Climate in the semi-arid Western U.S. exhibits considerable inter-annual variability; and the temporal and spatial distributions of precipitation, snowmelt, soil moisture, evapotranspiration and other hydrologic processes are sensitive to this variability. Sensitivity to climate change varies across gradients of physiography (e.g. elevation, vegetative community structure, and latitude) but the drivers and degree of this sensitivity in different mountainous landscapes are not fully comprehended. Similarly, the impact of these changes on basin-scale snowpack water storage cannot be determined because observations are not distributed across a range of elevations and other physiographic conditions that control snow distribution. As a result, statistical interpolation models of these scarce observations inadequately represent spatial patterns of snow accumulation. For nearly three decades, remotely sensed observations of snow cover depletion have been used to forecast seasonal snowmelt runoff and indirectly seasonal snow accumulation integrated over a watershed. The use of these data to reconstruct snow accumulation is based on the simple concept that deeper snow takes more time (or energy) to melt than shallower snow. More recently advances in remote sensing have enabled sub-pixel detection of snow cover depletion and the development of pixel-specific snow accumulation reconstruction models. The scarcity of ground-based observations needed to evaluate model performance and refine algorithms has restricted reconstruction modeling studies to small headwater catchments. Similarly, reconstruction techniques have not been used to resolve the temporal variability in snow distribution during the accumulation season. To address these inadequacies the proposed research will synergistically develop new observing and modeling systems for estimating the spatial distribution of snow accumulation. The proposed work will address questions related to temporal and spatial variability in snow distribution patterns in the central Sierra Nevada Mountains. A new method of snowfall estimation will be developed in which an Ensemble Kalman Smoother will be used to assimilate new remotely sensed snow measurement capabilities into physically based mass and energy balance models. Densely distributed clusters of ultrasonic snow depth sensors spanning the elevational gradients of the seasonally snow covered portions of the Sierra Nevada will be used to develop this new technique by accounting for both sub-grid variability and the spatial representativeness of the ground observations. The combination of these new modeling and measurement capabilities will enable new understanding of snow accumulation processes and have the potential to fundamentally change the way in which point observations are distributed over rugged terrain.
The objectives of this project was to develop new methods of snowfall estimation using remotely sensed snow measurement and physically based mass and energy balance models. To meet this objective three research questions were explored: 1) Do snow cover depletion patterns observed from satellite during the snowmelt season contain information with respect to the spatial distribution of snow accumulation during the winter season?; 2) To what extent can stratified in situ snow measurements and aforementioned remotely sensed data improve basin-scale estimates of snowfall relative to current operational techniques?; And 3) How does local-scale topography and landcover influence snow distribution and total water storage in the mountain snowpack? How consistent are local-scale snow distribution patterns from storm to storm throughout the accumulation season? With regard to local scale snow distribution, estimates of energy from the sun, derived from up-looking wide angle photographs, were used to explain the spatial distribution of snowmelt rates, explaining up to 58% of snowmelt rates. Using these solar radiation estimates as input to snowmelt models significantly improved model accuracy with an average normalized snow depth error of -0.2% and an error in the date of simulated snow disappearance of only 7.5 days. At the larger scale we combined satellite data and distributed models to simulate the distribution of snow water equivalent and determine the meteorological controls on snow distribution. Comparing these snow estimates versus April – July streamflow, the results indicate that our snow estimates correlate more strongly with streamflow than snow estimates developed by the National Oceanic and Atmospheric Administration or observed snowpack from snow stations. These results have far reaching implications for understanding the processes that control water fluxes in mountain systems. With the snow data described above we have explored the role of large scale atmospheric rivers (AR’s) in controlling this distribution. Atmospheric rivers were found to significantly contribute to the seasonal snow accumulation and its interannual variations. Our analyses of the processes leading to AR development indicate that AR activity, as measured by the number of high-impact ARs and snow accumulation, is significantly augmented when the Maden-Julian-Oscillation (MJO) is active over the far western tropical Pacific. The development of improved snow distribution information and relations to large scale controls on that distribution has broad implications regarding the sensitivity of mountain snow distribution to changes in ocean-atmosphere interactions associated with climate change.
