C31A-0275 INVITED 0800h
The Sierra Nevada Snowpack: Do the Snow Course Data Show Historical Trends?
Among the postulated consequences of global warming, the hypothesis that the Sierra Nevada snowpack may decline is worrisome because of the dependence of the California economy on snowmelt runoff. Some Sierra Nevada snow courses have data records extending back to 1910, and about 100 of the courses have more than 70 years of data. A simplistic analysis shows little other than the obvious interannual variability, but a closer look reveals a declining April snowpack at the lower elevations, especially in the southern parts of the range. The low-elevation deficit is apparently not compensated by more snow at higher elevations.
C31A-0276 0800h
Estimating Snowfall in the Arctic
A new method is described for reconstructing snowfall from observed snowdepth records and the NASA Seasonal-to-Interannual Prediction Project Catchment Land Surface Model (NSIPP CLSM). This model is applied to the Arctic, where estimates of snowfall from gauges are highly uncertain. Results show that using the CLSM, which has a three-layer snow model, snowfall can be accurately reconstructed which had to have fallen to produce the observed snowdepth, given the model physics.
C31A-0277 0800h
Measuring and Modeling Snow Accumulation, Movement, and Ablation in Colorado Shrublands
Basin shrublands, featuring {\it Artemisia tridentata} and {\it Sarcobatus vermiculatus} shrubs accompanied by various graminoids, are widespread constituents of intermountain western U.S. landscapes. These shrublands are characterized by spatially heterogeneous winter snow covers interrupted by transient melting periods. The spatial heterogeneity of snow in shrubland landscapes is determined by interactions among topography, wind, net radiation, and vegetation structure (shrub vs. graminoid canopy dominants). Transient melting periods are typically produced by intermittent warm and dry phases where snow melts over much of the landscape, especially in wind-scoured grassy areas where snow depths are relatively shallow. Because snow is an important factor in the annual water and energy budgets of shrublands, it is important to quantify and understand snow accumulation, movement, and ablation processes in these landscapes. We employed snow depth observations and a snow evolution modeling system (SnowModel) to quantify and simulate snow depths in North Park, an intermountain basin in Colorado. Observations and simulations focused on a 0.25 km$^{2}$ study area centered on a 34 m tall meteorological tower deployed for the FLuxes Over Snow Surfaces (FLOSS) project (http://www.atd.ucar.edu/rtf/projects/floss/). Snow observations were measured along a series of transects, extending 400-600 m upwind of the tower, every 10 days from late December 2002 to late March 2003. The transects were designed to capture snow characteristics within shrub and graminoid cover types. Further, observations were used to parameterize and validate SnowModel. Hourly model simulations were performed from 1 October 2002 to 1 April 2003 using topographic, meteorological, and vegetation distribution data for the domain. Observed and modeled snow characteristics are described. Over 30,000 snow depth observations were collected during the field season. Snow cover in graminoid-dominated areas was shallow, less variable, and ablated quickly. In contrast, shrubs accumulated deeper, more variable snow depths that persisted longer. Model simulations were compared with these observations and indicated that the model reproduces the seasonal characteristics of snow in this system. Model similarities and deviations from observed values are detailed and related to potential refinements in our approach, scale issues, and driving data.
C31A-0278 INVITED 0800h
A Distributed Snow Evolution Modeling System (SnowModel)
A spatially distributed snow-evolution modeling system (SnowModel) has been specifically designed to be applicable over a wide range of snow landscapes, climates, and conditions. To reach this goal, SnowModel is composed of four sub-models: MicroMet defines the meteorological forcing conditions, EnBal calculates surface energy exchanges, SnowMass simulates snow depth and water-equivalent evolution, and SnowTran-3D accounts for snow redistribution by wind. While other distributed snow models exist, SnowModel is unique in that it includes a well-tested blowing-snow sub-model (SnowTran-3D) for application in windy arctic, alpine, and prairie environments where snowdrifts are common. These environments comprise 68% of the seasonally snow-covered Northern Hemisphere land surface. SnowModel also accounts for snow processes occurring in forested environments (e.g., canopy interception related processes). SnowModel is designed to simulate snow-related physical processes occurring at spatial scales of 5-m and greater, and temporal scales of 1-hour and greater. These include: accumulation from precipitation; wind redistribution and sublimation; loading, unloading, and sublimation within forest canopies; snow-density evolution; and snowpack ripening and melt. To enhance its wide applicability, SnowModel includes the physical calculations required to simulate snow evolution within each of the global snow classes defined by Sturm et al. (1995), e.g., tundra, taiga, alpine, prairie, maritime, and ephemeral snow covers. The three, 25-km by 25-km, Cold Land Processes Experiment (CLPX) mesoscale study areas (MSAs: Fraser, North Park, and Rabbit Ears) are used as SnowModel simulation examples to highlight model strengths, weaknesses, and features in forested, semi-forested, alpine, and shrubland environments.
C31A-0279 0800h
A Meteorological Distribution System for High Resolution Terrestrial Modeling (MicroMet)
Spatially distributed terrestrial models generally require atmospheric forcing data on horizontal grids that are of higher resolution than available meteorological data. Furthermore, the meteorological data collected may not necessarily represent the area of interest's meteorological variability. To address these deficiencies, computationally efficient and physically realistic methods must be developed to take available meteorological data sets (e.g., meteorological tower observations) and generate high-resolution atmospheric-forcing distributions. This poster describes MicroMet, a quasi-physically-based, but simple meteorological distribution model designed to produce high-resolution (e.g., 5-m to 1-km horizontal grid increments) meteorological data distributions required to run spatially distributed terrestrial models over a wide variety of landscapes. The model produces distributions of the seven fundamental atmospheric forcing variables required to run most terrestrial models: air temperature, relative humidity, wind speed, wind direction, incoming solar radiation, incoming longwave radiation, and precipitation. MicroMet includes a preprocessor that analyzes meteorological station data and identifies and repairs potential data deficiencies. The model uses known relationships between meteorological variables and the surrounding area (primarily topography) to distribute those variables over any given landscape. MicroMet performs two kinds of adjustments to available meteorological data: 1) when there are data at more than one location, at a given time, the data are spatially interpolated over the domain using a Barnes objective analysis scheme, and 2) physical sub-models are applied to each MicroMet variable to improve its realism at a given point in space and time with respect to the terrain. The three, 25-km by 25-km, Cold Land Processes Experiment (CLPX) mesoscale study areas (MSAs: Fraser, North Park, and Rabbit Ears) will be used as example MicroMet simulation domains, to highlight model strengths, weaknesses, and applications.
