Past and Projected Changes in Snowpack and Soil Frost at the Hubbard Brook Experimental Forest, New Hampshire
Long-term empirical data from the Hubbard Brook Experimental Forest in New Hampshire show that air temperature has increased significantly by an average of 0.023 degrees C per year over the last 50 years of measurement. The warmer climate has caused significant declines in snow depth, snow water equivalent, and snow cover duration. Paradoxically, it has been suggested that warmer air temperatures may result in colder soil temperatures (and more soil frost) since there will be less snow cover insulating soils during winter. This prediction is supported by snow depth manipulation experiments, which have shown that soil frost depth increases when snow is removed. However it is unclear how well these experiments represent conditions associated with actual climate change. A thorough understanding of the influence of climate on soil frost is critical because it can have a profound effect on many hydrological, chemical and biological processes. Hubbard Brook has one of the longest records of soil frost field measurements in the northeastern US (45 years); however, high interannual variability and the infrequency of major soil frost events limit the ability to detect long-term trends. As an alternative to field measurements, soil frost can be modeled reliably using knowledge of the physics of energy and water transfer. In this application, we used the Simultaneous Heat and Water Model (SHAW) driven by statistically downscaled climate data from two Atmosphere-Ocean General Circulation Models (HadCM3, PCM). The SHAW model was run through the end of this century under two climate scenarios. Results indicate significant decreases in modeled soil frost due to a combination of warmer fall air temperatures and minimal declines in snowpack depth caused by greater winter precipitation. These findings counter the widely held belief that climate change will increase the depth, frequency and duration of soil frost.
Climate Change Influences on Snow Distribution and Melt in a Mountain Catchment
Assessing the effect of climate warming on snow distribution and melt is restricted by a lack of quality forcing data for simulation modeling. Twenty-five years (1984 – 2008 water years) of high quality meteorology, precipitation and snow data from multiple sites within a small mountain catchment in the Owhyee Mountains of Idaho, USA are available. These data provide forcing parameters for spatial simulation as well as independent validation of simulated results. The forcing data include air and soil temperature, humidity, solar and thermal radiation, wind speed and direction, and precipitation at multiple measurement sites within the catchment over the 25-year period of record. Data from continuously operated snow pillow along with by- weekly snow course data provide validation, along with detailed snow surveys in the later years of the record (after 2000). The Isnobal physics-based, spatially distributed snow model is used to simulate both the development and melting of the seasonal snowcover in each of the years of analysis. This analysis shows how the development, distribution and drifting, and melting of the seasonal snowcover has been affected by climate warming over the period of record.
Scale-Related Effects of Interactions Between Topography, Climate, and Seasonality on Snow Accumulation and Melt
The heterogeneity of mountain snow distribution and melt has vital implications across all spatial scales. Computationally efficient means of capturing this is required by fine-scale models applied to small mountain watersheds where heterogeneity affects runoff, erosion, water quality, and ecology; by moderate-scale models applied to larger river basins where its effect on runoff peaks and timing are seen; and in global-scale models where snow affects earth-atmosphere fluxes and circulation patterns. Further understanding of how the heterogeneity of snow distribution and melt scale will serve to guide determinations of appropriate model structures. Snow distribution is determined by meteorological forcings at the earth's surface with scaling largely dependent on the interactions between weather and topography. Meteorological forcings and associated scaling characteristics relative to snow distribution are constantly changing in response to local, synoptic, seasonal, and decadal cycles. Determining appropriate length scales for representing spatial variability across all conditions is essential for capturing both short-term impacts and those associated with long-term climate change. This research examines the interactions between weather, seasonality, and terrain and their effects on the scaling properties of the forcing processes controlling snow distribution and melt. The research is conducted using a physically-based, spatially distributed snow model with a process- based model scale to simulate snow distribution and melt over a 14 km2 mountainous catchment in southwestern Idaho, USA. Length scales of all forcing parameters are analyzed for different climatic scenarios and at different times of year. The sensitivity of modeled snow distribution and melt to the scaling of the forcing parameters is also characterized. This research will serve as a guide in determining which and under what conditions, snow-distribution processes can be appropriately up-scaled, and likewise which processes and under what conditions require an accounting of sub-grid variability.
