Evaluating observational methods to quantify snow duration under diverse forest canopies

Abstract

AbstractForests cover almost 40% of the seasonally snow‐covered regions in North America. However, operational snow networks are located primarily in forest clearings, and optical remote sensing cannot see through tree canopies to detect forest snowpack. Due to the complex influence of the forest on snowpack duration, ground observations in forests are essential. We therefore consider the effectiveness of different strategies to observe snow‐covered area under forests. At our study location in the Pacific Northwest, we simultaneously deployed fiber‐optic cable, stand‐alone ground temperature sensors, and time‐lapse digital cameras in three diverse forest treatments: control second‐growth forest, thinned forest, and forest gaps (one tree height in diameter). We derived fractional snow‐covered area and snow duration metrics from the colocated instruments to assess optimal spatial resolution and sampling configuration, and snow duration differences between forest treatments. The fiber‐optic cable and the cameras indicated that mean snow duration was 8 days longer in the gap plots than in the control plots (p < 0.001). We conducted Monte Carlo experiments for observing mean snow duration in a 40 m forest plot, and found the 95% confidence interval was ±5 days for 10 m spacing between instruments and ±3 days for 6 m spacing. We further tested the representativeness of sampling one plot per treatment group by observing snow duration across replicated forest plots at the same elevation, and at a set of forest plots 250 m higher. Relative relationships between snow duration in the forest treatments are consistent between replicated plots, elevation, and two winters of data.