C31A-0280 0800h
Snow, Shrubs, Grasses, and Footprint Theory: Measuring Moisture and Energy Fluxes in Patchy Landscapes
When measuring sensible and latent heat flux from a tower within a heterogeneous landscape, one must consider which part of the landscape influences the flux sampled by the instruments. This variable landscape fraction, known as a footprint, is dependent upon wind direction, wind speed and atmospheric stability (thermal and mechanical). From 1 December 2002 - 31 March 2003, the FLuxes Over Snow Surfaces II (FLOSS II) field campaign measured sensible and latent heat fluxes at various heights on a 34 m tower in North Park, Colorado. North Park is an intermountain basin covered with a mixture of shrubs and graminoids (grasses and sedges) that interact with winter snow and wind to produce heterogeneous snow covers and, depending on the depth, protruding vegetation. During this period, snow depth measurements were made along transects extending 400-600 m upwind of the tower roughly every ten days. These snow depth data, in combination with blowing-snow model (SnowTran-3D) simulations, provided daily snow-depth distributions on a 1-meter grid over the area surrounding the flux tower. In addition, shrub height and vertical biomass profiles were measured and combined with a vegetation map having a 1-meter sampling scale. Merging the snow-depth distributions with the vegetation-height map allowed us to quantify the amount of vegetation protruding above the snow. This, in turn, allowed us to analyze the influence of exposed vegetation on observed energy and moisture fluxes. In this poster we describe our model for identifying the landscape fraction gauged by the flux-tower instruments as a function of commonly observed atmospheric conditions.
C31A-0281 0800h
Multi-angle/Multi-spectral Mapping of Snow Covered Area and Vegetation Density Using MISR
Vegetation structure and density affect the dynamics of snow accumulation and ablation. The presence of vegetation also affects our ability to accurately estimate snow-covered area (SCA) from satellite-based sensors. The objective of this case study is to simultaneously retrieve subpixel estimates of snow covered area and vegetation density from multi-angle imagery. Imagery from the Multi-angle Imaging SpectroRadiometer (MISR) was acquired over Glacier National Park and a portion of the Colorado Rocky Mountains. Vegetation density can be related to the shape of the angular signature for a pixel. Here, we invert the Rahman-Pinty-Verstraete (RPV) model to compute values of a semi-empirical parameter (the k-parameter) that is statistically correlated with vegetation density. For the same pixels, we perform linear spectral unmixing using the four-band multi-spectral data at each of the nine MISR viewing angles. In areas with a mixture of vegetation and snow, SCA estimates vary as a function of viewing angle and vegetation density. Using both the multi-angle and multi-spectral data from MISR, we are able formulate corrections for SCA estimates based on the vegetation density. Moreover, the vegetation density information itself is an important retrieved parameter that can be used in snowmelt/runoff models.
C31A-0282 0800h
Enhanced Snow Cover Mapping on the Tibetan Plateau using NASA EOS Optical (MODIS) and Passive Microwave (AMSR-E) Remote Sensing Data
Snow cover on the Tibetan Plateau has been the focus of considerable attention since the inverse relationship between winter snow cover and the intensity of subsequent Indian summer monsoon rainfall was first suggested more than a century ago. However, the Tibetan Plateau typically receives only small snowfall amounts, often accompanied by strong winds. This results in a pattern of spatially intermittent, i.e. patchy, shallow snow cover. Over the past several decades both optical and passive microwave satellite data have been available to map snow on the Tibetan Plateau. However, the coarse resolution of these historical remote sensing data sets is often not capable of resolving the pattern of intermittent snow. The higher spatial resolution of the MODIS and AMSR-E data, however, provides an improved capability to more accurately characterize this patchy snow cover pattern. As background, we present the historical multi-sensor Northern Hemisphere snow products for the period 1978 to 2003. Snow cover climatologies derived from these optical and microwave data generally compare favorably and where they do not, a reasonable explanation exists. The single region over which these two data sets disagree is the Tibetan Plateau where the microwave data appear to consistently indicate more snow cover than the visible data. We offer several explanations for this which include: frozen ground, soil type, mixed pixels of frozen and unfrozen water, influence of a reduced atmospheric thickness and the application of non-representative station data for validation. The availability of the higher spatial resolution MODIS and AMSR-E data, and the blending of these data, allow new opportunities to investigate this apparent anomaly. The blending process incorporates MODIS data at approximately 5 km (0.05 deg.) with microwave-derived SWE at 25 km, resulting in a blended product that includes percent snow cover in the larger grid cell. Relationships between the percent area covered by snow as indicated by the MODIS data and the threshold for the appearance of snow as indicated by the passive microwave data are presented. Comparison of snow cover derived from MODIS, SSM/I and AMSR-E over the Tibetan Plateau provides an example of the higher spatial resolution of AMSR-E yielding more accurate snow extent maps.
C31A-0283 0800h
Observational Study on Sublimation from the Snow Surface in Northern Eurasia
In northern Eurasia region, both of snow-cover water equivalence and winter precipitation has been demonstrated to be range of 50 to 150 mm. Sublimation from snow surfaces has been identified as an important hydrological process at high altitudes and in high latitude regions, involving complex mass and energy exchanges. Since 2002, intensive observations, to dressing winter hydrological land processes, have been conducted in Mogot experimental watershed, which is locating in the southern mountain taiga region of eastern Siberia of Russia (55.5_E#8249;N, 124.7_E#8249;E), and also on snow-tussock grassland in Nalaikh of Mongolia (47o45'N, 107o20'E), which locating in southern periphery region of Eurasia cryosphere. Snow sublimation has been quantitatively dressed by aero-dynamics or Lysimeter method. Effects of topography and vegetation cover to sublimation have been investigated, and the seasonality has been presented at both study sites. In taiga region of eastern Siberia, sublimation from snow surfaces differed with atmospheric stability. During early spring, a significant difference in snow sublimation was observed between slopes and valley bottoms, despite variable vegetation cover. However, during observation period, only one episode of significant snow sublimation was observed, which was caused by strong wind and ensuing light snowmelt. As atmospheric stability decrease, the effect of forest cover on snow sublimation was clear, with a significant difference between forested areas and open fields. Later in the spring, increased net all-wave radiation did not lead to an increase in sublimation. The effect of forest cover on snow sublimation can be seen from the estimated bulk transfer coefficient for latent heat. The bulk transfer coefficient was larger for a larch forest than that of open site. At two study sites, sublimation shows similar seasonal variation, even if the snow depth varied differently. Peak sublimation was occurred earlier spring coupling light snow melting. Modeling works demonstrated sublimation proportioned 17 to 52% to winter precipitation. Key words: Northern Eurasia, snow cover, sublimation
C31A-0284 0800h
Eddy Covariance(EC) Over Snow in a Mountainous Environment to Determine Sensible and Latent Heat and Mass Fluxes.
Sensible and latent heat and mass fluxes can represent a significant component of the snowcover energy and mass balance in mountain environments. Though these fluxes are computed in energy balance snow models, few measurements exist for comparison or validation. This research investigates the methodology required and problems associated with the direct measurement of sensible and latent heat flux over snow. The sensible and latent heat and mass fluxes can be determined directly from the turbulent fluctuations measured by fast-response sensors using eddy covariance (EC) theory. The general site considerations and specific weather conditions common to mountain catchments that affect EC data collection over snow is explored. Corrections and post-processing of eddy covariance data is discussed. Examples from established EC measurement sites under adverse and optimal conditions will be presented. The two primary EC study sites are located in southwestern Idaho in a small headwater catchment of the Reynolds Creek Experimental Watershed, located approximately 80 km southwest of Boise, Idaho. A protected, below canopy site is located within a stand of aspen trees, and an exposed site is located on a ridge over big mountain sagebrush. The study investigates the conditions under which EC instrument systems can be used in mountainous, snow-dominated environments, the processing required to prepare the data for analysis, and several examples of post-processed data collected over snow. This research will improve our understanding of how heat and mass flux from the snowcover may impact water resources under the variable and changing climate conditions so common in the Western US.
