Grantee Research Project Results
Final Report: Improved Methods for Assessment of Hydrologic Vulnerability to Climate Change
EPA Grant Number: R824802Title: Improved Methods for Assessment of Hydrologic Vulnerability to Climate Change
Investigators: Lettenmaier, Dennis P. , Palmer, Richard , Wood, Eric F.
Institution: University of Washington , Princeton University
EPA Project Officer: Chung, Serena
Project Period: October 1, 1995 through March 31, 2000
Project Amount: $463,762
RFA: Regional Hydrologic Vulnerability to Global Climate Change (1995) RFA Text | Recipients Lists
Research Category: Ecological Indicators/Assessment/Restoration , Climate Change , Water
Objective:
Of the potential societal effects of global warming, the implications for hydrology and water resources are among the most significant. Particularly in areas where current water use courts supply failure, where projected water use is estimated to be unsustainable under the current climate, and in areas periodically devastated by flooding, even moderate projections of global warming and the concomitant hydrologic effects raise alarm. Over the past decade, the prospect of global warming has motivated a proliferation of methods aimed at assessing regional vulnerabilities to our best "educated guesses" at climate changes that may occur over the next several decades or even the next century. The objective of this research was to develop and evaluate, via case study applications spanning a range of hydroclimatic and geographic conditions and spatial scales, a balanced approach for assessing regional hydrologic sensitivity and water resources vulnerability to climate change.Summary/Accomplishments (Outputs/Outcomes):
The assessment approach is a sequence of steps linking simulations of climate change at the global scale to computer model simulations designed to evaluate forecasts and climate change sensitivity at the local scale. In our case studies, we applied a set of methods for climate change estimation that represents a reasonable tradeoff between reliance on the best available forecasts and methods and computational expense. This set of methods, in brief, begins with climate change estimates derived from general circulation model (GCM) simulations of global warming; translates these to the region of interest via a process called "downscaling"; simulates surface hydrology; evaluates water resources implications; and lastly, compares water resources system sensitivity to climate change with system sensitivity to nonclimate related changes.
Vulnerability Assessment Methodology
Specification of Large-Scale Climate Change Scenarios. As the first step in the sequence, the development of alternative climate scenarios can proceed simply from an arbitrary specification of changes in key surface variables, or with more complexity by extracting these changes at selected points from a run of a GCM. GCMs provide dynamic simulations of global climate over time horizons reaching several hundred years, based on the current state-of-the-science atmospheric physics and ocean-land-atmosphere interactions. GCMs are thought to yield the most reliable estimates of potential climate change available, yet estimates from different models fail to agree on the magnitudes of temperature change and the magnitude and direction of precipitation changes. Nonetheless, GCMs have progressed to the point at which we choose to constrain the scenarios of meteorological change using a set of GCM run outputs, bracketing a range of potential climate changes for a given future period. Lacking a rationale for choosing one GCM over another, consideration of the entire ensemble of results is preferable to relying on a single GCM.
In contrast to earlier GCM-based studies that used static climatic conditions (typically, a doubled CO2 climate), we adopt an approach which recognizes that shifts in climate are gradual. GCM research centers now simulate "transient" climate change scenarios, accounting for thermal lags in the ocean-atmosphere coupling. By extracting decadal averages of temperature and precipitation change at multiple points from a transient climate simulation, we form a sequence of projected changes from each transient run (e.g., one climate change estimate each for the decades centered on 2020, 2030, 2040, etc.). From each GCM in the set chosen for an assessment comes one such sequence, forming a range of possible future climate progressions.
