HR: 16:40h
AN: H54B-02    [Abstracts]
TI: The effect of model scale in reconstructing snow water equivalent over complex terrain
AU: * Molotch, N P
EM: molotch@seas.ucla.edu
AF: Department of Civil and Environmental Engineering, Univeristy of California, Los Angeles, 5732 Boelter Hall, Los Angeles, CA 90095, United States
AU: Durand, M
EM: durand@seas.ucla.edu
AF: Department of Civil and Environmental Engineering, Univeristy of California, Los Angeles, 5732 Boelter Hall, Los Angeles, CA 90095, United States
AU: Margulis, S A
EM: margulis@seas.ucla.edu
AF: Department of Civil and Environmental Engineering, Univeristy of California, Los Angeles, 5732 Boelter Hall, Los Angeles, CA 90095, United States
AB: Improving estimates of water and energy fluxes in the rugged landscapes of the American Cordillera is particularly challenging given the considerable topographic heterogeneity and paucity of observations within the region. In these complex systems, parameterizations of sub-grid variability in energy and mass transformations are highly sensitive to relationships between model spatial resolution and the correlation-length scale of the variable of interest - which is usually unknown. In the mountainous regions of the Western United States, the processes controlling the distribution of snow water equivalent are likely better known than any other hydrologic state. The observation record is relatively long (dating back to the early part of the 20th century) and snow cover is relatively easy to detect using remotely sensed observations in the visible and near infrared. Hence, distributed snowpack simulations provide an ideal case study for exploring relationships between process, model, and observation scales; potentially guiding efforts regarding other hydrologic states (e.g. soil moisture). To that end, this research uses a time series of fractional snow covered area (SCA) estimates from Landsat Enhanced Thematic Mapper (ETM+), Moderate Resolution Imaging Spectoradiometer (MODIS), and Advanced Very High Resolution Radiometer (AVHRR) data, in combination with a spatially distributed snowmelt model, to reconstruct snow water equivalent (SWE) in the Rio Grande headwaters (3,419 km²) of Colorado, USA. In this reconstruction approach, modeled snowmelt over each pixel is integrated over the time of satellite observed snow cover to estimate SWE. Despite the considerable differences in the magnitude of SWE in 2001 versus 2002, model performance - using ETM+ data aggregated to 100-m resolution - was robust with a mean absolute error (MAE) of 23% relative to observed SWE from intensive field campaigns. Model performance deteriorated when MODIS (MAE = 57%) and AVHRR (MAE = 90%) data were used and when simulations were run at coarser resolutions; MAE = 26, 34, and 47%, for ETM+ simulations run at 250-m, 500-m, and 1-km resolution, respectively. Basin-average maximum SWE using MODIS and AVHRR was 27% and 57% lower than ETM+ estimates, respectively. Maximum SWE decreased by 28% when ETM+ simulations were run at 1-km versus 100-m resolution. This research illustrates the utility and scale-dependent limitations of the reconstruction technique for obtaining SWE estimates at larger scales (e.g. > 1000 km²) and in locations where detailed hydrometeorological observations are scarce.
UR: http:cee.ucla.edu/faculty/molotch.htm
DE: 1847 Modeling
DE: 1855 Remote sensing (1640)
DE: 1863 Snow and ice (0736, 0738, 0776, 1827)
DE: 1880 Water management (6334)
DE: 1884 Water supply
SC: Hydrology [H]
MN: 2007 Joint Assembly