C31A-0285 0800h
The Effects of Vegetation Canopy Processes on Snow Surface Energy and Mass Balances
This paper addresses the effects of canopy physical processes on snow mass and energy balances in boreal ecosystems. We incorporate new parameterizations of radiation transfer through the vegetation canopy, interception of snow by the vegetation canopy, and under-canopy sensible heat transfer processes into the Versatile Integrator of Surface and Atmosphere (VISA) and test the model results against the Boreal Ecosystem-Atmosphere Study (BOREAS) data observed at South Study Area, Old Jack Pine (SSA-OJP). A modified two-stream radiation transfer scheme that accounts for the three-dimensional (3-D) geometry of vegetation accurately simulates the transferring of solar radiation through the vegetation canopy when the leaf and stem area index (LSAI) is reduced to match the observed, but the simulated wintertime surface albedos are higher than the observed. This overestimation can be removed by lowering the fractional snow cover on the canopy through the introduction of a snow interception model that explicitly describes the loading and unloading of snow and the melting and refreezing of snow. VISA overestimates the downward sensible heat fluxes from the canopy to the snow surface, which leads to earlier snow ablation and a shallower snowpack than the observed. Explicitly including a canopy heat storage term in the canopy energy balance equation decreases the spuriously large amplitude of the diurnal canopy temperature variation and reduces the excessive daytime sensible heat flux from the canopy downward to the snow surface. Sensitivity tests reveal that the turbulent sensible heat flux below the vegetation canopy strongly depends on the canopy absorption coefficient of momentum. During spring, the daytime temperature difference between the snow surface and the vegetation canopy forms a strongly stable atmospheric condition, which results in a larger absorption coefficient of momentum and a weak turbulent sensible heat flux. The modeled excessive downward sensible heat flux from the vegetation canopy to the snow surface is considerably reduced through the stability correction to the canopy absorption coefficient of momentum.
http://www.geo.utexas.edu/climate/Research/publications.htm
C31A-0286 0800h
Linking an Energy-Balance Snow Model (SNOBAL) to a Soil Temperature and Moisture Model (SHAW)
Accurate estimates of streamflow from mountain basins require an accounting for linkages between snow deposition and melt, and the moisture and temperature state of the soil. Though detailed snow deposition, energy state, and melt have been effectively simulated over mountain basins up to 2500 km$^{2}$ in the western US, equivalent soil moisture and temperature simulation has been limited to small plots or lumped extensions using land cover features over experimental catchments. As a step toward development of a fully coupled snow-soil energy and water balance model, we are testing a loose coupling of two models - SNOBAL for snowmelt and SHAW for below-ground temperature and moisture. This will involve forcing the below-ground component of SHAW with the output from SNOBAL, and will be limited to snow season conditions for this test. The objective of the initial coupling will be to determine the reliability of the simulation compared to measured conditions, and the sensitivity of the simulated snow-soil system to explicit rather than coupled feed-backs between the models. While the snowmelt model offers a numerically stable two-layer explicit solution that has been effectively extended over solution grids of 500,000 cells or more, the soil model uses a more numerically demanding central difference approach in which the number of nodes and layers vary with time and conditions. An objective of the test coupling of these models will be to determine how to simplify the representation of the soil thermal and moisture system, and still achieve an acceptable simulation of soil moisture and temperature. This research will result in the design requirements for a fully coupled snow-soil energy and water balance model that will improve our ability to manage limited water resources in the inter-mountain western US.
C31A-0287 0800h
Stratified Random Sampling Techniques for Snow Surveys of Mountainous Basins
Intensive snow surveys of mountain basins are the most accurate means of characterizing the heterogeneous mosaic of snow distribution typically present. As such, survey data is the preferred product for initializing and validating spatial snow models. The collection of survey data, even in small basins, is however costly and time-consuming. Snow survey data is typically collected along a standard grid, a spatial randomization based on a grid, or along pre-determined transects. Even in study areas of small to moderate size ($<$ 3 km$^{2}$) vital decisions on sample spacing and intensity must be made to adequately cover the entire basin while capturing large process-based snow-water-equivalent (SWE) differences that can occur at relatively small spatial scales. In this research, four years of survey data collected on a regularly spaced grid in the Reynolds Mountain East basin (0.36 km$^{2}$) in southwest Idaho were used to analyze stratified random sampling techniques designed to reduce the number of samples required to accurately portray snow distribution. It was found that a clustering algorithm based on prior survey data could substantially reduce the number of samples required to produce a surface of SWE similar to that produced by the full dataset. Clustering based exclusively on an a priori set of variables derived from the DEM and a distributed snowmelt model also produced satisfactory results.
C31A-0288 0800h
Streamflow Estimation from Hydrologic Model Assimilation of Remotely Sensed Snow Information in Snowmelt Dominated Basins
The USGS Precipitation Runoff Modeling System (PRMS) hydrological model was used to evaluate the utility of experimental, gridded, 1-km2 snow covered area (SCA) and snow water equivalent (SWE) products in improving the modeling of snowmelt runoff from three headwater basins in the Southwestern United Sates. The SCA product was the fraction of each 1-km2 pixel covered by snow and was derived from NOAA Advanced Very High Resolution Radiometer imagery. The SWE product was developed by combining the SCA product with SWE estimates interpolated from National Resources Conservation Service Snow Telemetry (SNOTEL) point measurements. An eight-year period (1995-2002) was used to compare PRMS simulated streamflow generated with and without the use of the SCA and SWE products. The test basins were the Upper Rio Grande (3,397 km2) in Colorado, and the White (1,634 km2) and Black (1,441 km2) which are tributaries to the Salt River in Arizona. In model runs using the SCA and SWE products, PRMS simulated SCA and SWE values were replaced with the SCA and SWE product values each time step the products were available. The simulated energy and mass balance states of PRMS were also adjusted based on the difference between the current model state and the assimilated estimate. The largest differences between PRMS simulations of SCA and SWE, and those estimated in the SCA and SWE products, occurred in the complex, higher elevation terrain. Simulated streamflow using the assimilated products were as much as 50% less than observed streamflow over the eight year period. The largest differences between observed streamflow and that simulated using the assimilated products occurred in the topographically complex Upper Rio Grande. Differences were smaller in the White and Black basins. Use of an averaging filter to smooth the SCA and SWE products prior to assimilation improved simulated streamflow volume, especially for the Upper Rio Grande basin
C31A-0289 0800h
Catchment Scale Study of the Spatial Variability of Snow Depths: Implications for Broader Long Term Snow Monitoring and Measurement Designs
In order to understand how the relation between snow depth and physiographic variables change over the accumulation and ablation periods a pilot study using ten continuous measurements was carried out at a mixed conifer 4-km$^{2}$ snow measurement site (Gin Flat) in the Upper Merced basin, California. Results from this winter/spring 2003/2004 study showed that despite differences in physiographic variables, the statistical distribution of the variability was consistent with that observed at nearby snow courses in the Upper Merced and Tuolumne basins; however the relative amount of snow water equivalent at individual snow-course points, which had similar physiographic characteristics, varied both seasonally and interannually. The results from this first season (2003/2004) provided motivation and guidance for a broader network design over a larger area that samples more physiographic variability. The new installation uses a sensor web to enhance the data recovery and analysis, and explore new methods for measuring seasonal snow cover. The specific aim is to use intensive point measurements to establish how physiographic variables interact to control SWE distribution, and define representative measurement areas based on the important physiographic variables.
