CHAPTER 1
A changing and uncertain future for freshwater
This chapter provides an overview of the range of complex impacts on freshwater that can be expected in a changing climate. It also highlights the main sources of uncertainty and information gaps associated with climate change impact assessments on water systems that pose challenges for informing practical, on-site adaptation decisions. Finally, it examines the policy implications of the proposition that "stationarity is dead", or in other words, that the future for freshwater will not look like the past.
Key messages
• Climate change is, to a large extent, water change. Climate change affects all aspects of the water cycle and water is the main way through which the impacts of climate change will be felt. The consequence of these impacts will depend on their nature, where and when they occur, and the exposure and vulnerability of the populations, ecosystems, and physical assets they affect.
• Typical assessments of the impacts of climate change on freshwater are of limited use when it comes to making practical, on-site decisions about adaptation. In general, the level of confidence in climate change projections decreases as their potential utility for making decisions on how to adapt increases. Adaptation decisions need to accommodate considerable uncertainty.
• One trend appears predictable: the future for freshwater will not look like the past. This shift calls for a flexible, dynamic, future-oriented approach that takes into account climate variability on all timescales.
Climate change impacts on freshwater resources are already evident and are projected to become more significant and to accelerate over time (Bates et al., 2008). There is also a growing recognition that climate change presents a singular challenge for water systems by rendering the historical assumption of stationarity2 increasingly irrelevant (Milly et al., 2008). This means that a fundamental assumption upon which water management, infrastructure design and planning, and ultimately many economic and resource management decisions are founded will no longer be a reliable basis for future planning and management. Decisions made today may lock us into management strategies and infrastructure for many decades that will not match future climatic conditions. The unprecedented rate of change and potential novel changes outside of historical experience introduce a greater degree of uncertainty beyond what water managers have traditionally had to cope with.
Uncertainty and knowledge gaps
Despite an ever-expanding scientific basis, reliable information about the nature, magnitude and timing of hydrological impacts at the scale needed for water resources planning and management is generally lacking. Improving this information base to attain the required level of detail and confidence needed to inform practical, on-site adaptation decisions will take time. This is due to limitations in data, modelling capacity, and computational requirements. Moreover, by their very nature, climate and hydrological systems are hard to predict. Given the current state of knowledge and limits to predictability of climate change impacts on water, effective decisions to adapt to climate change will need to be made in the absence of accurate and precise climate predictions (Dessai et al., 2009).
Most climate change impact studies rely on projections from global circulation models (GCMs). These models were originally designed to assess the global impact of various emissions pathways in order to make the case for mitigation efforts. The extension of their use to adaptation decision-making is a relatively recent development. The current suite of climate models were not developed to provide the level of accuracy required to inform adaptation decisions for water resources management (Kundzewicz and Stakhiv, 2010). The utility of GCMs for adaptation decisions for water resources and the most promising approaches for addressing their shortcomings or developing alternatives are the subject of widespread debate (Kundzewicz and Stakhiv, 2010). Even if some of the limitations of current approaches can be addressed, adaptation decisions will still need to accommodate considerable uncertainty (Anagnostopoulos et al., 2010; Wilby, 2010; Bates et al., 2008).
Although is it becoming "standard" practice for climate change impact assessments for water to link the results of a climate change model for temperature and precipitation with a hydrological model for runoff, these assessments have several limitations (Rodríguez-Iturbe and Valdés, 2011). However, the emergence of a "standard" approach does not imply that there is consensus about the utility or effectiveness of this approach. Novel approaches, such as "decision scaling" (Brown and Wilby, 2012), are emerging as a promising way forward (see Chapter 2 for further discussion). The main sources of uncertainty and knowledge gaps associated with typical climate change impact assessments for water resources are highlighted below.
Uncertainty related to scenarios, emissions trends, and models
There is significant uncertainty associated with climate models' reproduction of the current climate and simulation of the future climate. These models simulate some climatic processes in only a rudimentary fashion (Bates et al., 2008). Other sources of uncertainty are the scenarios used to estimate emission trends and the way that climate models simulate the impact of those trends on the climate. Depending on the climate model, the same emission trend can produce a wide variation of climate change projections. This is particularly acute in the case of precipitation and evapotranspiration. Hydrological models also add substantial uncertainty due to regional differences and limitations in the coverage of monitoring networks (Huntington, 2006).