Development of a one-dimensional vertical multilayer energy exchange model for blowing snow
In order to predict energy exchanges in the blowing snow layer, a simple vertical multilayer energy exchange model for blowing snow (FUBUKI model) has been developed. Fundamental elements characterizing energy exchanges in the blowing snow layer have been analyzed between 10-m and 0.1-m heights above the snow surface every 0.01-m height for friction velocities of 0.5m/s and 0.75m/s. The results reveal the following. When blowing snow occurs, wind profiles become weak. Air temperature and specific humidity change greatly near the lower layer due to increase in number of blowing snow particles. Downward shortwave radiation decreases and upward shortwave radiation increases with a decrease in height above the snow surface. Since blowing snow particles increase emissivity of the blowing snow layer, downward longwave radiation increases with an decrease in height above the snow surface and a change of upward longwave radiation is a little due to high emissivity of the snow surface. Blowing snow effects on downward longwave radiation have much in comparison with downward shortwave radiation. Although these results for fundamental elements characterizing energy exchanges in the blowing snow layer are consistent with those obtained by the field observations and the wind tunnel experiments, the reduction of the uncertainty of parameters describing the characteristics of blowing snow would require the incorporation of parameters based on systematic ground- based studies of the structure of blowing snow in specific fields.
Evaluation of regional simulations of snow cover over the Austrian Alps
Better representation of snow processes in regional climate models is a key important requirement in the development of strategies for successful management of water resources and natural hazards in the future. The objective of the research presented here is to compare a new implementation of a grid-based hydrological model embedded in a model of land-surface climatology (the Joint UK Land Exchange Scheme; JULES) with observations of snow cover available in Austria over the past 30 years. The JULES model was driven with observed precipitation and air temperature. Additional meteorological driving variables for the land surface model (air humidity, wind speed, net radiation) were obtained from regional climate simulations driven by the ERA-40 reanalysis. The results of simulations were evaluated against daily snow depth observations at 754 climate stations. We found the overall accuracy with which the model simulated snow cover was 89% for the entire year and 75% during the winter months. The comparison of spatial and temporal snow cover patterns indicates that the model accurately reproduces the seasonal accumulation and melt in lowland and mountain regions. The implications of these findings will be discussed in the context of climate change simulations using the modeling framework we have developed.
Reconstructing Snowmelt in Idaho's Watershed Using Historic Streamflow Records: Methods and Applications
Warming climate appears to have influenced Idaho's snowpack and the timing of spring snowmelt; however, records of the timing of snowmelt only extend to the 1980's. Stream gage data can extend records of the timing of snowmelt in snowmelt-driven systems. We used SNOTEL and historic streamflow records to test the applicability of Short Time Fourier Transform (STFT) wavelet analysis of hydrographs as a method to reconstruct final snowmelt dates into the early 19th century. STFT reconstructions of snowmelt tested against known historic final snowmelt dates indicate final snowmelt dates can be determined within ± 4 days ~ 95% of the time and within ± 7 days 100% of the time. Comparison of the STFT method with the center of timing method indicates the STFT method may limit errors in interpretation associated with changes in discharge not related to snowmelt. Of the sites that have streamflow records that extend into the early 1900's, most show early snowmelt years in the 1920's-1930's, later and less variable snowmelt dates in the 1940's-1970's, and both variable and early snowmelt from the mid-1980's to present. Early and variable snowmelt during the last ~ 20 years in the Boise National Forest is associated with large wildfires. This method appears to work equally well in extending records of final snowmelt most western snow dominated watersheds of North America. Ongoing research compares the extended snowmelt data with historical fire seasons while focusing on the development of soil moisture drying curves to be used with current final snowmelt dates as a predictor to the onset of fire conditions in Idaho. Future work will extend additional final snowmelt records in the western US states and will examine relationships between final snowmelt dates and the onset of fire conditions in the western US and their teleconnections with regional and global scale climate-forcing mechanisms (i.e. ENSO, PDO and AO).
Seasonal Characteristics of Surface Meteorological and Radiative Fluxes on the East Rongbuk Glacier in Mt. Qomolangma (the Mt. Everest) Region
Ground-based measurements are essential for understanding alpine glacier dynamics, especially in remote regions where in-situ measurements are extremely limited. The meteorological and radiative fluxes were measured over the accumulation area on the East Rongbuk Glacier, Mt. Qomolangma (the Mt. Everest) at elevation of 6,560 m a.s.l. Measurements were conducted using an automatic weather station (AWS) from May 1 through July 22, 2005 (spring-summer period) and from October 2 of 2007 through January 20 of 2008 (autumn-winter period). Surface meteorological and radiative characteristics were strongly controlled by two major synoptic circulation regimes: the southwesterly Indian monsoon regime in summer and the westerlies in winter. At the AWS site on the East Rongbuk Glacier, north or northwest winds prevail with higher wind speed (up to 35 ms-1 in January) in winter and south or southeast winds predominate after the onset of the southwesterly Indian monsoon with relatively low wind speed in summer. Intensity of incoming shortwave radiation is extremely high due to its high elevation and high reflective surrounding surface. The striking feature is that the observed 10-minute mean incoming shortwave radiative fluxes around local noon were frequently higher than the solar constant at the top of the atmosphere from May through July, 2005. The observed higher-than-solar-constant values are mainly due to the impact of local convective broken clouds and high surface reflectivity over the surrounding terrains. We estimated that horizontal component of received diffusive solar radiation from surrounding terrains ranged from 140 to 310 Wm-2, accounting for about 10 to 25% of the observed incoming shortwave radiation under clear sky conditions. This value could be even higher under overcast cloudy days. The mean surface albedo ranged from 0.72 during summer- spring period and 0.69 during the autumn-winter period. The atmospheric incoming longwave radiation was strongly controlled by cloud conditions and atmospheric moisture content. Overall impact of clouds on net radiation balance is negative in Mt. Qomolangma region. The daily mean net all-wave radiation was positive during the entire spring-summer period and mostly positive during the autumn-winter period except a few overcast cloudy days. On monthly basis, net all-wave radiation was always positive.
