Abstract

A methodology for assimilating ground-based and remotely sensed snow data within a snow-evolution modeling system (SnowModel) is presented. The data assimilation scheme (SnowAssim) is consistent with optimal interpolation approaches in which the differences between the observed and modeled snow values are used to constrain modeled outputs. The calculated corrections are applied retroactively to create improved fields prior to the assimilated observations. Thus, one of the values of this scheme is the improved simulation of snow-related distributions throughout the entire snow season, even when observations are only available late in the accumulation and/or ablation periods. Because of this, the technique is particularly applicable to reanalysis applications. The methodology includes the ability to stratify the assimilation into regions where either the observations and/or model has unique error properties, such as the differences between forested and nonforested snow environments. The methodologies are introduced using synthetic data and a simple simulation domain. In addition, the model is applied over NASA’s Cold Land Processes Experiment (CLPX), Rabbit Ears Pass, Colorado, observation domain. Simulations using the data assimilation scheme were found to improve the modeled snow water equivalent (SWE) distributions, and simulated SWE displayed considerably more realistic spatial heterogeneity than that provided by the observations alone.

Abstract

SnowModel is a spatially distributed snow-evolution modeling system designed for application in landscapes, climates, and conditions where snow occurs. It is an aggregation of four submodels: MicroMet defines meteorological forcing conditions, EnBal calculates surface energy exchanges, SnowPack simulates snow depth and water-equivalent evolution, and SnowTran-3D accounts for snow redistribution by wind. Since each of these submodels was originally developed and tested for nonforested conditions, details describing modifications made to the submodels for forested areas are provided. SnowModel was created to run on grid increments of 1 to 200 m and temporal increments of 10 min to 1 day. It can also be applied using much larger grid increments, if the inherent loss in high-resolution (subgrid) information is acceptable. Simulated processes include snow accumulation; blowing-snow redistribution and sublimation; forest canopy interception, unloading, and sublimation; snow-density evolution; and snowpack melt. Conceptually, SnowModel includes the first-order physics required to simulate snow evolution within each of the global snow classes (i.e., ice, tundra, taiga, alpine/mountain, prairie, maritime, and ephemeral). The required model inputs are 1) temporally varying fields of precipitation, wind speed and direction, air temperature, and relative humidity obtained from meteorological stations and/or an atmospheric model located within or near the simulation domain; and 2) spatially distributed fields of topography and vegetation type. SnowModel’s ability to simulate seasonal snow evolution was compared against observations in both forested and nonforested landscapes. The model closely reproduced observed snow-water-equivalent distribution, time evolution, and interannual variability patterns.

Abstract

Twenty-five years ago, we published a global seasonal snow classification now widely used in snow research, physical geography, and as a mission planning tool for remote sensing snow studies. Performing the classification requires global datasets of air temperature, precipitation, and land cover. When introduced in 1995, the finest-resolution global datasets of these variables were on a 0.5° × 0.5° latitude–longitude grid (approximately 50 km). Here we revisit the snow classification system and, using new datasets and methods, present a revised classification on a 10-arc-s × 10-arc-s latitude–longitude grid (approximately 300 m). We downscaled 0.1° × 0.1° latitude–longitude (approximately 10 km) gridded meteorological climatologies [1981–2019, European Centre for Medium-Range Weather Forecasts (ECMWF) Reanalysis, 5th Generation Land (ERA5-Land)] using MicroMet, a spatially distributed, high-resolution, micrometeorological model. The resulting air temperature and precipitation datasets were combined with European Space Agency (ESA) Climate Change Initiative (CCI) GlobCover land-cover data (as a surrogate for wind speed) to produce the updated classification, which we have applied to all of Earth’s terrestrial areas. We describe this new, high-resolution snow classification dataset, highlight the improvements added to the classification system since its inception, and discuss the utility of the climatological snow classes at this much higher resolution. The snow class dataset (Global Seasonal-Snow Classification, Version 1) and the tools used to develop the data are publicly available online at the National Snow and Ice Data Center (NSIDC).

