USGS Publications Warehouse

Abstract

Rain-on-snow events are common on mountain slopes within the transient-snow zone of the Pacific Northwest. These events make more water available for runoff than does precipitation alone by melting the snowpack and by adding a small amount of condensate to the snowpack. In forest openings (such as those resulting from clearcut logging), the amount of snow that accumulates and the turbulent- energy input to the snowpack are greater than below forest stands. Both factors are believed to contribute to a greater amount of water available for runoff during rain-on-snow events in forest openings than forest stands. Because increased water available for runoff may lead to increased downstream flooding and erosion, knowledge of the amount of snowmelt that can occur during rain on snow and the processes that control snowmelt in forest openings is useful when making land-use decisions. Snow accumulation and melt were simulated for clearcut conditions only, using an enery- balance approach that accounts for the most important energy and mass exchanges between a snowpack and its environment. Meteorological measurements provided the input for the simulations. Snow accumulation and melt were not simulated in forest stands because interception of precipitation processes are too complex to simulate with a numerical model without making simplifying assumptions. Such a model, however, would need to be extensively tested against representative observations, which were not available for this study. Snowmelt simulated during three rain-on-snow events (measured in a previous study in a clearcut in the transient-snow zone of the H.J. Andrews Experimental Forest in Oregon) demonstrated that melt generation is most sensitive to turbulent- energy exchanges between the air and the snowpack surface. As a result, the most important climate variable that controls snowmelt is wind speed. Air temperature, however, is a significant variable also. The wind speeds were light, with a maximum of 3.3 meters per second during one event and average wind speeds for all three events ranging from 1.7 to 2.1 meters per second. For observed and estimated conditions, the average simulated snowmelt ranged from 0.2 to 0.8 millimeter liquid water per hour, and turbulent-energy exchange provided 51 percent of the energy that led to snowmelt during the largest of the three rain-on-snow events. When wind speeds were multiplied by a factor of 4, the simulated snowmelt ranged from 1.0 to 2.5 millimeters per hour. Similarly, when wind speeds were multiplied by a factor of 6, the simulated snowmelt ranged from 1.6 to 3.7 millimeters per hour. Turbulent-energy exchange provided a dominant 88 and 92 percent of the energy input to the snowpack during the largest rain-on-snow event when average wind speeds were multiplied by factors of 4 and 6, respectively. During the same event, the contribution to melt by the sum of net solar and net thermal radiation (net all-wave radiation) was roughly equal to the contribution of sensible energy carried by the precipitation itself (advective heat). Estimates of snowmelt resulting from rain on snow for climate conditions other than those observed and estimated in the simulated plot-scale data were expanded by simulating snowmelt for 24-hour presumed rain-on-snow events extracted from the reconstructed, long-term historical climate records for Cedar Lake and Snoqualmie Pass National Weather Service stations in Washington State. The selected events exceeded 75 millimeters of precipitation in 24 hours. When clearcut conditions were assumed to be identical to those at the H.J. Andrews Experimental Forest site and a ripe snowpack that never completely melted was assumed to be available, simulated 24-hour snowmelt ranged from 4.2 to 47.0 millimeters (0.2 to 2.0 millimeters per hour) for low wind speeds (1.5 meters per second) and from 10.3 to 178.8 millimeters (0.4 to 7.5 millimeters per hour) for high wind speeds (8.2 meters per second). The ranges in melt for a given wind speed resulted from the different combinations of air temperature, dewpoint temperature, and precipitation depth that were characteristic of the synthetic events. The average of the median 24-hour snowmelt at Cedar Lake and Snoqualmie Pass was 15.1 millimeters (0.6 millimeters per hour) at low wind speeds and 49.6 millimeters (2.1 millimeters per hour) at high wind speeds. Condensation could increase water available for runoff by a small percentage of the melt. The climate conditions used to generate the range in melt estimates are representative of the transient-snow zone of the western Cascade Range of the Pacific Northwest because Cedar Lake and Snoqualmie Pass are located near the bottom and top of the zone, respectively. Hourly plot-scale data available from previous studies for clearcut, forested, and plantation conditions in the western Pacific Northwest could not be used to simulate snow accumulation and melt over extended periods of time to investigate the effects of different climate and physical conditions. Measurements of snowpack properties were too infrequent; precipitation-density information was absent; and water-available-for-runoff measurements on vegetated plots were not considered representative of larger areas because lysimeters were too small to account for the lateral variability of snow accumulation and melt due to interception processes in the canopy. Lack of representative data for vegetated land precluded the testing of a numerical model that would simulate precipitation-interception processes in the forest canopy. Even for the plot-scale simulations that were done, basic data had to be estimated, and as a result, the three plot-scale rain-on-snow events, as well as the 24-hour events, could be considered synthetic. To ensure adequate data sets for future studies of climate and physical factors in snowmelt generation during rain on snow, data-collection efforts would include frequent (at least every few days) visits to obtain measurements of snowpack thickness, density, liquid-water content, and temperature and to verify that climate data suitable for use in energy-balance numerical models are being collected. In addition to climate variables such as average hourly wind speed, incoming solar radiation, air temperature, and dewpoint temperature, variables such as incoming thermal radiation, reflected solar radiation, and precipitation density would be measured. Soil temperature would be measured, except at study sites at altitudes where snowpacks remain close to isothermal at 11 degrees Celsius, where those measurements could be optional. Studies of melt generation during rain on snow on forested land could be designed to account for the lateral variability of snow accumulation and melt that occurs below the vegetative canopy. Plot-scale studies that use small lysimeters to measure water available for runoff are not appropriate for the study of rain on snow in forested settings; instead, a combination of data collection at both the plot and catchment scale could be used. At the plot scale, water available for runoff would need to be measured in a few extremely large lysimeters, or many small ones. At the catchment scale, water available for runoff would have to be computed from streamflow measurements by correcting it for such variables as baseflow, interflow, soil-moisture storage, evapotranspiration, and bank storage. Plot- and catchment-scale data could be analyzed simultaneously, because a nested, duplicate approach is more likely to produce useful results for simulating water available for runoff during rain on snow in forest stands than analysis of either data type alone.