Cellular Automata Model for Simulating Wind Transport of Snow and the Interaction with Topography and Alpine Vegetation
A cellular automata model is proposed for simulating the evolution of snow packs in areas in which wind transport of snow and the interactions with terrain and short alpine vegetation are dominant. The model is designed to work at small spatial scales (1 m) and over weekly time steps. The processes simulated include a layered snow pack formed by accumulating weekly precipitation, a physically based densification scheme that accounts for the compaction of the snow layers, and the interaction between the blowing snow with small- scale topographic features and vegetation. Other features include the possibility of time-variable transport trajectories, space- and time-variable precipitation, and time-variable initial density. The removal and deposition of particles is determined according to a predefined set of probabilities dependent upon the relative location of the grid cells to aerodynamic obstacles and the vertical angles with such obstacles. The interaction with the vegetation is simulated using a set of probabilities that depend on the height of the vegetation, and a relationship that relates the exposed vegetation height to the vegetation effectiveness. The model is applied to combinations of synthetic topographic and vegetation fields. Preliminary results illustrate that the correlation structure of the snow depth fields becomes stronger as the amount of snow transported increases, while the probability distributions of the fields progress from a Gaussian distribution for small transport values to positively skewed probabilities for high transport values. These results are similar to what we have observed for Light Detection and Ranging (LIDAR) snow depth fields in alpine and wind dominated environments, for which strong correlation structures and positively skewed distributions have been obtained. Further results regarding the effect of vegetation, dynamic wind patterns and precipitation on the characteristics of the spatial organization of the fields will be presented.
Arctic Snow and Diurnal Changes in Albedo
Local snow albedo on ultraviolet (UVA, UVB, erythemally weighted UV) and visible (VIS) region, at wavelengths of photosynthetically active radiation (PAR, 400-700 nm), has been measured in at Sodankylä (67 o N, 26 o E, 179 m a.s.l.) during the spring of 2007. SL501 and NILU-UV radiometers at 2 m height were used. The data were logged at 1-min intervals. The measurement period covered accumulation and melting of snow. The accumulation of snow was up to 68 cm. The surface layer thickness varied from 0.5 to 35 cm with the snow grain size between 0.2 and 2.5 mm. According to SL501 results, the midday erythemally weighted UV albedo ranged from 0.6 to 0.8 in the accumulation period, and from 0.5 to 0.7 during melting. During the snow melt period, under cases of an almost clear sky and variable cloudiness, an unexpected diurnal decrease of 0.05 in erythemal albedo soon after midday, and recovery thereafter, was detected. This diurnal decrease in albedo was found to be asymmetric with respect to solar midday, thus possibly indicating a change in the properties of the snow. Independent NILU-UV results on UVA and UVB confirm these findings. In addition, in some cases diurnal asymmetry in albedo (with or without diurnal decrease and recovery) was observed. The results are discussed in context of corresponding Antarctic results, too.
Snow and Soil Moisture Response Across Elevation, Aspect and Canopy Variables in a Mixed-conifer Forest, Southern Sierra Nevada
A water-balance instrument cluster that included over 27 snow-depth measurements and vertical profiles of
soil temperature and volumetric water content was deployed in summer and fall of 2007 at the Southern
Sierra Critical Zone Observatory (CZO), at an elevation of 1,600-2,000 m. The CZO is co-located with the
Kings River Experimental Watersheds, a U.S. Forest Service integrated watershed research site. Instruments
were deployed to capture both north- and south-facing aspects, as well as differences in canopy cover
across the instrument cluster. Snow on south-facing slopes melted before than on north-facing slopes,
resulting in drier, warmer soils after spring snowmelt. Soils in the open dried faster than those at the canopy
drip edge, also reflecting earlier snowmelt. Soils reached near saturation during a multi-day summer rain
event, followed by more rapid drying of near-surface layers. Soils are largely decomposed granite; however,
organic content in near-surface soils influenced measured values of volumetric water content. Macropores,
observed at some sites during installation of the soil moisture probes (Echo TE, Decagon) also influenced
relative vertical values of soil-water content.
