Bureau of Reclamation : Research and Development | Research and Development Office

Enhancing Predictions of Climate Change Impacts on Snow Distribution and Melt Patterns in the Mountain West

Project ID: 8106
Principal Investigator: Eric Rothwell
Research Topic: Water Operation Models and Decision Support Systems
Priority Area Assignments: 2014 (Climate Change and Variability Research), 2015 (Climate Change and Variability Research), 2016 (Climate Change and Variability Research)
Funded Fiscal Years: 2014, 2015 and 2016
Keywords: None

Research Question

1. How are VIC model outcomes related to process parameterizations and calibrations, scale, and model structure?
a)What improvements can be obtained with fine-scaled depictions with greater physical representation of processes?
b)What improvements can be obtained with more rigorous calibration?
c)How does model scale impact large-scale models of snow accumulation and melt in mountainous environments?

2. How can future projections of the effects of climate change on streamflow patterns in the Western U.S. be improved?
a)Can the VIC/DHSVM snow module be improved, particularly in the intermountain West?
b)Can a fine-scaled model's explicit representations of snow processes better inform and define the implicit sub-grid representations and distributions in large-scaled models?

Need and Benefit

The SECURE Water Act authorized the Bureau of Reclamation to assess climate change risks in eight major western U.S. river basins. Reclamation is accomplishing these tasks within its WaterSMART Basin Study Program which includes the West-Wide Climate Risk Assessments (WWCRA). Initial WWCRA projections for these eight river basins were based on an ensemble of 112 VIC hydrologic model runs forced with statistically downscaled meteorological data from a variety of global climate models covering three different CMIP3 carbon emission scenarios. In order to meet these vast modeling objectives, computational efficiency was a necessity. The complex processes governing snowmelt and runoff had simplified model representations often requiring the calibration of model parameters and modeling was conducted on a 1/8° spatial grid (~12km x 12km). The heterogeneity of mountain snowcover mass and energy fluxes however occurs at much finer scales (Bales et al., 2006; and others). These vital differences – not captured in the VIC simulations – can strongly affect streamflow timing and magnitudes (Luce et al., 1998; and others).
Reclamation's Technical Memorandum # 86-68210-2011-01 evaluating the VIC future scenarios concluded that calibration issues were a concern requiring refinements to the VIC applications and/or the introduction of more appropriate hydrologic models in future assessments. High biases in VIC-simulated flows in smaller sub-catchments were also observed. VIC-modeled streamflow in snow-dominated catchments – even at 1/16° resolution – was biased early by about 12 days in the Pacific Northwest (Wenger et al; 2010a) and one month early in Wyoming (Wenger et al., 2010b). Winstral (2012) has shown that model scale alone can produce an early bias in simulated snowmelt rates in regions like the Western U.S. where preferential snow accumulation typically occurs on solar-shaded slopes.
In order to best predict long-range climate effects on water supplies, it is imperative that we better understand the strengths, weaknesses, sensitivities, and biases of selected models. This research will be focused on the Boise River Basin (BRB). The BRB has a diverse forest cover ranging from heavily forested evergreens to non-forested alpine and sagebrush environments. Basin area is typical of the smaller sub-catchments the VIC model has struggled in. We will test and compare several hydrologic (e.g. VIC, DHSVM) and snow models (e.g.DHSVM/VIC, Isnobal) operated at range of scales from 100 meters up to 1/8°. Initial results from S&T Project #2264, "Application of a Physically-Based Distributed Snowmelt Model in Support of Reservoir Operations and Water Management", have shown that Isnobal, a high-resolution mass and energy balance snow model, can be successfully applied to the BRB. The fine-scaled, detailed models will: a) shed light on large-scale model tendencies including how scale and complexity affects model outcomes and b) provide information and guidance on vital sub-grid heterogeneity that can be translated to the larger scale models.
We have been collaborating with NCAR researchers – the same ones working on S&T Project #1646, "Sensitivity of Hydrologic Impacts to Downscaling Methodology and Spatial Resolution" – on improving sub-grid representations of SWE in large-scaled models. We hope to continue this productive relationship to enhance both their S&T project and the proposed research. Our work will be conducted at much finer scales that can further inform their downscaling methods. Their research on hydrologic models and scaling can guide ours. Together we can test findings across a range of scales and watersheds. At the conclusion of these two studies our understanding of large-scale hydrologic models in snow-dominated regions and our capacity to model long-term climate change effects on snow accumulation, melt, and streamflow will be greatly improved.

Contributing Partners

Contact the Principal Investigator for information about partners.

Research Products

Bureau of Reclamation Review

The following documents were reviewed by experts in fields relating to this project's study and findings. The results were determined to be achieved using valid means.

Enhancing Predictions of Climate Change Impacts on Snow Distribution and Melt Patterns in the Mountain West (final, PDF, 1.0MB)
By Eric Rothwell, Scott Havens, Daniel Marks
Publication completed on January 31, 2017

Results from models run at different scales, from the regional down to the hill slope scale, are dependent on the process parameterizations and calibrations of the physics being represented. This project compared the application of two models that could be used for climate change predictions of snowmelt, a physically based gridded snowmelt model and a coarse lumped parameter model, over the Boise River Basin. The results showed that, while the calibrated parameter model could estimate the streamflow volumes at a yearly and monthly time scale, the volume of water stored in the snowpack was underestimated. Whereas the physically based snow model could accurately estimate the snowpack accumulation and ablation when compared to the in-situ measurement sites. When addressing climate change impacts on the prediction of snow distribution and melt patterns, USDA ARS recommends to use physically based models that have limited calibration and represent the underlying physical processes of the hydrologic system. While lumped parameter models can be quickly developed, robust calibration must be performed to ensure that the resulting parameters represent the physical processes of the system. If the parameters are not representative, then the climate change predictions will have a high level of uncertainty.

Enhancing Predictions of Climate Change Impacts on Snow Distribution and Melt Patterns in the Mountain West (final, PDF, 1.0MB)
By Scott Havens, Daniel Marks, Eric Rothwell
Research Product completed on September 30, 2017

This research product summarizes the research results and potential application to Reclamation's mission.

Return to Research Projects

Last Updated: 6/22/20