Predicted Changes of the Relative Magnitude of Water Storage

Download Report

Transcript Predicted Changes of the Relative Magnitude of Water Storage

Predicted Changes of the Relative Magnitude of Water Storage Capacity in Seasonal
Snowpack to Artificial Reservoirs at the Global Scale
Jennifer C. Adam & Jennifer M. Johnston
Civil and Environmental Engineering, Washington State University, PO Box 642910,Pullman, WA 99164-2910.
4. Basin-Scale Results
1. Introduction
Many watersheds that receive most of their precipitation during the winter rely heavily on mountain snowpack to store
water to meet water demands during the dry summer months. Climate change projections predict a warming trend over the
next century, leading to a reduced snowpack (due to more winter precipitation falling as rain) and an earlier melting of
snowpack, shifting the availability of water away from the dry summer months when it is needed most. In addition to
snowpack, humans rely on stocks of water stored in artificial (man-made) reservoirs, which offer year-round water access.
However, such reservoirs are limited in capacity. In many regions, prospective dam construction projects have slowed or
halted due to the possibility of negative environmental impacts and many previously constructed dams have recently been
removed or scheduled for removal.
Our objective is to compare the predicted reduction in pre-snowmelt snow water equivalent (SWE) due to climate change
and compare this to the existing storage capacity in each global river basin. The rationale for this study is that river basins
that have a low reservoir storage capacity in comparison to the predicted change in SWE will be less able to adapt to the
impacts of climate change on water resources availability because they will not be able to store increases in winter runoff
into the summer season. Understanding which regions will be more susceptible to climate change is vital for long-term
water resources planning.
2. Methods and Data
•Early spring SWE for historic (1970-1999) and future (2025-2054) periods were simulated.
•Early spring SWE is defined as the average SWE during March and April for the Northern
Hemisphere or during September and October for the Southern Hemisphere.
•We simulated SWE using the Variable Infiltration Capacity (VIC) model run at a half degree
resolution.
• For future climate, we utilized 48 scenarios: 16 global climate models (GCMs) and 3
emission scenarios (B1, A1B, and A2).
•Model output was averaged over global river basins and inter-basins and compared to
reservoir storage capacity and population.
•Reservoir capacities are from the Global Water System Project (GWSP) Atlas Dams and
Capacity of Artificial Reservoirs dataset.
•Population Data are from the Center for International Earth Science Information Network
(CIESIN), for year 2004.
Global Basins
•Historical and future spring SWE were
averaged over each global basin (see
left) and inter-basin (see below).
•For the 48 simulations of future SWE,
the min, max, and 25%, 50%, and 75%
SWE values were found.
•These points were subtracted from
historical SWE to determine “loss of
SWE Storage”.
•They were plotted against reservoir
storage capacity and population (see
plots below).
•The points were ordered (left to right)
according to loss of SWE storage (greatest to
least).
•Only basins/inter-basins with significant
historical SWE were included.
•Regions with the highest susceptibility to these
climate change effects are those with a high loss
of SWE storage and/or high future SWE
uncertainty, low reservoir capacity, and high
population (highlighted in yellow).
Global Inter-Basins and Reservoir Locations
Maximum
75% Quantile
3. Global-Scale Results
1970-1999 Early Spring SWE, mm
Change in Early Spring SWE (Future-Historical), mm
50% Quantile
25% Quantile
15
10
5
Minimum
0
-5
-10
-15
Reservoir
Capacity
(km3)
•Early spring SWE for the historic (1970-1999) period was
simulated using observed meteorology.
•Regions pole-ward of the 40° latitude and mountainous areas are
considered snowmelt-dominated (Barnett et al. 2005).
Standard Deviation of Future Early Spring SWE, mm
•The historical spring SWE was subtracted from the mean of the 48
future simulations.
•Early spring SWE increased in some regions and decreased in
others due to combined precipitation and temperature changes.
Coefficient of Variation of Future Early Spring SWE
Population
(Millions)
20
10
0
-10
-20
-30
-40
-50
-60
5. Conclusions
•The standard deviation of the 48 future simulations was
calculated.
•The variation among future predictions is greatest in mountainous
areas, suggesting that uncertainty in these regions is greatest.
•The coefficient variation was determined by dividing the standard
deviation by the mean (of the 48 future simulations).
•When normalized by the mean, the areas showing the greatest
uncertainty are the boundaries of the snowmelt-dominated region.
 We found that in some regions of the globe the models predict a significant decrease in the early spring snow
storage from the historical to future periods. Some of these regions of SWE loss are in basins with low reservoir
storage and high population, making these regions particularly sensitive to climate change impacts.
 The uncertainty in future SWE due to climate change uncertainty is largest at the boundaries of snowmeltdominated areas, which is close to large population centers. This creates a further vulnerability to climate
change.
 A number of limitations exist in this study and include scale issues related to reservoir location relative to
snowpack and the basin boundaries, excludes the effects of climate change on glaciers (an additional
vulnerability in many locations), additional uncertainty due to model parameterization, mode structure, and
downscaling technique, and stresses and limitations on existing reservoirs.