C31A-0290 0800h
Mountain Snow System Interactions - An Integrative Approach
Snow scientists now have capabilities and opportunities unimagined in the 1950's due to refinements in field techniques and instrumentation, and the advent of remote sensing platforms. These technical advances enable snow scientists to observe the mountain snow system at virtually any spatial scale. Mountain snow covers are essential water resources in many regions and are increasingly recognized as sensitive bellwethers of global change. Earth system science requires datasets that capture the 'vital signs' of system states and interactions at multiple spatio/temporal scales. Snowmelt processes are influenced by complex interactions that occur over a range of spatial scales. Surface energy exchange states and storage of melt water within the snowpack are expected to dominate snowmelt at the point scale. At larger spatial scales, the influence on lateral movement of water through the snowpack by basin topography and stream network traits may begin to dominate runoff. At still larger scales, reductions in basin- scale snow albedo caused by aerosols or dusts originating from distant sources may become the dominant forcing agent. Models based on an understanding of snowpack processes at the point scale will tend to allow point-scale processes to dominate when integrated to the basin scale. Knowledge of how processes at different scales interact, and which processes dominate at which scales, is essential to the development of new models. Traditional snow observation protocols and existing datasets often fail to capture or represent earth-surface interactions and processes in ways that enhance the integrated investigation of the mountain snow system as a system. The Center for Snow and Avalanche Studies and its collaborators seek to facilitate the interdisciplinary, integrative development of a ?mountain snow system observation protocol? or MSSOP. A multi-modal, multi-scale, integrative MSSOP observation set would identify proxy measures of system behavior for routine and sustained observation at mountain snow system observatories such as the Senator Beck Basin Study Area in the San Juan Mountains of southwest Colorado. An MSSOP would facilitate mountain climate and weather modeling and verification, would support research and applications in global change science and regional resource and hazard management, provide a framework for analyzing, enhancing, and publishing existing snow datasets and observation programs, and represent a basis on which further hydrologic/snow system observatories are grown.
C31A-0291 0800h
Landscape Vegetation Structure and Snow Cover Relationships at Multiple Scales in Glacier National Park, Montana, USA
A two year study was undertaken to examine the relationships between snow cover and landscape vegetation structure across a range of landscape types and climate regimes in Glacier National Park, Montana, USA. Nine snow survey transects, each consisting of 30 survey points spaced at 30 meter intervals, were installed at low, medium, and high elevation areas on both sides of the Continental Divide in Glacier National Park. Results from the first year of snow surveys indicate that, while distinctly different landscape types (e.g. forests and meadows) exhibit substantial differences in snow cover, subtle differences in forest cover density seem to exert only a weak influence on the distribution of snow cover at the plot scale. Survey measurements taken at 1 meter increments along several survey transects indicate that the scale of snow cover variability (quantified by the correlation length) varies between transects and ranges from about 15 meters to over 50 meters. Additional snow surveys designed to investigate the relationship between forest cover density and snow cover distribution at the forest stand scale will be conducted over the winter of 2004-2005. Results from these surveys, along with the results from a second year of snow surveys along the original nine transects, should clarify the effects of landscape vegetation structure on snow cover distribution at multiple scales and across a range of landscape types and climate regimes.
C31A-0292 0800h
Climate Sensitivity to Realistic Solar Heating of Snow and Ice
Snow and ice-covered surfaces are highly reflective and play an integral role in the planetary radiation budget. However, GCMs typically prescribe snow reflection and absorption based on minimal knowledge of snow physical characteristics. We performed climate sensitivity simulations with the NCAR CCSM including a new physically-based multi-layer snow radiative transfer model. The model predicts the effects of vertically resolved heating, absorbing aerosol, and snowpack transparency on snowpack evolution and climate. These processes significantly reduce the model's near-infrared albedo bias over deep snowpacks. While the current CCSM implementation prescribes all solar radiative absorption to occur in the top 2 cm of snow, we estimate that about 65% occurs beneath this level. Accounting for the vertical distribution of snowpack heating and more realistic reflectance significantly alters snowpack depth, surface albedo, and surface air temperature over Northern Hemisphere regions. Implications for the strength of the ice-albedo feedback will be discussed.
C31A-0293 INVITED 0800h
Eddy correlation measurements of sensible and latent heat transfer coefficients over a high altitude glacier snow surface in the Andes Mountains, Bolivia.
An eddy correlation system was installed at 5100 m elevation on the Zongo Glacier in the Andes Mountains near La Paz, Bolivia in order to evaluate the transfer of sensible and latent heat to ablating snow on the glacier surface. Concomitant measurements of all radiation components, snow surface temperature, vertical gradients of air temperature, humidity and wind speed and of surface melt and sublimation were made. The site was approximately level within several hundred metres, with drainage winds at night and dry conditions prevailed in the austral sub-tropical winter. The results show that in the exceeding dry and thin atmosphere, surface temperatures below freezing are maintained throughout melt with significant cooling of the surface due to sublimation and low longwave radiation input. Turbulent transfer to the snow surface was in the rough flow regime with roughness heights similar to that of seasonal snowcovers. Transfer coefficient of sensible heat and water vapour was enhanced several fold with respect to that for momentum diffusivity, despite frequent stable conditions in the lowest few metres. Over mid-continental seasonal snowcovers, turbulent transfer of latent and sensible heat is usually dampened with respect to momentum transfer; the influence of surrounding melting snowcover on large scale air masses and consequent entrainment is often invoked as the cause. However in the Andes, melting snow on glaciers is surrounded by arid high plateau and mountains which were snowfree at the time of observations. Complex flows over the mountains and drainage winds would permit substantial entrainment of dry, warm air. Cool surface temperatures driven by longwave loss ensured strong temperature gradients. The resulting sublimation rates at the surface exceeded 100 W/m2 latent heat flux at times and were matched by opposing sensible heat fluxes, which are amongst the highest reported for surface snow in the world. These aspects of turbulent transfer help to explain the rapid ablation rates that IRD has observed for tropical glaciers and make important limitations on the water resources that can be derived from the same.
C31A-0294 0800h
Long-wave Irradiance Measurement and Modeling during Snowmelt, a Case Study in the Yukon Territory, Canada
At high latitudes, long-wave radiation emitted by the atmosphere and solar radiation can provide similar amounts of energy for snowmelt due to the low solar elevation and the high albedo of snow. This paper investigates temporal and spatial variations of long-wave irradiance at the snow surface in an open sub-Arctic environment. Measurements were conducted in the Wolf Creek Research Basin, Yukon Territory, Canada (60°36'N, 134°57'W) during the springs of 2002, 2003 and 2004. The main causes of temporal variability are air temperature and cloud cover, especially in the beginning of the melting period when the atmosphere is still cold. Spatial variability was investigated through a sensitivity study to sky view factors and to temperatures of surrounding terrain. The formula of Brutsaert gives a useful estimation of the clear-sky irradiance at hourly time steps. Emission by clouds was parameterized at the daily time scale from the atmospheric attenuation of solar radiation. The inclusion of air temperature variability does not much improve the calculation of cloud emission.
