Lidar-based models of precipitation interception in the boreal forest of the Eklutna Valley, Alaska

Abstract:

Precipitation interception by forest canopies can significantly influence regional hydrological processes, however, accurate measurement of interception over large areas and across heterogeneous forests can be difficult. Lidar has emerged as an effective tool to measure forest structure, and may be used to model large-scale forest processes such as precipitation interception. We developed a non-linear regression model from first principals to predict the percent of precipitation interception from forest canopies using lidar. To find the best parameters for the model, we measured throughfall of rain (n = 21), fresh snow (n = 21), and old snow (n = 26) on plots in the boreal forest of the upper Eklutna Valley, Alaska. The valley contains heterogeneous forest structure, with open and closed deciduous, needleleaf, and tall shrub stands. For each plot we calculated a set of twelve lidar metrics, and found the combined metric of mean height * cover to be the lidar metric most highly correlated to ln(throughfall) for rain (r = -0.81), fresh snow (r = -0.79), and old snow (r = -0.73). Using mean height * cover in the precipitation interception model we created a 15 m resolution spatial model of interception across the vegetated portions of the upper Eklutna Valley. Using this model, we calculated the mean interception for rainfall (20% ± 3%), fresh snow (29% ± 4%), and old snow (20% ± 3%) across the modeled region. These results suggest lidar can be an effective tool to predict both rain and snow interception across structurally varied forests.