Downscaling Method. The most commonly applied "downscaling" (i.e., bridging the scale mismatch between global models and regional assessments) approaches have been: (a) nested modeling?the use of a detailed atmospheric model at an intermediate scale; (b) statistical approaches relating GCM scale atmospheric outputs to regional scale meteorology; and (c) simple perturbation approaches, in which predicted future large-scale changes in surface variables are applied to shift or scale local meteorological records. From the standpoint of uncertainty, the downscaling step is among the least quantifiable links in the sequence of models used in the vulnerability assessment; hence, the allocation of a great deal of effort to downscaling is, we believe, unjustified. The perturbation method offers a reasonable tradeoff between simplicity of application and ability to link GCM scenario output to the hydrologic modeling analysis. Downscaling via the perturbation method proceeds as follows:
- The large-scale changes extracted from the GCM grid locations are spatially interpolated to locations of interest in the region (i.e., surface observation station locations, the river basin centroid, or grid cell locations in a distributed hydrologic model);
- The interpolated temperature and precipitation changes for each decadal average in each future climate sequence described in part (a) above are used to shift (in the case of temperature) or scale (in the case of precipitation) local observed daily surface records of length 4-5 decades, producing altered, historical time series that correspond to one decadal-average future climate scenario.
Hydrologic Assessment. The downscaled temperature and precipitation time series for each climate change scenario are used to drive a hydrologic model, producing estimates of runoff (hence, streamflow) and evapotranspiration. Comparison of the output for these scenarios with output from a base case, historical (unaltered meteorology) scenario reveals the regional hydrologic sensitivity to the range of changes forecasted by the set of GCMs for the region.
The hydrologic model used in the case studies is physically-based, "macro-scale" (for large river basins), "distributed" (using different parameters for different locations, rather than "lumped", in which parameters are spatially aggregated), and "continuous" (adhering to a moisture and energy budget throughout a simulation). While any well-calibrated hydrologic model will suffice to translate precipitation and temperature changes into runoff changes, physical, continuous, spatially-distributed models are arguably better suited to evaluate streamflow sensitivity than models that lack these attributes. Continuous models preserve the water balance over a number of years; physical models are more likely than regression-based models to be reliable outside of the historical climate boundaries in which they have been calibrated; and spatially-distributed models will better capture spatial variability in responses to climate change. Although the implementation of such models is more time-consuming and computationally intensive than that of simpler hydrologic models, this effort is justified by the physical realism the models offer.
Water Resources System Evaluation. For many regions of the world, the effects of climate change on hydrology are buffered by a water resources operational system consisting of, for example, reservoirs, locks and dams, hydropower turbines and pumping facilities. The performance of the system, and indirectly, the value of system operation to society (e.g., for flood control, hydropower or water supply), depends on the streamflow entering the system and also upon the rules for operating the system (e.g., making releases or withdrawals). At this point in the assessment sequence, each climate scenario has been translated into a set of streamflow time series which, when input to the water resources system model, yields estimates of system performance. Contrasted with performance results for current climate streamflows and with performance results arising from operational changes that reflect potential nonclimate related changes in system priorities, these estimates characterize the regional water resources vulnerability to climate change.
Water resources systems operation models have been in use since the early 1950s, and there is a wealth of literature describing their application. To the extent that they are able to mimic the operating guidelines of a given system and incorporate the physical components of the system, such models reproduce well most observed system outputs (e.g., flows, performance measures). Any system model that is validated through comparison of observed and simulated current climate outputs should suffice for evaluation of water resources vulnerability to climate change. Performance measures should be chosen that reflect the primary operations of concern for a given region.
Demand Projections/Alternative Operations Scenarios. Demand on water resources will likely increase as population increases, and the implications of these demand changes for water resources system performance as well as for regional hydrology (e.g., the effect of groundwater extraction on baseflow) may be at least as significant as the implications of potential climate change. Furthermore, the uncertainties arising in the estimation of future demand or operational changes are comparable to those associated with forecasting climate change, and equally problematic for vulnerability assessments. Demand or operational changes may compound climate-related difficulties facing water resources systems, or may offer a source for adaptations mitigating the effects of climate change. Therefore, development of a set of demand/operational change scenarios is a critical step in a hydrologic and water resources vulnerability assessment.
Methods for developing these scenarios vary from detailed econometric analyses that incorporate not only regional demographic and economic projections, but also the sensitivity of water and energy demand to climate shifts. Absent the wherewithal to complete detailed analyses, arbitrary sensitivity analysis may at least be employed to gauge the influence of different drivers of water resources system performance. The methods adopted depend greatly on the water resources development and available information in the region under study.