C31A-0295 0800h
Evaluation of the NOAA/NWS National Snow Analyses Snow Model
The National Operational Hydrologic Remote Sensing Center (NOHRSC) of NOAA's National Weather Service has developed and operates a high-resolution (1-km2) multi-layer energy- and mass-balance model for snow and soil to estimate snow pack properties throughout the coterminous U.S. (CONUS) in support of operational hydrologic forecasting. In its operational mode, the spatially distributed model is run hourly on a 1-km grid, and all available in situ observations of snow water equivalent and snow depth and satellite-derived snow cover are assimilated daily into the model. To examine the physical performance of the model, it was run in a 1-D mode for the 2002-2003 accumulation and ablation season at five micrometeorological stations established for the NASA Cold Land Processes Experiment (CLPX). Forcing data sets from these stations were used to drive the 1-D model. The output was compared to data from snow pits collected approximately once per month at each site and additional hourly data sets from the met stations. A second 1-D model, SNTHERM, was also run at each site to provide comparative and diagnostic information. Comparison of the NSA model to empirical snow pit data generally shows good monthly-seasonal agreement in snow water equivalent (SWE) and snow depth at each site. This poster focuses on evaluation of major components of the model (albedo, surface temperature, vertical temperature profile, and energy fluxes) by comparison to additional empirical data collected at the met stations and to the SNTHERM model.
http://www.nohrsc.nws.gov
C31A-0296 0800h
Snowfall Retrivals Using a Video Disdrometer
A video disdrometer has been recently developed at NASA/Wallops Flight Facility in an effort to improve surface precipitation measurements. One of the goals of the upcoming Global Precipitation Measurement (GPM) mission is to provide improved satellite-based measurements of snowfall in mid-latitudes. Also, with the planned dual-polarization upgrade of US National Weather Service weather radars, there is potential for significant improvements in radar-based estimates of snowfall. The video disdrometer, referred to as the Rain Imaging System (RIS), was deployed in Eastern North Dakota during the 2003-2004 winter season to measure size distributions, precipitation rate, and density estimates of snowfall. The RIS uses CCD grayscale video camera with a zoom lens to observe hydrometers in a sample volume located 2 meters from end of the lens and approximately 1.5 meters away from an independent light source. The design of the RIS may eliminate sampling errors from wind flow around the instrument. The RIS operated almost continuously in the adverse conditions often observed in the Northern Plains. Preliminary analysis of an extended winter snowstorm has shown encouraging results. The RIS was able to provide crystal habit information, variability of particle size distributions for the lifecycle of the storm, snowfall rates, and estimates of snow density. Comparisons with coincident snow core samples and measurements from the nearby NWS Forecast Office indicate the RIS provides reasonable snowfall measurements. WSR-88D radar observations over the RIS were used to generate a snowfall-reflectivity relationship from the storm. These results along with several other cases will be shown during the presentation.
C31A-0297 0800h
Observation and Modelling of the Thermal Boundary Layer over Snow and soil Patches
The melting of the snow cover, in most cases, is characterized by a transition from complete snow cover, to the appearance of bare soil patches, to the increase in size of the bare patches, and the decrease in size and subsequent disappearance of the snow patches. Thus, for a significant portion of the snowmelt period, the local advection of energy plays an important role, and the correct determination of the melting of a patchy snow cover requires an understanding of this advection process. The energy advected from the exposed bare soil to the snow surface depends on the fetch and size of snow and bare ground patches. Lightweight, portable instrument masts were designed and constructed for the purpose of measuring the growth of the thermal boundary layer over snow patches and bare soil patches within a snow field. The results of a field study are presented. The thermal boundary layer height is shown to be a power function of the fetch distance; the coefficient and exponent are related to the upwind surface roughness. A simple boundary-layer model is developed for calculating the advection of energy from bare ground to melting snow patches. Results from the model are compared with observations of temperature profiles and boundary-layer heights. The model is then used to investigate advection for a wider range of atmospheric stability and surface roughness.
C31A-0298 INVITED 0800h
Variability of the Below Canopy Thermal Structure over Snow
Due to the complexity of energy exchange in forested environments, existing canopy models have difficulty capturing the impact of canopy elements on thermal signatures. Prior research suggests that below canopy solar and thermal fluxes, and forest snow pack variability, is controlled by canopy type and structure. Considerable research has been conducted on the interactions of solar radiation (visible through the near infrared) with canopy elements. This degree of research has not been conducted for the interaction of longwave infrared radiation with canopy elements and the underlying surface. The thermal contribution from the different tree elements varies spatially and temporally due to differential effects of point sources (i.e. the sun) and extended sources (terrain, sky, canopy, and canopy gaps). The objectives of this work are to document the complexity and variability of the thermal signature within a forest canopy, and to make simple calculations to asses the thermal impact of the canopy elements on the energy balance incident on the snow surface. Measurements of the thermal environment beneath forest canopies were made in April 2004 at the Reynolds Creek Experimental Watershed in Idaho as part of a larger effort to characterize the sub-canopy energetics at the snow surface in a sub-watershed complicated by forest cover and variable terrain. We made measurements of the thermal environment using a Flir System ThermaCAM S60 infrared camera (spectral range, 7.5 to 13$\mu$m; thermal sensitivity, $0.08\deg$C). Our measurements captured the spatial and temporal variability of thermal radiation in both conifer and deciduous stands. We used the infrared camera to obtain images of the trees from all cardinal orientations over a 24-hour period, as well as images of forest litter on the snow surface. Surface temperatures of deciduous tree trunks ranged from 2 to $25\deg$C depending on orientation and time of day. Conifer stem temperatures were measured $20\deg$C higher than the surrounding air temperature. The temperature of a single fir cone on the snow surface was measured at $13\deg$C; while small, thin fir needles on the snow surface reached $2\deg$C. A dead tree limb on the snow surface reached an extreme temperature of $37\deg$C. Simple calculations show the contribution of these tree elements and forest litter to the energy balance at the snow surface.
C31A-0299 0800h
Distribution and Ablation Patterns of an Arctic Snowpack in a Windy Environment
Because of topographic variation, most catchments exhibit a degree of unevenness in their snow distribution patterns. For treeless environments in the high Arctic, additional heterogeneity is induced because of wind events. Quantifying this variability at the watershed scale is a challenge. Data has been collected at 80 plus sites over the whole of the north draining Kuparuk River basin on the North Slope of Alaska since 1996. Measurements, 50 depth and 5 snow water equivalent (SWE), have been made at numerous representative land types in/adjacent to the basin. In addition, seven sites on a north-south transect from the headwater foothills to the coast have been monitored to define the pattern of melt. Slope, aspect, elevation, direction of prevailing winds, and vegetation (shrubs, tussocks, grasses, etc.) are all variables that impact snow distribution. The average SWE over the watershed can vary by a factor of two or more between years. Although typically the melt pattern can progress from south (higher elevations) northward towards the coast, the opposite has been observed occasionally. Snowmelt has also started in the middle of the watershed with the melt both progressing southward towards the higher elevations and northward down to sea level. In this study, the patterns of snow distribution and end of winter ablation over eight years are presented.
