The integration of in-situ snow cover measurements with multi-scale ground based and airborne passive microwave data in the Canadian tundra
Tundra snow cover is an extremely important cryospheric parameter to understand and monitor as it influences local, regional, and global scale surface water balance, energy fluxes, and ecosystem and permafrost dynamics. Seasonal snow accumulation patterns are controlled at the regional scale by differences in precipitation, but influenced considerably at the local scale by interaction with vegetation and terrain driven by wind redistribution. This variability, coupled with the sparse network of conventional observing stations across tundra regions, mean current tundra snow cover data sets are highly uncertain. Spaceborne passive microwave sensors offer promise to provide operational snow cover data over tundra landscapes, however, problems retrieving snow water equivalent (SWE) and snow depth exist due to the influence of sub-grid scale snowpack and terrain variability on microwave emission. To address these scaling issues, data were collected in the Upper-Coppermine River Basin in the NWT, Canada. Six years (2003 to 2008) of spatially intensive in- situ snow measurements were obtained over the study area along with two airborne passive microwave radiometer campaigns (2005 and 2008) and the deployment of ground based passive microwave radiometers in 2007. The aim of these in-situ campaigns was to achieve a more complete understanding of late winter, pre- melt tundra snow cover properties and distribution across a large scale basin for the development and validation of snow cover models and satellite passive microwave algorithms. Analysis of multiscale airborne brightness temperatures and ground observations show that it is possible to retrieve SWE at homogenous terrestrial sites using high resolution ground based (2 m) and airborne (50 m) passive microwave measurements. However, estimates of SWE are complicated when frozen lakes and complex terrain are integrated into satellite footprints with 25 km grid cell dimensions. Nonetheless, there are systematic relationships between sub-grid lake cover, terrain complexity, and microwave emission that can be exploited at the satellite scale. The results from this analysis have allowed for a better understanding of how multiple scales of passive microwave observations and in-situ data can be utilized to improve operational SWE algorithm development.
Evaluating Terrestrial Snow Cover in the Canadian Regional Climate Model
The Canadian Regional Climate Model has been coupled with the latest version of the Canadian Land Surface Scheme (CLASS 3.4). The new scheme includes a number of improvements in snow process algorithms, affecting the treatment of snow density, canopy interception and unloading, and turbulent exchange with the atmosphere. The impact of these improvements on the regional climate is examined in a 10 year simulation (Apr. 1997 to Dec. 2007) over western Canada. The surface air temperature and precipitation from this simulation are evaluated against a station based gridded monthly observed surface climate dataset produced by Environment Canada. The modeled snow water equivalent and snow cover fraction are compared with observations derived from both satellite passive microwave data and Canadian Meteorological Centre objective analysis over three contrasting surface types: cropland, boreal forest and sub-arctic tundra. Both improvements and ongoing issues in the new coupled modeling system will be discussed.
Topographic and Hydrometeorological Controls of Remotely-Sensed Snow Cover Distribution in the Quesnel River Basin of British Columbia, Canada
This presentation will focus on the topographic and hydrometeorological controls of remotely-sensed snow cover distribution in the alpine Quesnel River Basin (QRB) of western Canada. The MODIS 8-day maximum snow cover extent products (MOD10A2) from 2000-2007 are first filtered to reduce cloud coverage over the period 2000-2007 and evaluated with ground-based snow measurements. The resulting snow cover data are used to evaluate the evolution of snow cover duration (SCD) and snow cover fraction (SCF) in the QRB where elevations range from about 500 m to 3000 m above sea level. Elevation, slope, and aspect greatly influence the distribution and duration of snow cover in the watershed. For instance, mean gradients of SCD with elevation are 3.8, 4.3, and 11.6 days (100 m)-1 for the snow onset season, snowmelt season, and entire year, respectively. The gradient of SCF with elevation (d(SCF)/dz) during the snowmelt season is 8% (100 m)-1. The average ablation rates of SCF are similar for different 100 m elevation bands at about 5.5% (8 days)-1 for altitudes < 1500 m with decreasing values with elevation to near 0% (8 days)-1 for altitudes > 2500 m where perennial snow and glaciers dominate the landscape. Further investigations with regional climatic data show that a 1°C rise in average air temperature during spring leads to a 10-day advance in reaching the 50% SCF threshold in the QRB. Propagation of the snowmelt signal to the Quesnel River is then tracked by correlating the passage of the 50% SCF threshold with the timing of the center of volume in spring runoff. This study thus demonstrates the potential usefulness of applying near-real time MODIS snow cover data, in conjunction with station-based meteorological data, to track snowmelt and the timing and intensity of the spring freshet in alpine watersheds of western Canada such as the QRB.
