And, it would be good to acquire more data on rainfall intensity characteristics to compare with those from the Hayman study, especially the size and number of rain cells of a given intensity within groups of spring frontal and summer convective rainstorms. To verify and at the same time improve the method, I am planning to deploy a rain gage network in collaboration with NOAA and NCAR and a network of flumes to measure peak discharge. Please let me know if you are interested in any aspects of this effort.