A Modeling Framework for Improved Agricultural Water Supply Forecasting
The National Water and Climate Center (NWCC) of the USDA Natural Resources Conservation Service is moving to augment seasonal, regression-equation based water supply forecasts with distributed-parameter, physical process models enabling daily, weekly, and seasonal forecasting using an Ensemble Streamflow Prediction (ESP) methodology. This effort involves the development and implementation of a modeling framework, and associated models and tools, to provide timely forecasts for use by the agricultural community in the western United States where snowmelt is a major source of water supply. The framework selected to support this integration is the USDA Object Modeling System (OMS). OMS is a Java-based modular modeling framework for model development, testing, and deployment. It consists of a library of stand-alone science, control, and database components (modules), and a means to assemble selected components into a modeling package that is customized to the problem, data constraints, and scale of application. The framework is supported by utility modules that provide a variety of data management, land unit delineation and parameterization, sensitivity analysis, calibration, statistical analysis, and visualization capabilities. OMS uses an open source software approach to enable all members of the scientific community to collaboratively work on addressing the many complex issues associated with the design, development, and application of distributed hydrological and environmental models. A long-term goal in the development of these water-supply forecasting capabilities is the implementation of an ensemble modeling approach. This would provide forecasts using the results of multiple hydrologic models run on each basin.
Simple Method to Evolve Daily Ground Temperatures From Surface Air Temperatures in Snow Dominated Regions
The Simple Land-Interface Model (SLIM) was modified to evolve daily ground temperatures from surface air temperatures (SAT) in snow-dominated areas. Daily ground surface temperature (GST) was modeled as a function of daily surface air temperature and snow depth. Analytical solution to the 1-D thermal diffusion equation was then used to simulate subsequent propagation of the GST signals into the subsurface. Time dependent apparent thermal diffusivity was modeled as a simple sinusoid and incorporated into the framework to account for the effects of seasonal latent heat and nonconductive heat transfer processes. The model was tested in snow-dominated sites such as Barrow, Council and Ivotuk in Alaska, and Reynolds Creek Experimental Watershed in Idaho. The model captured much of the seasonal dynamics of the ground thermal regime at all sites. It underestimated the fall ground temperatures in Barrow and Ivotuk in some years and overestimated spring ground temperatures in Council.
Snow and Soil Improvements to the Noah Land Surface Model for use in the Arctic System Reanalysis
Although the Noah land surface model is used globally, several aspects of the model as not suited for Arctic land processes. In this study, the effects of three model configuration and physics changes are tested. To better simulate the evolution of the snowpack, a three-layer snow model was implemented, replacing the existing one-layer model in Noah. When compared to river flow data, this model produces better timing of spring runoff. The three-layer model also improves heat transfer into the underlying soil. To improve the simulation of soil temperature and active layer depth, more and deeper soil layers were added to Noah and the bottom boundary condition was changed from a difficult to determine fixed temperature to a zero-flux bottom boundary. All of these changes were made to improve land surface state simulation in the upcoming Arctic System Reanalysis.
Importance of Ground Heat Flux in the Energy Balance of Shallow Snow
Ground heat flux is commonly, and rightfully, neglected in energy balance models of deep snowpacks.
However, ground heat flux can be important in the patchy, shallow snow that characterizes large land areas in
the semi-arid mountains of the western United States. Shallow snow, 10cm or less, has an energy balance
distinct from deep snowpacks primarily because solar radiation can effectively penetrate to the ground
through less than 10 cm of snow. Additional atmospheric energy through the shallow snow to the soil results
in elevated soil temperatures causing a significant increase in ground heat flux. This study quantifies the
ground heat flux (G) and its fraction of the energy balance beneath shallow and deep snow in complex
terrain. Ground heat flux was calculated from snow/soil interface temperatures logged using distributed
temperature sensing (DTS). The fiber-optic DTS collected high spatial resolution (1m) soil-snow interface
temperatures from all aspects and slopes of a semi-arid watershed for the eight-week duration of the spring
2008 snowmelt. Results show significantly higher ground fluxes and higher G as a percent of the snow
energy balance beneath shallow snow when compared to deeper snow.
Determining subgrid variability in snow water equivalent surrounding operational snow stations of the Western U.S.
