Augmenting flood frequency estimates using flood-scarred trees.

Abstract

Flood-damaged trees along streams and rivers in 20 localities in Arizona, Colorado, New Mexico, and Utah were examined to assess the feasibility of reconstructing the frequency and magnitude of floods over the last several hundred years. Tree-ring dating of the flood damage produced evidence for 17 floods during the last 125 years, and for at least four floods prior to 1866. Most of the flood-scar dates from the historic period were found to coincide with the dates of major floods on the waterways from which they were collected, or from nearby streams. Flood damage to trees was found to be very abundant and accessible, with the most productive situation probably being that of seasonal streambeds in narrow deep canyons with relatively steep gradients. Amount of effort required to produce a flood chronology appears to be relatively small. Collections at most sites involved sampling from 1 to 5 trees, by extracting 3 to 6 increment cores from each visibly scarred tree, and 2 cores from undamaged trees to facilitate crossdating. At two sites larger collections were made, including 3 cross sections (V-cuts), which were very helpful in establishing dates of scars. For a given channel gradient, scars seem to develop only when the flood depth is above a particular threshold level. This threshold was defined by plotting scar height for scarred trees and flow depth for other floods not producing scars against the logarithm of channel slope. The scarred and unscarred individual points were separated into two well-defined fields, separated by a fairly straight line. Discriminant function analysis showed the separation to be highly significant, and classification of individual points as scarred or unscarred was usually correct. The implication of the scar threshold is that even in drainages with no gage or historical record at all, the presence or absence of flood scars on channelside trees would indicate the occurrence or nonoccurrence of floods of a certain depth over the lifetime of the trees. Basin analysis and field measurements of appropriate channel characteristics would allow this depth and the corresponding discharge to be calculated.