Radiative Forcing of Dust in Mountain Snow from MODIS surface reflectance data
Here I present an algorithm that retrieves the radiative forcing by desert dust in mountain snow cover from surface reflectance data from NASA Moderate Resolution Imaging Spectroradiometer (MODIS). Dust emitted from natural and disturbed lands frequently deposits to mountain snow cover through dry and wet deposition, particularly in spring when synoptic scale storms entrain material from recently dried surfaces. Dust decreases snow spectral albedo, primarily in the visible wavelengths where the imaginary parts of the complex refractive indices of dust and ice have the greatest contrast. This surface radiative forcing accelerates melt and contributes to the snow-albedo feedback. In the Rocky Mountains of Colorado, this has been shown to shorten the duration of snow cover by approximately a month. The algorithm presented here, MODIS Dust Radiative Forcing in Snow (MOD-DRFS), determines the per pixel radiative forcing by dust in snow from a coupled radiative transfer model that infers the reflectance difference between clean snow spectra and dust- laden snow spectra according to a grain size matching in the near infrared and shortwave infrared wavelengths that are not affected by dust absorption. The spectral residuals are splined to a high spectral resolution and convolved with the at surface spectral irradiance modulated by local topography, and subsequently integrated to the instantaneous surface radiative forcing. I demonstrate the model with retrievals in the Zagros Mountains, Iran and the San Juan Mountains, Colorado, USA. Preliminary validation of the model with in situ detailed pyranometer measurements in the San Juan Mountains indicates that the model has uncertainties of < 7 W/m2.
Acoustic Observation of Snowpack Physical Properties
Recent research has demonstrated the ability to determine Snow Water Equivalent (SWE) by the digital signal processing of an audible sound wave produced by a loudspeaker situated at nadir to the snow surface. Sound waves reflected by the snowpack are recorded by a sensor situated above the snow surface at an offset distance from the source. To determine the impulse response of a snowpack, two types of source signals were considered. Frequency-swept sound pulses were sent into the snowpack, and digitally homodyned with the source signal to determine reflection events. The reflections were related to SWE by a recursive algorithm. Broadband acoustic pulses comprised of Maximum Length Sequences (MLS) were also used as source signals. Processing of the reflected MLS sound wave with the Fast Hadamard Transform (FHT) allowed for determination of the snowpack impulse response, from which the attenuation coefficient spectrum and the SWE were determined. The accuracy of SWE determination by these two methods was assessed by comparison to snowpit measurements of depth and density. The possibility of measuring the structural and related thermal conduction properties of snow from the impulse response was also considered. Application of a Gabor filter suggests that the speed of the sound wave in the pore spaces of the snowpack is related to the frequencies present in the broadband MLS source spectrum. From this speed-frequency relationship, the basic parameters of the Biot model of sound propagation through porous media were calculated. By using a mixture theory approach and coupling the Biot model with the Jackson-Black theory, structural metrics and thermal conductivity were estimated and compared to thermal and structural measurements in the field. These non-invasive measurements of snow physical properties could be potentially used as inputs for models of snowpack evolution. Further aspects of measurement system implementation and limitations of the models will also be discussed.