C31A-0300 0800h
Fractal Dimension as a Measure of Snow Depth Complexity
Snowpack properties vary dramatically over a wide range of spatial scales, from crystal microstructure to regional snow climates. The variability in snow depth in subalpine and alpine environments has importance in understanding and modeling hydrologic, ecologic, and avalanche processes. The driving forces of wind and energy balance interact with topographic and vegetation roughness elements to dominate the observed variability in snow depth at scales from 1 to 1000 meters. Despite this apparent and intuitive relation to topography, efforts to relate topographic parameters to variability in snowpack properties have been largely unsatisfactory. The measure of the fractal dimension of surfaces, i.e. land surface, vegetation, and snow cover, shows promise as a tool for assessing the relative complexity of the interrelated surface morphologies. Fractal dimension is an index of the roughness and self-similarity of an object derived from its scaling properties. This study uses LiDAR-derived land surface elevation, vegetation surface elevation, and snow depth data collected at the Buffalo Pass, Walton Creek, and Alpine Intensive Study Areas as part of the NASA Cold Lands Processes Experiment (CLPX) in April and September, 2003. Fractal dimensions are estimated from the slope of a log-transformed variogram, and demonstrate scale-invariant, fractal behavior over several orders of magnitude in the elevation, vegetation, and snow depth datasets. Directional differences in the snow depth fractal dimension are examined relative to the prevailing wind direction.
C31A-0301 0800h
Local Scale Radiobrightness Modeling During the Intensive Observing Period-4 of the Cold Land Processes Experiment-1
The NASA Cold Land Processes Field Experiment (CLPX-1) was designed to provide microwave remote sensing observations and ground truth for studies of snow and frozen ground remote sensing, particularly issues related to scaling. CLPX-1 was conducted in 2002 and 2003 in Colorado, USA. One of the goals of the experiment was to test the capabilities of microwave emission models at different scales. Initial forward model validation work has concentrated on the Local-Scale Observation Site (LSOS), a 0.8~ha study site consisting of open meadows separated by trees where the most detailed measurements were made of snow depth and temperature, density, and grain size profiles. Results obtained in the case of the 3rd Intensive Observing Period (IOP3) period (February, 2003, dry snow) suggest that a model based on Dense Medium Radiative Transfer (DMRT) theory is able to model the recorded brightness temperatures using snow parameters derived from field measurements. This paper focuses on the ability of forward DMRT modelling, combined with snowpack measurements, to reproduce the radiobrightness signatures observed by the University of Michigan's Truck-Mounted Radiometer System (TMRS) at 19 and 37~GHz during the 4th IOP (IOP4) in March, 2003. Unlike in IOP3, conditions during IOP4 include both wet and dry periods, providing a valuable test of DMRT model performance. In addition, a comparison will be made for the one day of coincident observations by the University of Tokyo's Ground-Based Microwave Radiometer-7 (GBMR-7) and the TMRS. The plot-scale study in this paper establishes a baseline of DMRT performance for later studies at successively larger scales. And these scaling studies will help guide the choice of future snow retrieval algorithms and the design of future Cold Lands observing systems.
C31A-0302 0800h
Evaluation of Two Ultrasonic Snow Depth Sensors for National Weather Service Automated Surface Observation System Sites
In the late 1980's the National Weather Service (NWS) deployed the Automated Surface Observing System (ASOS) at airport observing sites, eliminating the need for human observers. At the time there were no reliable sensors to measure snow depth and the traditional snow measurements of 6 hour snowfall and snow water equivalent (SWE) were abandoned at most locations. The National Weather Service is currently exploring the feasibility of installing ultrasonic snow depth sensors at ASOS sites to restore snowfall measurements to the historic data record. In the 2003-2004 snow season testing of the Judd Communications ultrasonic depth sensor began at three sites: Fort Collins, CO; Stove Prairie, CO; and New Brunswick, OH. Preliminary results show that due to scattering of the sound pulse the Judd sensor performs poorly under windy conditions and when low density snow is present on the snow surface. In addition to the automated data, 6 and 24 hour manual measurements of snowfall, snow depth, snow water equivalent and gauge precipitation were collected. For the 2004-2005 snow season, 15 sites across the U.S. will test both the Judd Communications and the Campbell Scientific sensors. This poster will show three aspects of the project: i) the magnitude and characteristics of noise in the sensor data, ii) an algorithm to convert continuous sensor total snow depth data to the traditional NWS 6 hour snowfall measurements, and iii) a comparison of the performance of the two sensors.
C31A-0303 0800h
A spatially explicit snow model in a mid-latitude alpine basin
The snowpack in an alpine basin in the Sierra Nevada is analyzed with a spatially-explicit snowmelt model. The 1900 hectare basin lies almost entirely above timberline. We run CRREL's point snow model SNTHERM.89 in each spatial element of a 30 m grid. Three field campaigns during the ablation season obtained depth, density, and temperature measurements for estimating the distribution of snow water equivalent and thermal properties of the snow. Nine remotely sensed images of subpixel snow-covered area provide initialization, validation, and re-initialization. SNTHERM.89 uses grain size retrieved by remote sensing to calculate snow albedo. The initial snow water equivalent image was derived from the first gridded field survey of 429 depth and 33 density samples. We distribute meteorological data at hourly time steps from three stations within the basin to the elevation grid based on calculated lapse rates with elevation. Shortwave radiation inputs to the model are calculated with a topographic radiation model and adjusted for cloud cover based on measured solar radiation. Longwave radiation inputs are calculated with a model incorporating air temperature and relative humidity, with corrections for cloud cover. We validate the model's calculation of the snow cover by comparing snow-covered area, snow water equivalent, and grain size with remote sensing data and field measurements, and we also compared the model snowmelt flux with the hydrograph measured at the basin outlet. Model results are sensitive to correct estimation of albedo. Overall the model matches the hydrograph well, except it misses the very beginning of runoff in the stream. This error is perhaps caused by lateral flow within the snowpack.
http://www.snow.ucsb.edu
C31A-0304 0800h
A comparison of physically based and degree-day representations of snowpack / atmosphere turbulent fluxes in an alpine watershed
The degree-day turbulent flux algorithm of the Snowmelt Runoff Model (SRM) was compared with a physically based turbulent flux algorithm to evaluate the trade-offs between model complexity, data input intensity and model accuracy. The two different turbulent flux representations were used in a snowpack mass and energy balance model that explicitly represents net radiation. Snowmelt simulations were coupled with a time-series of remotely sensed snow covered area (SCA) data to simulate daily snowmelt volume in the 19.1-km$^{2}$ Tokopah Basin of the southern Sierra Nevada, California. Modeled daily snowmelt rates for each 30-m pixel were scaled by the SCA and integrated over the snowmelt season to obtain estimates of maximum SWE accumulation. Evaluation of model performance using remotely sensed snow covered area indicated that snow cover depletion rates were reasonably simulated using the physically based model but were overestimated using the degree-day model, with total snow disappearance occurring at least 16 days prior to observed disappearance. Un-calibrated correlations between modeled basin-wide snowmelt volume and observed runoff indicated that the timing of snowmelt was simulated adequately with the degree-day model (R$^{2}$ = 0.59) relative to the physically based model (R$^{2}$ = 0.62). Modeled peak snowmelt occurred 18 days prior to peak observed runoff in both cases. Results from a regression tree and co-kriging interpolation of 397 snow depth observations obtained at maximum SWE were compared with the maximum SWE reconstructions from the snowmelt models. Model SWE reconstruction error was 36 percent for the degree-day model versus 12 percent for the physically based model. Relative to the physically based model, the degree-day model was able to adequately represent the timing of snowmelt with a significantly lower computational and input data requirement.