Abstract

A blowing-snow model (SnowTran-3D) was combined with field measurements of end-of-winter snow depth and density to simulate solid (winter) precipitation, snow transport, and sublimation distributions over a 20 000-km² arctic Alaska domain. The domain included rolling uplands and a flat coastal plain. Simulations were produced for the winters of 1994/95, 1995/96, and 1996/97. The model, which accounts for spatial and temporal variations in blowing-snow sublimation, as well as saltation and turbulent-suspended transport, was driven with interpolated fields of observed temperature, humidity, and wind speed and direction. Model outputs include local (a few hundreds of meters) to regional (several tens of kilometers) distributions of winter snow-water-equivalent depths and blowing-snow sublimation losses, from which the regional winter precipitation distributions are computed. At regional scales, the end-of-winter snow depth is largely equal to the difference between winter precipitation and moisture loss due to sublimation. While letting SnowTran-3D simulate the blowing-snow sublimation fluxes, the precipitation fields were determined by forcing the regional variation in model-simulated snow depths to match measured values. Averaged over the entire domain and the three simulation years, the winter precipitation was 17.6 cm, with uplands values averaging 19.0 cm and coastal values averaging 15.3 cm. On average, 21% of the precipitation was returned to the atmosphere by blowing-snow sublimation, while in the windier coastal regions 34% of the winter precipitation sublimated.

Abstract

An intermediate-complexity, quasi–physically based, meteorological model (MicroMet) has been developed to produce high-resolution (e.g., 30-m to 1-km horizontal grid increment) atmospheric forcings required to run spatially distributed terrestrial models over a wide variety of landscapes. The following eight variables, required to run most terrestrial models, are distributed: air temperature, relative humidity, wind speed, wind direction, incoming solar radiation, incoming longwave radiation, surface pressure, and precipitation. To produce these distributions, MicroMet assumes that at least one value of each of the following meteorological variables are available for each time step, somewhere within, or near, the simulation domain: air temperature, relative humidity, wind speed, wind direction, and precipitation. These variables are collected at most meteorological stations. For the incoming solar and longwave radiation, and surface pressure, either MicroMet can use its submodels to generate these fields, or it can create the distributions from observations as part of a data assimilation procedure. MicroMet includes a preprocessor component that analyzes meteorological data, then identifies and corrects potential deficiencies. Since providing temporally and spatially continuous atmospheric forcing data for terrestrial models is a core objective of MicroMet, the preprocessor also fills in any missing data segments with realistic values. Data filling is achieved by employing a variety of procedures, including an autoregressive integrated moving average calculation for diurnally varying variables (e.g., air temperature). To create the distributed atmospheric fields, spatial interpolations are performed using the Barnes objective analysis scheme, and subsequent corrections are made to the interpolated fields using known temperature–elevation, wind–topography, humidity–cloudiness, and radiation–cloud–topography relationships.

Abstract

Air and snow–ground interface temperatures were measured during two winters at 33 stations spanning the 180-km-long Kuparuk basin in arctic Alaska. Interface temperatures averaged 7.5°C higher than air temperatures and varied in a manner that was more complex, and on a spatial scale more than 100 times smaller, than the air temperature. Within the basin, two distinct thermal regimes could be identified, with the division at the boundary between coastal and uplands provinces. When each station was classified into one of three snow exposure classes (exposed, intermediate, or sheltered), accounting for variations in snow depth and thermal properties, 87% of the variation in the average interface temperature could be predicted from air temperature. Individual station interface temperature records were fit using a beta curve that captured the slow decrease in autumn and the rapid rise in spring. Beta curves were specified by three parameters (α, β, and γ) that could be predicted if province and snow exposure class were known. A model based on the beta curves was developed and applied over the basin to predict interface temperatures in both time and space. Tests of the model against data from within the domain and from arctic Alaskan locations outside the domain suggest an accuracy of ±2°C when simulating average winter interface temperatures, and ±3°C when simulating daily interface temperatures.

Abstract

Snow interception in a coniferous forest canopy is an important hydrological feature, producing complex mass and energy exchanges with the surrounding atmosphere and the snowpack below. Subcanopy snowpack accumulation and ablation depends on the effects of canopy architecture on meteorological conditions and on interception storage by stems, branches, and needles. Mountain forests are primarily composed of evergreen conifer species that retain their needles throughout the year and hence intercept snow efficiently during winter. Canopy-intercepted snow can melt, fall to the ground, and/or sublimate into the air masses above and within the canopy. To improve the understanding of snow–canopy interception processes and the associated influences on the snowpack below, a series of model experiments using a detailed, physically based snow–canopy and snowpack evolution model [Alpine Multiscale Numerical Distributed Simulation Engine (AMUNDSEN)] driven with observed meteorological forcing was conducted. A cone-shaped idealized mountain covered with a geometrically regular pattern of coniferous forest stands and clearings was constructed. The model was applied for three winter seasons with different snowfall intensities and distributions. Results show the effects of snow–canopy processes and interactions on the pattern of ground snow cover, its duration, and the amount of meltwater release, in addition to showing under what conditions the protective effect of a forest canopy overbalances the reduced accumulation of snow on the ground. The simulations show considerable amounts of canopy-intercepted snowfall can sublimate, leading to reduced snow accumulation beneath the forest canopy. In addition, the canopy produces a shadowing effect beneath the trees that leads to reduced radiative energy reaching the ground, reduced below-canopy snowmelt rates, and increased snow-cover duration relative to nonforested areas. During snow-rich winters, the shadowing effect of the canopy dominates and snow lasts longer inside the forest than in the open, but during winters with little snow, snow sublimation losses dominate and snow lasts longer in the open areas than inside the forest. Because of the strong solar radiation influence on snowmelt rates, the details of these relationships vary for northern and southern radiation exposures and time of year. In early and high winter, the radiation protection effect of shadowing by the canopy is small. If little snow is available, an intermittent melt out of the snow cover inside the forest can occur. In late winter and spring, the shadowing effect becomes more efficient and snowmelt is delayed relative to nonforested areas.