We characterized the spatial distribution of snow depth and snow water equivalent (SWE) within 16-, 4-, and 1-km2 grid elements surrounding 13 operational snow stations (i.e. snow telemetry (SNOTEL) and California Snow Sensors) across four regions of the Western U.S.. Field observations of snowpack properties and binary regression tree models were used to characterize relationships between at-snow pillow snow depth and the depth of surrounding areas. Field campaigns involved intensive field surveys of the distribution of snowpack properties (snow depth, grain size, SWE, and snow temperature) across the 1- km2 area surrounding the snow stations, including over 690 snow depth measurements at each site; with repeated sampling over the season this included nearly 15,000 sample points. The root mean square deviation between measured snow depth at the snow stations versus the mean of the surrounding areas was 17, 14, and 12% relative to the mean observed snow depth at the Idaho-Oregon, California, and Colorado sites, respectively. Snow depth values at the snow pillows were as much as 70% greater than the corresponding mean grid-element SWE; indicating considerable overestimation of total snow accumulation. These relationships were not consistent from site to site or from region to region as some sites underestimated mean snow depth by as much as 30%. These analyses have relevance for land- atmosphere studies that require ground observations for updating snow information; an example is shown using the SWE product of the National Operational Hydrologic Remote Sensing Center. Field campaigns will continue in 2009 and 2010 and results and data will be disseminated to the broader community for use in regional scale land surface modeling studies and operational streamflow forecasting.
Spatio-temporal Variability of Melt Intensity over the Greenland ice sheet from 2000-2005 using coupled MODIS Optical and Thermal Measurements
Increased ice sheet velocity in the equilibrium zone of western Greenland Ice Sheet coincident with periods of summer melting has been demonstrated and attributed to infiltrated melt water that enhances glacial sliding. The assessment of surface melting beyond a binary classification of melt and no-melt events using passive microwave techniques, has been demonstrated using a liquid water fraction (LWF) retrieval model applied to higher resolution, cloud-free, composited MODIS optical and thermal data. Estimates of LWF were derived for composited periods from May through August for 2000 through 2005. An increase in the areal distribution of estimated LWF varies from (0-1%) during May to upwards of 15% later in the season inter- annually. A comparison to QuikSCAT derived melt zones indicate low LWF amounts associated with dry snow zones and higher LWF amounts with wet snow zones. This relationship holds spatially and temporally during the analysis period.
Improving Snow Measurement Technology to Better Parameterise Cold Regions Hydrometeorology Models
Marmot Creek Research Basin, in the Rocky Mountains of Alberta, Canada constitutes a long term cold regions hydrometeorological observatory with over 45 years of intensive observations in alpine and forested zones. Recently, novel combinations of measurement technology to snow have been deployed in Marmot Creek to advance the understanding of snow processes and to improve hydrometeorological models of streamflow and atmospheric variables. One advance has been the development and application of portable acoustic reflectometry to measure the density and structure of seasonal snowpacks using an audible sound wave. This has permitted the non-invasive measurement of snow water equivalent for both stationary and snow survey applications. Another advance has been the use of oblique time-lapse digital photography which is corrected for elevation and view angle from a LiDAR DEM to produce daily orthogonal snow covered area images of the alpine zone. These images are used to calculate snowcovered area and to develop and test improved snowcover melt and depletion algorithms. Deployment of 3-axis ultrasonic anemometers and fast hygrometers with collection of 10 Hz data and full correction for non-stationarity, axis rotation and other effects has shown that horizontal turbulence is often advected into mountain clearings and causes failure of traditional bulk transfer calculations of latent and sensible heat. For forest snow a hanging, weighed spruce tree and hanging, weighed sub-canopy troughs are used to capture intercepted snow load and unloaded snow fluxes respectively. These quantities provide the information needed to test detailed models of the snow interception and unloading processes. To quantify variations in sub-canopy energy for snowmelt, infrared imaging radiometers and narrow beam radiometers are used to measure thermal radiation exitance from needles, stems and trunks in forests of varying structure. These measurements are being used to develop improved models of longwave radiation transfer between forests and snow. Improved algorithms resulting from this application of field technology are being used to update a modular, object-oriented computer simulation of the cold regions hydrological cycle, the Cold Regions Hydrological Model, CRHM. CRHM can be easily and frequently updated as improved algorithms become available and used to test the sensitivity of snow hydrology calculations to these improvements.
Accessing and Sharing Data Using the CUAHSI Hydrologic Information System
The Consortium of Universities for the Advancement of Hydrologic Science, Inc (CUAHSI) has a Hydrologic
Information System (HIS) project, which is developing infrastructure to support the sharing of hydrologic data
through web services and tools for data discovery, access and publication. Centralized data services
support access to National Datasets such as the USGS National Water Information System (NWIS) and
SNOTEL, in a standard way. Distributed data services allow users to establish their own server and publish
their data through CUAHSI HIS web services. Once such a data service is registered within HIS Central, it
becomes searchable and accessible through the centralized discovery and data access tools. The HIS is
founded upon an information model for observations at stationary points that supports its data services. This
is implemented as both XML and relational database schema for transmission and storage of data
respectively. WaterML is the XML based data transmission model that underlies the machine to machine
communications, while the Observations Data Model (ODM) is a relational database model for persistent data
storage. Web services support access to hydrologic data stored in ODM and transmitted using WaterML
directly from applications software such as Excel, MATLAB and ArcGIS that have Simple Object Access
Protocol (SOAP) capability. A significant value of web services derives from the capability to use them from
within a user's preferred analysis environment, rather than requiring a user to learn new software. This
allows a user to work with data from national and academic sources, almost as though it was on their local
disk. This poster will be computer-based with internet access for demonstration of HIS tools and functionality.