C31A-0305 0800h
Improved short-term operational streamflow forecasting for snow-melt dominated basins
The Snowmelt Runoff Model (SRM) is a conceptually-based model that uses three commonly available input variables (precipitation, temperature, and percent snow covered area) and the degree-day (temperature-index) concept to simulate and forecast streamflow in snowmelt-dominated basins. This research investigates techniques for adding more physically-based methods to SRM while maintaining its conceptual framework and appeal to operational use. This includes the incorporation of relative humidity and wind speed data to improve model performance during rain-on-snow events and the evaluation of various basin disaggregation schemes to better account for sub-catchment variability. We will demonstrate the use of SNOTEL data to estimate model parameters. Results from running SRM in its traditional, conceptual fashion will be compared to results obtained by running a modified version of the model that utilizes more physically-based parameter estimation techniques.
C31A-0306 0800h
Snow-Elk Interactions In The Northern Elk Winter Range, Yellowstone National Park
Distributed estimates of snowpack properties and the processes that control them are important for gaining a more comprehensive understanding of complex ecosystems. The Northern Elk Winter Range, located in the Greater Yellowstone Ecosystem, is an exemplar of such a complex ecosystem. Ungulates, specifically elk (Cervus Elaphus), are a key biotic component of the Northern Elk Winter Range. Migration, predation, forage utilization strategies and herd distribution of elk are all greatly influenced by the heterogeneous distribution of snowpack properties. This study developed an approach to model the spatial and temporal distribution of snowpack properties for two basins within the Northern Elk Winter Range in an attempt to better understand the energetic expenditures of elk with respect to variable snowpack conditions. Our approach used SNTHERM, a process based energy and mass balance model, to predict snowpack properties for the 2004 winter season. SNTHERM takes initial snowpack conditions and observed meteorological conditions over a given time period, and using mathematical equations based on known physical processes, calculates melt and other snowpack fluxes. We spatially distributed the model by classifying the study area into zones of similar physical characteristics and running the model for each classification region.. Discrete classification regions were derived using combinations of elevation, aspect and landcover type. The performance of the distributed SNTHERM model was tested by comparing the predicted versus measured snow depth, snow density, snow water equivalences, snowpack temperatures, and snowpack stratigraphy. This poster presents the results from these measured versus modeled comparisons and assesses the major sources of uncertainty in the modeling process. Ultimately, this research looks to provide insights into the complexities of snow/elk interactions as well as making a contribution to the development and improvement of spatially distributed snow models within the field of snow hydrology.
C31A-0307 0800h
Snowmelt in a High Latitude Mountain Catchment: Effect of Vegetation Cover and Elevation
The energetics and mass balance of snowpacks in the premelt and melt period were compared from three elevation bands in a high latitude mountain catchment, Wolf Creek Research Basin, Yukon. Elevation is strongly correlated with vegetation cover and in this case the three elevation bands (low, middle, high) correspond to mature spruce forest, dense shrub tundra and sparse tundra (alpine). Measurements of radiation, ground heat flux, snow depth, snowfall, air temperature, wind speed were made on a half-hourly basis at the three elevations for a 10 year period. Sondes provided vertical gradients of air temperature, humidity, wind speed and air pressure. Snow depth and density surveys were conducted monthly. Comparisons of wind speed, air temperature and humidity at three elevations show that the expected elevational gradients in the free atmosphere were slightly enhanced just above the surface canopies, but that the climate at the snow surface was further influenced by complex canopy effects. Premelt snow accumulation was strongly affected by intercepted snow in the forest and blowing snow sublimation in the sparse tundra but not by the small elevational gradients in snowfall. As a result the maximum premelt SWE was found in the mid-elevation shrub tundra and was roughly double that of the sparse tundra or forest. Minimum variability of SWE was observed in the forest and shrub tundra (CV=0.25) while in the sparse tundra variability doubled (CV=0.5). Snowmelt was influenced by differences in premelt accumulation as well as differences in the net energy fluxes to snow. Elevation had a strong effect on the initiation of melt with the forest melt starting on average 16 days before the shrub tundra and 19 days before the sparse tundra. Mean melt rates showed a maximum in middle elevations and increased from 860 kJ/day in the forest to 1460 kJ/day in the sparse tundra and 2730 kJ/day in the shrub tundra. The forest canopy reduced melt while the shrub canopy enhanced it relative to the sparsely vegetated tundra. Duration of melt was similar in the forest and shrub tundra at 20 days while the sparse tundra was shorter at 13 days; the differences due to differing snow accumulation and melt rates. The greatest variability in the timing and rate of melt was found in the shrub tundra, where the effect of the shrub canopy over snow depends on snow depth and insolation and is reduced in years with high snow accumulation or extensive cloudy periods in spring. The results show that it is necessary to consider the combination of elevation and vegetation effects on snow microclimate and melt processes in high latitude mountain catchments, but that weather patterns induce substantial variability on the effect these factors.
C31A-0308 0800h
Air Temperature Lapse Rate Dynamics in a Snow-Dominated Mountainous Watershed
Spatial variation in snowpack processes owing to topographic effects is complicated by air temperature inversions that may be an important factor in determining the timing and rate of snowmelt. This process is of great interest in mountainous regions where snowmelt is one of the largest surface water inputs controlling runoff. Localized temperature inversions, coupled with topographic shading of solar radiation and windspeed reductions can produce ablation patterns that proceed from ridgetops to valley bottoms in some catchments. To develop a better understanding of this process, 38 temperature data loggers were installed in transect spanning three second-order catchments in the Mica Creek Experimental Watershed (MCEW) in northern Idaho. Results indicate that strong temperature inversions occur from low to upper mid slopes during daylight hours. Between the months of November 2003 and April 2004, inversions typically spanned the lower 140 vertical meters. These inversions resulted in lapse rates of 54.6, 12.8, and 11.4 C/km at 12:00, 15:00, and 18:00hrs respectively on sunny days, and lapse rates of 19.9, 6.4, and 2.1 C/km at 12:00, 15:00, and 18:00hrs respectively on cloudy days. At approximately 140 m the temperature pattern began to revert back to normal lapse rates. At this scale (i.e. less than 1km), the observed lapse rates could lead to highly variable impacts on snowmelt due to the dramatic deviation from average environmental lapse rates of approximately -5 to -6 C/km. This work has shown that the classic assumption of normal lapse prevailing in mountainous catchments can be incorrect. The violation of the standard assumption that air temperature decreases with altitude holds important implications for development of distributed climate surfaces to drive distributed snowmelt models.
