Providing a Fuller Characterization of Uncertainty in Climate Impact Assessments to Better Inform Local-to-Regional Scale Decisions in the Water Resources Sector

Abstract:

Assessments of projected climate change impacts on water resources at local-to-regional scales typically use a combination of numerical models for evaluating the various differences in projected effects resulting from different driving emissions scenarios and different climate models. Regional climate models, used to describe how large-scale changes in the Earth system effect change in the local or regional climate, and hydrologic models, used to describe how changes in local or regional climate effect change in local or regional hydrologic processes, are both typically used in these assessments. However, with both model types, assessments most often include only one or a small number of different models, and inter-model differences are often not constrained to define the uncertainties in the simulations consistently. More generally, the impacts assessment modeling community has focused very intently on characterizing uncertainty in climate projections using multiple climate models and multiple emission scenarios, doing less work to characterize and understand uncertainties in the regional climate and hydrologic models, and their important interactions, for impacts assessment and local-to-regional scale decision-making.

The US Army Corps of Engineers is working with partners on a number of projects to help provide fuller characterizations of some previously undescribed uncertainties with the goal of helping inform its local-to-regional applications. Some salient results from those projects include: (1) choice of methods to produce gridded meteorological fields can have effects on projected hydrologic outcomes as large as the climate change signal; (2) many statistical downscaling methods popular in the water management community produce hydroclimate representations with too much drizzle, too small extreme events, and improper representation of spatial scaling characteristics relevant to hydrology; and (3) outcomes depend significantly on subjective decisions made in calibrating hydrologic models, such as the choice of forcing data, the choice of calibration scheme, and the choice of objective function. This presentation will explain these and other results from testing model uncertainties and will suggest possible decision outcomes which can be sensitive to those uncertainties.