Cryospheric Dynamics in the Central Chilean Andes: Multi-decadal Reconstruction and Multi-annual Monitoring of Rock Glaciers and a Debris-covered Glacier
In the semiarid Central Chilean Andes at 33.5°S, permafrost is widely present above 3500-4000 m a.s.l., especially in the form of ice-rich debris accumulations such as rock glaciers. While Chilean rock glacier are among the largest known rock glaciers, glaciers are mostly restricted to the highest summits and are affected by a significant retreat during the last decades. Rock glaciers display a characteristic morphology produced by the deep-seated steady-state creeping of their permafrost body, and they are of great geomorphic and hydrological importance in poorly glaciated mountains. Their response to Global Warming is however still poorly understood. Our study area in the Laguna Negra catchment provides up to two-thirds of the drinking water supplies to Chile's capital Santiago (5.5 million inhabitants) during the dry summer months. We investigate the multi-decadal and multi-annual dynamics of a complex cryospheric system composed of a rock glacier and debris-covered glacier. We use aerial photogrammetry to produce diachronic orthophotos and high-resolution digital elevation models of 1956 and 1996, and to detect changes in cryospheric features such as glacier retreat, thermokarst development and rock glacier creep. Surface dynamics on rock glacier transects are furthermore analyzed using differential GPS measurements (2004-2008), indicating horizontal surface displacement rates between 20 and 130 cm/yr. Ground temperatures monitored in 2003/04 with miniature data loggers between 2500 and 4000 m a.s.l. in combination with regional meteorological measurements provide further insight into periglacial process environments in the Andes of Santiago, including information on frost penetration depth, influence of snow cover, and altitudinal gradients. Ongoing research in the Laguna Negra area include a spatially distributed ground thermal monitoring program and, for the first time in the Andes, terrestrial laser scanning of rock glaciers. The cryospheric monitoring network at Laguna Negra constitutes one of the first in the Andes. Ground temperatures are monitored with 45 two-channel data loggers placed in a wide range of topoclimatic and geomorphic contexts in order to statistically assess the influence of relevant parameters (snow depth, solar radiation, surface type, etc.) on the ground thermal regime. Geodetic networks on rock glaciers and the debris-covered glacier have also been enhanced. GPS and laser surveys will be repeated in annual to multi-annual intervals.
Downscaling Temperature and Precipitation from NARR Reanalysis for Glacier Modelling
The North American Regional Reanalysis (NARR) provides a high-quality retrospective on climate in the Northwest hemisphere by providing atmospheric variables at 3 hour intervals on a roughly 32 km grid spacing with 29 vertical levels. Despite this fairly high spatial resolution, the NARR data are too-widely spaced to directly drive spatially distributed alpine glacier models. We present techniques for downscaling the two variables of primary interest for glacier mass balance - temperature and precipitation. We then apply those data to a glacier surface mass balance model for the North American Cordillera. For precipitation, we apply a linear model of orographic precipitation with modifications for air mass tracking and dynamic calculation of nucleation and fallout timescales. Our temperature downscale calculates free-air and inversion lapse rates and uses these to adjust mid-tropospheric temperature to ground-level on the high- resolution target DEM. The regions of focus for our study are southern British Columbia and northern Washington state where observational station density is high and validation of our downscale is possible. Our downscaling efforts produce a high resolution 30 year climatology of temperature and precipitation for the years 1979 through 2008. We have validated our downscaling against Environment Canada's observational data, British Columbia and NRCS SNOTEL snow pillow data for validating precipitation and temeprature in mountainous terrain. We also compare our results to the PRISM gridded precipitation and temperature products. Our downscaling shows good agreement with annual precipitation amounts and mean annual temperatures and captures shorter time-scale events as well. Downscaled temperature shows best performance in summer likely due to simpler boundary layer structure than winter in which cold air pooling can predominate. The physics-based precipitation downscale significantly improves precipitation fields in high altitude, mountainous regions relative to the raw NARR data. Our temperature downscaling yields similar performance to the PRISM product but at a higher spatial and temporal resolution. Using the downscaled temperature and precipitation yields seasonal specific glacier mass balance with RMSE of 0.70 m w. eq. for summer balance and 0.6 m w. eq. for winter mass balance when compared to measurements made on glaciers throughout our domain.
Regional Glaciation Modeling With Debris Transport
When mountain glaciers waste away they commonly become shrouded with surficial debris and render themselves less vulnerable to subsequent ice loss. Regional glaciation models that aim to predict the rate of disappearance of Earth's glaciers in response to climate warming should therefore include this process. As part of a large-scale study of climate change impacts on glaciers in northwestern North America we have added this feature to a two-dimensional (vertically-integrated) time-evolving regional glaciation model. The model solves conservation equations for englacial and supraglacial debris and simulates the exchange from englacial to supraglacial debris which occurs in response to surface melting. An increasing thickness of supraglacial debris insulates the glacier from summer warmth and reduces the rate of surface melt. We illustrate model performance by applying it to synthetic examples and to the real-world problem of predicting changes in the volume and extent of glaciers in the region of Silverthrone and Klinaklini Glaciers in the Coast Mountains of southwestern British Columbia, Canada.