Snow-Depth Variability and Implications to Tundra Travel on the North Slope of Alaska
Information regarding the timing and amount of snow accumulation/ablation is essential for tundra-travel decisions by industry and management agencies on the North Slope, as well as for spring snowmelt estimations and for the modeling of hydrologic processes. Research projects in the Sagavanirktok River/Bullen Point area and Kuparuk River basin contain meteorological stations, eighteen of which utilize Campbell Scientific (SR50 and SR50A) ultrasonic depth sensors to collect hourly snow-depth observations. A number of the more accessible stations are visited on a monthly basis throughout the winter season to conduct snow surveys, including snow-depth transects, and to measure snow-depths directly under the sensors. In addition to these visits, an extensive snow survey campaign is conducted between April and May to estimate the end-of-winter snow water equivalent (SWE). Snow sensor data in conjunction with field observations were used to estimate snow-depth at different spatial scales. Snow-depth cumulative distribution functions at each station show the applicability of relating point data to area-wide scales. Generally, the North Slope receives relatively little snow fall on the coastal plain, with slightly deeper snow in the foothills regions. The height variability in the tussock land surface is often similar to the depth of cumulative snowpack. However, wind redistribution of snow tends to fill in these depressional areas during the early winter. The combination of thin snow covers, variable ground cover roughness, and wind redistribution makes this a challenging environment to quantify the early winter snow cover conditions; data that is critical for resource management decisions.
Snowpack and Streamflow Trends in the Cascade Mountains of Northern Washington, USA and Southern British Columbia, Canada
Most studies of snowpack trends focus on the snow water equivalent (SWE) measured at snowcourses on or near 1 April. Although these data are generally the longest time series of SWE available in North America, the single measurement per year misses other behaviors of the snowpack, which may be as sensitive or even more sensitive indicators of the effects of climate change and global warming. Daily SWE data from automated snow pillows can be used to assess trends in these other behaviors, such as the peak SWE and the date of its occurrence, the date of initiation of permanent winter snowpack, and the date of melt out. Although the time series available for daily SWE measurements are of significantly shorter duration than the 1 April snowcourse measurements (20-30 years rather than 70-80), it is nevertheless worth looking at the snowpack characteristics that can be derived from daily data to see what trends can or cannot be seen. Such analyses were carried out for a few selected sites from the USDA Natural Resources Conservation Service SNOTEL network and the British Columbia automated snow pillow network in the Cascade Mountains of northern Washington and southern British Columbia. Trends for the four quantities mentioned above (peak SWE, date of peak SWE, date of initiation of permanent winter snowpack, and date of melt out) were examined. Although most of the trends were not statistically significant due to the short record lengths and high degree of variability, they were generally what one would expect in a warming climate, such as later initiation of winter snowpack and earlier date of peak SWE. A simple analysis of streamflow timing was also carried out to see if these trends were consistent with the snowpack trends. The analysis was carried out on five rivers, three on the east side of the Cascades and two on the west side, all of which have their headwaters in the same vicinity of the snow sites analyzed. The quantity examined was the monthly fraction of total water year flow. These were computed as time series, and changes in behavior were noted over time. These data were also grouped according to Pacific Decadal Oscillation (PDO) eras, and average monthly fractions for each era were computed. The primary trend noted was an increase in the winter (November-January) flow fraction and an increase of the March fraction coupled with a decrease in the April fraction. There was also a notable difference in flow fractions between the two warm PDO eras represented in the time series (1925-1946 and 1977-2008) in that the current warm PDO era has had more flow in the winter and early spring than did the previous PDO warm era, suggesting that the current PDO warm era is different from (warmer than) the previous one.
Avalanche Crown-Depth Distributions
The literature disagrees about the statistical distribution of snow avalanche crown depths. Large datasets from Mammoth Mountain, California and the Westwide Avalanche Network show that the three-parameter generalized extreme value distribution provides the most robust fit, followed by a two-parameter variation, the Fréchet distribution. The most parsimonious explanation is neither self-organized criticality nor other complex cascades, but the maximum domain of attraction, implying that distributions of individual avalanche crown depths are scaling. We also show that avalanches do not have a universal tail index. Rather, they range from 2.8 to 4.6 over different avalanche paths, consistent with other geophysical phenomena such as wildfires, which show similar variability.