C31A-0309 0800h
Variations in Below Canopy Turbulent Flux From Snow in North American Mountain Environments
Sensible and latent heat and mass fluxes from the snow surface are modulated by site canopy density and structure. Forest and shrub canopies reduce wind speeds and alter the radiation and thermal environment which will alter the below canopy energetics that control the magnitude of turbulent fluxes between the snow surface and the atmosphere. In this study eddy covariance (EC) systems were located in three experimental catchments along a mountain transect through the North American Cordillera. Within each catchment, a variety of sites representing the local range of climate, weather, and canopy conditions were selected for measurement of sensible and latent heat and mass flux from the snow surface. EC measurements were made 1) below a uniform pine canopy (2745m) in the Fraser Experimental Forest in Colorado from February through June melt-out in 2003; 2) at an open, unforested site (2100m), and below an Aspen canopy (2055m) within a small headwater catchment in the Reynolds Creek Experimental Watershed, Owyhee Mts., Idaho from October, 2003, through June melt-out, 2004; and 3) at five sites, representing a range of conditions: a) below a dense spruce forest (750m); b) a north-facing shrub-tundra slope (1383m); c) a south-facing shrub-tundra slope; d) the valley bottom between b) and c) (1363m); and e) a tundra site (1402m) in the Wolf Creek Research Basin (WCRB) in the Yukon, Canada during the 2001 and 2002 snow seasons. Summary data from all sites are presented and compared including the relative significance of sublimation losses at each site, the importance of interception losses to the snowcover mass balance, and the occurrence of condensation events. Site and weather conditions that inhibit or enhance flux from the snow surface are discussed. This research will improve snow modeling by allowing better representation of turbulent fluxes from snow in forested regions, and improved simulation of the snowcover mass balance over low deposition, high latitude sites such as WCRB, and during drought conditions at mid-latitude sites such as Fraser, Colorado, and RCEW in Idaho.
C31A-0310 0800h
A Water Balance Approach to Characterizing the Hydroclimatology of a Mountainous Semi-arid Catchment
A long-term water balance is needed to understand the hydrology of mountainous semi-arid catchments, which exhibit considerable interannual variability in precipitation and temperature as well as spatial variation in snow accumulation, soils, and vegetation. Long-term data sets reduce the uncertainty associated with estimating water balance quantities that are difficult to measure in practice. In this study, the data required to compute a long-term water balance are assembled from on-site and nearby locations to create a continuous 21-year hourly record of precipitation, meteorological parameters, and streamflow for the Upper Sheep Creek (USC) catchment, a 26 ha, snow-fed, semi-arid rangeland headwater drainage within the Reynolds Creek Experimental Watershed in southwestern Idaho, USA. This study will allow us to extend a previous 10-year water balance (water years 1985-1994) computed for the USC catchment, enabling a more thorough consideration of climate variability including periods of drought and flood. It also sets the stage for analyzing the hydrologic response of the USC catchment to a prescribed fire planned for 2006. Statistical correlations between on-site and nearby meteorological stations were used to develop a complete 21-year hourly data set (water years 1984-2004) of climate and precipitation records. These data will be used to drive the Simultaneous Heat and Water (SHAW) model to simulate evaporation and transpiration, precipitation, storage, and stream discharge. Water balance quantities will be computed for separate landscape units and then aggregated for the overall watershed. This research will improve our ability to manage water resources in semi-arid mountain regions.
C31A-0311 0800h
Snow and climate change research in two mountain catchments, Glacier National Park, Montana
Snow is a dominant driver of many mountain ecosystem processes, determining forest growth, basin hydrology and glacier dynamics in many temperate mountain ranges. A 14-year research program in the Northern Rocky Mountains, USA, has focused on how mountain ecosystem processes in Glacier National Park, Montana have responded to changes in climate patterns. Modeling and monitoring the spatial and temporal variability in seasonal snowpacks has been focused on two catchments, each approximately 450km2. We have conducted intensive snow surveys across these two catchments since 1993 and have over 10,000 monthly measurements of snow water equivalence (SWE) and snow depths. Because some of the oldest Natural Resource Conservation Service (NRCS) snow courses (dating back to 1922) and SNOTEL sites (dating back to 1969) are located in our catchments, our spatially extensive snow datasets complement the single site snow records from SNOTEL installations and NRCS snow courses. Snow chemistry has been investigated for five seasons at multiple sites and elevations to determine spatial variability of major ion concentrations and loadings in addition to a 10-year single site record to track interannual variability. Backcountry stream gages and water chemistry sampling sites monitor snowmelt in both catchments to augment long-term historical records from USGS gages. Snow avalanche research has provided several hundred snowpack profiles, a history of avalanche magnitude and frequency, and an ongoing forecasting program for natural avalanche releases. Glaciers and perennial snowfields have been mapped, monitored and modeled in the catchments and four high-elevation backcountry weather stations have provided climatic data for up to 10 years. The integrated research program has produced or modified models of snow distribution, snowpack seasonal evolution, decadal-scale variability, and catchment-scale water balance. These snow oriented products are scaled up to mountain ecosystem models that address impacts of climatic change for western mountain regions.
http://www.nrmsc.usgs.gov
C31A-0312 0800h
On-Board Cryospheric Change Detection By The Autonomous Sciencecraft Experiment
The Autonomous Sciencecraft Experiment (ASE) is operating on-board Earth Observing - 1 (EO-1) with the Hyperion hyper-spectral visible/near-IR spectrometer. ASE science activities include autonomous monitoring of cryopsheric changes, triggering the collection of additional data when change is detected and filtering of null data such as no change or cloud cover. This would have application to the study of cryospheres on Earth, Mars and the icy moons of the outer solar system. A cryosphere classification algorithm, in combination with a previously developed cloud algorithm [1] has been tested on-board ten times from March through August 2004. The cloud algorithm correctly screened out three scenes with total cloud cover, while the cryosphere algorithm detected alpine snow cover in the Rocky Mountains, lake thaw near Madison, Wisconsin, and the presence and subsequent break-up of sea ice in the Barrow Strait of the Canadian Arctic. Hyperion has 220 bands ranging from 400 to 2400 nm, with a spatial resolution of 30 m/pixel and a spectral resolution of 10 nm. Limited on-board memory and processing speed imposed the constraint that only partially processed Level 0.5 data with dark image subtraction and gain factors applied, but not full radiometric calibration. In addition, a maximum of 12 bands could be used for any stacked sequence of algorithms run for a scene on-board. The cryosphere algorithm was developed to classify snow, water, ice and land, using six Hyperion bands at 427, 559, 661, 864, 1245 and 1649 nm. Of these, only 427 nm does overlap with the cloud algorithm. The cloud algorithm was developed with Level 1 data, which introduces complications because of the incomplete calibration of SWIR in Level 0.5 data, including a high level of noise in the 1377 nm band used by the cloud algorithm. Development of a more robust cryosphere classifier, including cloud classification specifically adapted to Level 0.5, is in progress for deployment on EO-1 as part of continued ASE operations. [1] Griffin, M.K. et al., Cloud Cover Detection Algorithm For EO-1 Hyperion Imagery, SPIE 17, 2003.