Keywords
global warming thresholds, large-ensemble model, signal-to-noise ratio, snow drought, uncertainty
Start Date
17-9-2020 12:00 PM
End Date
17-9-2020 12:20 PM
Abstract
Uncertainties in hydro-climatic projections arise out of the use of different climate forcings, models and methods, with potentially large noise compared to change signal. In this study, we used two structurally different approaches: (i) a large ensemble (50 realizations) of Canadian Regional Climate Model (CanRCM4-LE) and (ii) Variable Infiltration Capacity (VIC) hydrologic model driven by statistically downscaled GCM ensemble, and analysed the maximum snow water equivalent (SWEmax) projections over North-Western North America – a region heavily dependent on the snowpack freshwater storage. We considered the spatial and temporal variability of changes under 1.0°C to 4.0°C warming above the preindustrial global mean temperatures. The results indicate consistent direction of change from both sets of projections. Specifically, steep SWEmax decline in the warmer coastal/southern basins, moderate decline in the milder interior basins, and either small increase or decrease in the colder northern basins, are projected. A key factor for these spatial differences is the proximity to freeze/melt threshold, with larger SWEmax declines for the basins closer to the threshold. Furthermore, under a categorical framework of below-normal SWEmax defined as snow drought (SD), both CanRCM4-LE and VIC results indicate predominant SD occurrences under above-normal temperature and precipitation. This implies a limited capacity of the precipitation increase to compensate the temperature driven snowpack decline. However, the magnitude of changes from the two sets of projections show considerable differences, with larger SWEmax losses and more frequent and severe SD occurrences for CanRCM4-LE compared to VIC based projections. Hence, the differences in GCM/RCM model structures and their internal variability, downscaling methods and model parameterizations have a larger influence in the quantitative change signals compared to the qualitative change signals. Nevertheless, consistent results, including extreme snow loss in the southern basins, indicate the highest impacts in the region where current water demands are also the highest.
Uncertainties in Snowpack Projections over North-Western North America from a Large-Ensemble RCM and a Hydrologic Model
Uncertainties in hydro-climatic projections arise out of the use of different climate forcings, models and methods, with potentially large noise compared to change signal. In this study, we used two structurally different approaches: (i) a large ensemble (50 realizations) of Canadian Regional Climate Model (CanRCM4-LE) and (ii) Variable Infiltration Capacity (VIC) hydrologic model driven by statistically downscaled GCM ensemble, and analysed the maximum snow water equivalent (SWEmax) projections over North-Western North America – a region heavily dependent on the snowpack freshwater storage. We considered the spatial and temporal variability of changes under 1.0°C to 4.0°C warming above the preindustrial global mean temperatures. The results indicate consistent direction of change from both sets of projections. Specifically, steep SWEmax decline in the warmer coastal/southern basins, moderate decline in the milder interior basins, and either small increase or decrease in the colder northern basins, are projected. A key factor for these spatial differences is the proximity to freeze/melt threshold, with larger SWEmax declines for the basins closer to the threshold. Furthermore, under a categorical framework of below-normal SWEmax defined as snow drought (SD), both CanRCM4-LE and VIC results indicate predominant SD occurrences under above-normal temperature and precipitation. This implies a limited capacity of the precipitation increase to compensate the temperature driven snowpack decline. However, the magnitude of changes from the two sets of projections show considerable differences, with larger SWEmax losses and more frequent and severe SD occurrences for CanRCM4-LE compared to VIC based projections. Hence, the differences in GCM/RCM model structures and their internal variability, downscaling methods and model parameterizations have a larger influence in the quantitative change signals compared to the qualitative change signals. Nevertheless, consistent results, including extreme snow loss in the southern basins, indicate the highest impacts in the region where current water demands are also the highest.
Stream and Session
false