Very High Resolution Climate Change Projections for Hydrologic Impacts Assessments over the Lake Champlain Basin in Vermont

Wednesday, 17 December 2014: 5:45 PM
Jonathan Winter1, Brian Beckage2 and Gabriela Bucini2, (1)Dartmouth College, Hanover, NH, United States, (2)University of Vermont, Burlington, VT, United States
The Lake Champlain Basin is a critical socioeconomic resource for the Northeastern US and Southern Quebec. While global climate models (GCMs) provide an overview of climate change in the region, they lack the spatial resolution necessary to fully assess the effects of increasing greenhouse gas concentrations on hydrologic and ecologic processes. One approach to address this limitation is statistical downscaling, which both bias corrects and increases the spatial resolution of GCM simulations. However, even the increased spatial resolution of most statistically downscaled products (~1/8°) is not sufficient for detailed hydrologic, ecologic, and land-use modeling in the small watersheds of the Lake Champlain Basin.

An ensemble of very high resolution (30”) precipitation and temperature projections (1950-2099) is developed for the Lake Champlain Basin by applying an additional level of downscaling based on topography and the dense station network in Vermont and Quebec to intermediately downscaled (1/8°) data. The additional downscaling consists of two main steps. First, maximum likelihood estimation is used to derive the observed relationships between precipitation and elevation, and temperature and latitude and elevation. Then, these relationships are combined with spatial interpolation to enhance the resolution of intermediately downscaled GCM projections.

The resulting very high resolution dataset is analyzed both for its ability to reproduce station observations over a historical period (1970-1999), as well as add value when compared with spatial interpolation only. The sensitivities of the projections and additional downscaling to GCM, greenhouse gas emissions scenario, interpolation method, and intermediately downscaled dataset are evaluated.