SnowSTAR2002 transect reconstruction using a multilayered energy and mass balance snow model
The lateral and vertical variability of snow stratigraphy was investigated through comparison of measured profiles of snow density, temperature and grain size obtained during the SnowSTAR2002 1200-km transect from Nome to Barrow with model reconstructions from SNTHERM, a multilayered energy and mass balance snow model. Model profiles were simulated at the SnowSTAR2002 observation sites using ERA-40 reanalysis 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 provide good simulations for snowpack temperature, with larger errors for grain size and density. A spatial similarity analysis using semivariograms of measured profiles shows that there is diverse spatial and profile variability for snow properties along the SnowSTAR2002 transect resulting from differences in initial snow deposition, influenced by wind, vegetation, topography, and post-depositional mechanical and thermal metamorphism. The correlation length in snow density (42km) is quite low, whereas it is slightly longer for snow grain size (125km) and longer still for snow temperature (130km). 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. We used the Microwave Emission Model for Layered Snowpacks (MEMLS) 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) 1.5K for vertical polarization, increasing to 8.7K for horizontal polarization), and somewhat larger (5.9K for vertical polarization, and 6.2K for the horizontal polarization) at 37 GHz. These results successfully capture the variability along the transect and imply that simulation of snow microphysical profiles is a viable strategy for passive microwave satellite-based retrievals of SWE.
Pattern formation in snow during temperature gradient metamorphism
Temperature gradient metamorphism causes sublimation and growth of crystals. This process causes a dramatic change in thermal and geometrical properties. Using a time-series of snow evolution, we simulated the evolution of the thermal conductivity parallel and perpendicular to the temperature gradient direction. Thermal conductivity changed within a few days from an isotropic property to a strongly anisotropic property. Surprisingly, these changes are only marginally reflected in the geometrical anisotropy of the full snow microstructure. We also observed that the heat flux in the microstructure is concentrated in a small part of the ice matrix, which causes a high tortuosity. The percentage of the ice matrix involved in high heat fluxes was almost constant over time. However, the connectivity of these heat-conducting ice structures increased. The formation of an anisotropic temperature conductivity could have important consequences in terrain where temperature gradients are not perpendicular to the surface, as in shallow snowpacks over hummocky terrain or in boulder areas, or where the snowpack has a strong surface topography, e.g. due to sastrugi formation.
Spatial, Seasonal, and Interannual Variability of Snow Accumulation Control Mechanisms in two Neighboring Alpine and Sub-alpine Catchments in California's Seasonally Snow- covered Southern Sierra Nevada
Accurate representation of annual snow accumulation in space and time is needed to properly model the hydrologic system, but is complicated by considerable spatial, interannual and seasonal variability as well as its sensitivity to changes in climate. Regression tree model predictions of distributed snow depth, while statistical in nature, have been shown to capitalize on the hypothesis that snow accumulation patterns are governed by identifiable and measurable local-scale physiography. The independent variables net solar radiation, elevation, slope, aspect, vegetation density, and maximum upwind slope have previously been successfully applied to snow distribution in the southern Sierra Nevada. Common trends in the relative importance of these independent variables are examined here for different survey periods, water years, and physiographic environments. This work presents new results from snow surveys conducted within the 19.1 km2 alpine Tokopah Basin (April 2008) and the neighboring 7.5 km2 forested, sub-alpine Wolverton Basin (February, March and April 2008) with an average of 206 depth measurements per survey. Results are compared to previous findings in the Tokopah for years 1997 - 2005. Regression tree results from the Tokopah survey show an optimal tree size of eight terminal nodes (R2=0.39). The tree structures and normalized predictions are similar to previous years despite significant differences in precipitation magnitude (e.g. April 1 precipitation was 60% and 2% above average for 1997 and 2008, respectively, but tree model structure was nearly identical). The relative importance of snow accumulation control mechanisms differed both seasonally as well as between the alpine and sub-alpine basins. Solar radiation was the primary independent variable in the Tokopah Basin whereas elevation and maximum upwind slope governed the snow distribution in the forested Wolverton Basin. Current and on-going work involves the incorporation of regression tree results into a physically-based snowmelt model and data assimilation scheme to resolve the spatial distribution of snowfall and accumulation patterns in the region and will provide a basis for inter-storm and inter-site comparisons under a variety of physiographic environments.