Abstract

The presence of snow and its relationship to surrounding vegetation significantly impacts the surface energy balance. For accurate atmospheric model simulations, the degree to which a snowpack can cover vegetation must be realistically represented. Both vegetation height and snow depth must be reasonably known to determine the amount of masking.

The Regional Atmospheric Modeling System/Land Ecosystem–Atmosphere Feedback, version two (RAMS/ LEAF-2) snow model was modified to simulate snow depth in addition to snow water equivalent and was driven offline with observed atmospheric forcing data. The model was run for five of the Boreal Ecosystem–Atmosphere Study (BOREAS) surface mesonet stations over the 1995/96 winter. The time evolution of simulated snow depth was compared with the observed snow depth. Averaged over the winter, the modeled snow depth at the four low-wind stations was within 0.09 m of the observations, and the average percent error was 27%, while the one wind-blown station was considerably worse. The average depth error at all five stations was ±0.08 m. This is shown to be sufficient to reasonably account for the surface energy balance effects of vegetation protruding through the snow.

Abstract

A field measurement program was undertaken as part NASA’s Cold Land Processes Experiment (CLPX). Extensive snowpack and soil measurements were taken at field sites in Colorado over four study periods during the two study years (2002 and 2003). Measurements included snow depth, density, temperature, grain type and size, surface wetness, surface roughness, and canopy cover. Soil moisture measurements were made in the near-surface layer in snow pits. Measurements were taken in the Fraser valley, North Park, and Rabbit Ears Pass areas of Colorado. Sites were chosen to gain a wide representation of snowpack types and physiographies typical of seasonally snow-covered regions of the world. The data have been collected with rigorous protocol to ensure consistency and quality, and they have undergone several levels of quality assurance to produce a high-quality spatial dataset for continued cold lands hydrological research. The dataset is archived at the National Snow and Ice Data Center (NSIDC) in Boulder, Colorado.

Abstract

The lateral and vertical variability of snow stratigraphy was investigated through the comparison of the measured profiles of snow density, temperature, and grain size obtained during the Snow Science Traverse—Alaska Region (SnowSTAR2002) 1200-km transect from Nome to Barrow with model reconstructions from the Snow Thermal Model (SNTHERM), a multilayered energy and mass balance snow model. Model profiles were simulated at the SnowSTAR2002 observation sites using the 40-yr European Centre for Medium-Range Weather Forecasts Re-Analysis (ERA-40) as meteorological forcing. ERA-40 precipitation was rescaled so that the total snow water equivalent (SWE) on the SnowSTAR2002 observation dates equaled the observed values. The mean absolute error (MAE) of measured and simulated snow properties shows that SNTHERM was able to produce good simulations for snowpack temperature but larger errors for grain size and density. A spatial similarity analysis using semivariograms of measured profiles shows that there is diverse lateral and vertical variability for snow properties along the SnowSTAR2002 transect resulting from differences in initial snow deposition, influenced by wind, vegetation, topography, and postdepositional mechanical and thermal metamorphism. The correlation length in snow density (42 km) is quite low, whereas it is slightly longer for snow grain size (125 km) and longer still for snow temperature (130 km). An important practical question that the observed and reconstructed profiles allow to be addressed is the implications of model errors in the observed snow properties for simulated microwave emissions signatures. The Microwave Emission Model for Layered Snowpacks (MEMLS) was used to simulate 19- and 37-GHz brightness temperatures. Comparison of SNTHERM–MEMLS and SnowSTAR2002–MEMLS brightness temperatures showed a very good match occurs at 19 GHz [a root-mean-square error (RMSE) of 1.5 K (8.7 K) for vertical (horizontal) polarization] and somewhat larger [5.9 K (6.2 K) for vertical (horizontal) polarization] at 37 GHz. These results imply that the simulation of snow microphysical profiles is a viable strategy for passive microwave satellite–based retrievals of SWE.