Lidar-Derived Forest Metrics Predict Snow Accumulation in the Central Sierra Nevada, USA

Abstract

Snowmelt from montane forests is a critical water resource in the western United States. Forest managers use treatments like selective thinning to encourage resilient ecosystems for wildfire mitigation and wildlife habitat. There is also interest in managing forests to optimize snowpack retention to improve water resources in a changing climate, but detailed studies and management recommendations are limited. We explore the controls on snowpack accumulation using a newly developed light detection and ranging (lidar) point-cloud filtering method and a local open-reference area approach using data collected over gradients in forest structure across multiple snow seasons in the Sagehen Creek Basin (SCB) in the central Sierra Nevada, California, USA. Consistent with previous studies with much more limited snow and vegetation measurements, we show there is ~25% more snow accumulation in open areas relative to forested areas. Random forest (RF) outputs indicate that forest structure metrics explain a greater amount of accumulation variance than terrain metrics, and the greatest potential to increase snow accumulation via thinning occurs when the fraction of vegetation (fVEG) is > 30%. Our results suggest that considering both coarse (e.g., fVEG) and fine-scale (e.g., the arrangement of canopy) canopy information is integral to predict snowpack response to canopy disturbance at many relevant management scales (i.e., 100 m² to 100 km²). The corresponding simple decision support tool, developed with data from SCB, can assess the utility of completed and planned forest restoration over a larger spatial extent to strategically target areas with the highest potential snowpack response. Our new lidar processing methods are easily transferrable to other areas where they could improve snowpack management from forest restoration.