Five Snow Sublimation Estimates From Barrow, Alaska Using Eddy Correlation and One- and Two-Level Profile Methods
Published rates of snow sublimation range from 3 to 80% of total winter precipitation, depending on the site, type of instrumentation, and the methods used. Here we present estimates of snow sublimation from Barrow, Alaska for March through May 2005. Three independent sets of meteorological instruments were operated during this period, including an eddy correlation tower with a sonic anemometer and krypton hygrometer, a small (3-m) gradient tower with four levels of wind and temperature, and a large (40-m) gradient tower with wind and temperature measured at two heights. In addition, high frequency humidity data (20 Hz) was collected at two heights on the eddy correlation tower. Using these data, we have computed surface sublimation estimates using five methods: 1) the eddy correlation method using the moisture results from the krypton hygrometer, 2) the eddy correlation method using a humidity sensor (HMP), 3) the aerodynamic profile method using temperature and wind at two heights, 4) the K-theory method using wind and temperature at two heights, and 5) the one-level bulk profile method. Using the HMP in eddy correlation computations is not customary but showed surprisingly similar results as the krypton hygrometer. Depending on method, we estimate that sublimation removed between 1 and 30% of total winter precipitation for the measured period. The highest sublimation rates were produced by the aerodynamic profile method and the lowest rates were from the eddy correlation method using the HMP. Computed rates for all methods except eddy correlation methods were sensitive to which lower temperature sensor was selected (we had several choices), suggesting that an accurate near-snow surface temperature is crucial to producing reliable values. Based on a survey of the existing literature, and our computational results, we suggest that the wide range of reported sublimation rates arises in part due to problems of both sensors and computational errors.
Contributions to Glacier Energy Balance From Turbulent Heat Fluxes and Meltwater Percolation Dynamics
An effective way of resolving the amount of melt from a glacier in a given period is to conduct an energy balance survey on its surface. The energy fluxes between the snow or ice surface, the underlying snow or ice, and the atmosphere all contribute to the amount of resulting energy available for melt. We will present results from experiments directly above and below the snow/ice surface, lending insight into two elements contributing to glacier energy balance: turbulent heat fluxes, and meltwater percolation effects. Field experiments took place at the Opabin Glacier (Yoho National Park, BC, Canada) and at the Haig Glacier (Peter Lougheed Provincial Park, AB, Canada) during the 2007 and 2008 melt seasons. Preliminary findings show emerging properties of turbulent heat flux parameters as the surface evolves over the melt season, as well as the capability of detecting meltwater percolation and refreezing within the supraglacial snowpack.
Parsimonious snow model explains reindeer population dynamics and ranging behavior
Winter snow is a key factor affecting polar ecosystems. One example is the strong negative correlation of winter precipitation with fluctuations in population in some high-arctic animal populations. Ice layers within and at the base of the snowpack have particularly deleterious effects on such populations. Svalbard reindeer have small home ranges and are vulnerable to local "locked pasture" events due to ground-ice formation. When pastures are locked, reindeer are faced with the decision of staying, living off a diminishing fat store, or trying to escape beyond the unknown spatial borders of the ice. Both strategies may inhibit reproduction and increase mortality, leading to population declines. Here we assess the impact of winter snow and ice on the population dynamics of an isolated herd of Svalbard reindeer near Ny-Ålesund, monitored annually since 1978, with a retrospective analysis of the winter snowpack. Because there are no long-term observational records of snow or snow properties, such as ice layers, we must recourse to snowpack modeling. A parsimonious model of snow and ground-ice thickness is driven with daily temperature and precipitation data collected at a nearby weather station. The model uses the degree-day concept and has three adjustable parameters which are tuned to correlate model snow and ground-ice thicknesses to the limited observations available: April snow accumulation measurements on two local glaciers, and a limited number of ground-ice observations made in recent years. Parameter values used are comparable to those reported elsewhere. We find that modeled mean winter ground-ice thickness explains a significant percentage of the observed variance in reindeer population growth rate. Adding other explanatory parameters, such as modeled mean winter snowpack thickness or previous years' population size does not significanly improve the relation. Furthermore, positioning data from a small subset of reindeer show that model icing events are highly correlated to an immediate increase in range displacement between 5-day observations, suggesting that Svalbard reindeer use space opportunistically in winter, a behavioral trait that may buffer some of the negative effects of the expected climate change in the Arctic.
Evaluation of a Coupled Snow Hydrology-Emission Model using SWE observations from L- Band Laboratory Measurements and from Field Experiments
An existing distributed snow hydrology model (Devonec and Barros 2002) was modified to incorporate transient snowpack physics from SNTHERM (Jordan, 1991) and radiative transfer processes in multilayered snowpacks from MEMLS (Matzler and Wiesmann, 1999). The ultimate objective is to use the model as a tool to integrate high space-time resolution ground-based measurements of snow water equivalent to generate analysis fields of snowpack properties for characterizing the sub-grid scale heterogeneity of snow fields in coarse scale remote-sensing measurements. To evaluate the model, we rely on two different types of observations: 1) the radiative transfer component is evaluated based on L-band measurements of SWE in a heterogeneous snowpack modeled in the laboratory by various combinations of foam layers with distinct dieletric and structural properties, as well as water content; and 2) the full model is validated against field measurements from CLPX and from the Himalayas.