Case Studies
The three case study applications were the Apalachicola-Chattahoochee-Flint River (ACF) basin, the Columbia River basin, and the "Global Rivers" basins, a number of large river basins around the world. The ACF, the smallest of the three at 50,800 km2, is located in the semi-humid southeastern United States and drains into the Gulf of Mexico. The Columbia River basin, at 579,000 km2, is a snowmelt driven system in the northwestern United States and British Columbia, Canada. The Global Rivers basins, which range in size up to the Amazon River basin at 4,619,000 km2 and in climate from arid to tropical, demonstrate the application of the methodology in data sparse regions of the world.
Apalachicola-Chattahoochee-Flint River Basin. The ACF system is composed of two major tributaries of roughly equal drainage area, the Chattahoochee and the Flint rivers, which combine to form the Apalachicola River. The Flint River is essentially unregulated, while the Chattahoochee River has about 15 dams and locks operated for navigation, flood control, water supply, recreation, and hydropower generation. Transient simulations from three GCM centers were used to form climate scenarios for 2005, 2020, 2030, 2040, and 2050, which were compared to a current conditions (~1990) scenario. Annual average temperature changes for 2050 ranged from 2.4 to 3.1 oC, and precipitation changes were all positive, from 3 to 14 percent. Hydrologic changes ranged from large (25 percent) increases to moderate declines (7 percent) in runoff in the later decades, mostly connected to precipitation increases or decreases. Climate change effects for the ACF system were largely related to changes in inflow under each climate scenario and the relative priorities of system uses. Water supply, with highest priority, was insensitive to differences in the climate scenarios. Flood control, hydropower generation, recreation, and navigation were all vulnerable to changes in climate, with positive or negative effects (depending on the direction of streamflow changes) as large as 40, 10, 20, and about 5 percent, respectively. The alternative demand/operational changes had a much greater influence than climate for the water supply and recreation objectives, but only a small effect on the other objectives. Demand management may, therefore, offer a partial remedy for water supply declines should the more severe of the forecasted climate change impacts occur.
Columbia River Basin. The Columbia River, the largest natural water resource of the Pacific Northwest, is managed for flood control, hydropower production, irrigation, and navigation, with ecological considerations (e.g., preservation of salmon) and river recreation becoming ever more important. The total reservoir storage, operation of which has extensively altered the natural flow regime, can hold about 30 percent of the mean annual flow. Seasonal accumulation and melt of snow in the mountainous portions of the basin is a critical determinant of basin hydrology and system operation. Transient climate simulations at four modeling institutes were examined, and the two GCMs that roughly bracket the changes predicted by the suite of models were selected for detailed analysis. Decadal average projections for 2025, 2045, and 2095 were developed, in addition to several operational change scenarios related to change in priority from hydropower to salmon preservation and recreation. Annual average temperature changes for 2050 ranged from 2.6 to 3.0 oC, and precipitation changes varied in direction, from -9 to 5 percent. The primary effect of climate change under these scenarios was to reduce snowpack, shifting the seasonal runoff cycle forward from spring to winter. Streamflow sensitivity was on the order of +/- 10-15 percent. The sensitivity of system performance to operational changes was similar, suggesting that operational adaptations (particularly, a shift in hydropower to match the runoff shift) may significantly offset certain climate impacts.
Global Rivers. Nine large-scale river basins were selected to study the effects of climate change: the Amazon, Amur, MacKenzie, Mekong, Mississippi, Xi, Yellow, Yenisei, and Severnaya Dvina River regions. The basins range from tropical (Amazon) to arctic (Yenisei) and from mid-latitude rainy (Mississippi) to arid and cold (Yellow). Transient climate simulations from eight climate models were examined, and the four models with the best spatial resolution were selected for further study. Decadal average projections for 2025 and 2045 were formed. By 2025, annual average temperatures were predicted to increase at least 1 oC for all nine basins. Larger changes were foreseen for 2045, particularly in the high northern latitudes (e.g., +4.2 oC for the Yenisei). Precipitation changes varied by basin, with an annual average increase of about 10 percent predicted for most. The climate change scenarios led to increased flow in the winter for cold climate basins due to the earlier onset of snow melt. Increased summer evapotranspiration during the summer reduced flows in those basins lacking a simultaneous increase in summer precipitation. The Yellow River, which is arid under current climate conditions, was the only basin to suffer a consistent reduction in streamflow during all seasons for all four climate scenarios. Unfortunately, demand change and water resources system vulnerability assessments were not completed within the period of the study; future work in this area is needed.
Conclusions:
In summary, given the demonstrable uncertainties in the generation of climate change scenarios by even the most technologically advanced methods (GCMs), and the additional difficulties associated with downscaling from predictions on the global scale to those for the regional scale, it is appropriate at this stage to approach climate change vulnerability assessment as a sensitivity analysis of regional hydrology and water resources system operation, using GCM derived estimates to define a "plausible range" of climate changes. Some regions may be hydrologically robust to this range of meteorological changes, whereas others may currently exist on a threshold at which the changes will have significant effects. The methodology outlined above is useful in identifying regions of hydrologic sensitivity and water resources vulnerability, together with potential nonclimate related adaptations that may help to ameliorate potentially deleterious effects of global warming.Journal Articles on this Report : 5 Displayed | Download in RIS Format
Other project views: | All 8 publications | 5 publications in selected types | All 5 journal articles |
---|
Type | Citation | ||
---|---|---|---|
|
Hamlet AF, Lettenmaier DP. Effects of climate change on hydrology and water resources in the Columbia River basin. Journal of the American Water Resources Association December 1999;35(6):1597-1623. |
R824802 (Final) |
Exit |
|
Lettenmaier DP, Wood AW, Palmer RN, Wood ER, Stakhiv EZ. Water resources implications of global warming: a U.S. regional perspective. Climatic Change September 1999;43:537-579. |
R824802 (Final) |
Exit |
|
Nijssen B, Schnur R, Lettenmaier DP. Global retrospective estimation of soil moisture using the variable infiltration capacity land surface model, 1980-93. Journal of Climate 2001;14(8):1790-1808. |
R824802 (Final) |
Exit Exit |
|
Nijssen B, O'Donnell G, Hamlet AF, Lettenmaier DP. Hydrologic sensitivity of global rivers to climate change. Climatic Change 2001;50(1-2):143-175. |
R824802 (Final) |
Exit Exit |
|
Nijssen B, O'Donnell GM, Lettenmaier DP, Lohmann D, Wood EF. Predicting the discharge of global rivers. Journal of Climate 2001;14(15):3307-3323. |
R824802 (Final) |
Exit Exit |
Supplemental Keywords:
watersheds, global climate, vulnerability, flooding, drought, public policy, decisionmaking, cost benefit, hydrology, modeling, general circulation models, regional circulation models, climate models, pacific northwest, British Columbia, Columbia River, Apalachicola River, Chattahoochee River, Flint River, Senegal, Amazon River, Amur, Severnaya Dvina, Xi, MacKenzie River, Mekong River, Mississippi River, Yellow River, Yenisei River, agriculture, water supply, navigation, fisheries, hydropower, water-based recreation., RFA, Scientific Discipline, Air, Geographic Area, Hydrology, climate change, Northwest, Pacific Northwest, Atmospheric Sciences, Ecological Risk Assessment, EPA Region, water resources, environmental monitoring, regional hydrologic vulnerability, energy generation, Columbia River, Boston Metropolitan Area, drinking water supplies, hydrologic models, climate models, vulnerability assessment, temperature variables, land and water resources, Region 10, stochastic weather generators, Region 1, climate variability, climatic modelsRelevant Websites:
http://www.hydro.washington.edu/Lettenmaier/CurrentResearch.htmlProgress and Final Reports:
Original AbstractThe perspectives, information and conclusions conveyed in research project abstracts, progress reports, final reports, journal abstracts and journal publications convey the viewpoints of the principal investigator and may not represent the views and policies of ORD and EPA. Conclusions drawn by the principal investigators have not been reviewed by the Agency.