Modelling uncertainty may be reduced to some extent by running an ensemble (several slightly different models of the climate system) or thousands of runs from a single model. Yet, in practice, this is a complex and resource-intensive task. Also, the results from an ensemble run may diverge significantly. Even multiple runs of a single model can show significant variation between projections at coarse spatial and temporal scales. Lack of agreement between climate models does not mean that there will be no impact, or that any given impact is unlikely. Instead, it may mean that there is a large range of possible futures, including significant potential increases or decreases in a given climate parameter.
Coarse resolution/scale mismatch
In general, the level of confidence in climate change projections decreases as their potential utility for adaptation decision-making increases. As the spatial scale decreases, projections become less consistent between models. Thus, it is widely recognised that findings from global assessments of climate and hydrological change are not directly usable by decision makers at regional, national and subnational levels for adaptation (UNFCCC, 2011).
The coarse "resolution" of global climate models means that outputs are insufficiently detailed for climate impact studies at finer geographic scales. While climate models have been able to reproduce broad features of the past climate at large geographic scales (continental and above), in general, they still cannot reconstruct the important details of the climate at finer scales (Kundzewicz and Stakhiv, 2010). Because of these limitations, outputs from GCMs are typically "downscaled". These techniques require significant information on the ground for calibration (Rodríguez-Iturbe and Valdés, 2011). However, efforts to address the scale mismatch of global models to provide more site specific information through downscaling also have serious practical limitations (Wilby and Dessai, 2010).
Low confidence for key climate parameters
There is a higher degree of confidence in estimates of shifts in temperature, than for changes in precipitation and evapotranspiration. Unfortunately, precipitation, the principal input to freshwater systems, is not adequately simulated in present climate models (Kundzewicz et al., 2008). With some exceptions, models generally disagree about the magnitude of precipitation changes and sometimes also the direction of the change. Low confidence in precipitation projections also precludes the reliable estimate of changes in flood frequency and magnitude.
A focus on shifts in the mean
Changes in averages are easier to project than changes in extremes. However, projections of increasing average temperatures and changes in average annual rainfall are of limited use for adaptation decision-making. Reliable estimates of changes in the forms of precipitation (rain or snow), seasonal timing of precipitation, inter annual variability, shifts in runoff and river discharge are generally lacking. Moreover, most models do not even try to simulate shifts in extremes, often assuming that current levels of variability will continue relative to a shifting mean. This assumption may overlook some of the most severe, sudden, and costly impacts of climate change on water. Figure 1.1 illustrates several modes of climate change. Climate models tend to focus on changes in the "mean", which is not likely to be either the most likely mode of change (Le Quesne et al., 2010), nor the most relevant for adaptation decisions.
A widening range of possible futures
Uncertainties in climate change impact assessments for water resources arise in all stages of the impact assessment process. Figure 1.2 illustrates how the sources of uncertainties compound to multiply the range of possible futures.
Scientific advances can either reduce or expand the range of uncertainty. For example, as previously unknown processes are identified and described uncertainties about those processes may expand the existing range of uncertainty. Even if perfect climate models could be built, uncertainty about all the other non-climatic pressures means that regional hydrological projections would still be highly uncertain (Wilby, 2010).
A future for freshwater unlike the past
Despite all of the uncertainty, one trend appears predictable: the future will not look like the past. The notion that "stationarity is dead" (Milly et al., 2008) and is no longer an adequate guide for future water resources risk assessment and planning has gained widespread acceptance (Wilby, 2010). This marks a significant departure from the past, as much of the experience to date in managing water resources and infrastructure is based on the historical record of climate variability during a period of relatively stable climate.
Stationarity has served as a central, default assumption for water resources management. Probability-based design using historical climate data informs the construction and operation of levees, dams, spillways, and water supply and sewage treatment systems. Flood frequency characteristics, although meaningful under the assumption of stationarity, are questionable in the nonstationary environment (Kundzewicz and Somlyody, 1997). Water allocation regimes are oftentimes based on historical climate data, as are the basic tools that the insurance industry uses to communicate about water risks. Climate change alters the assumptions, data and modelling techniques required to develop information about water risks, such as flood zone maps and, ultimately, their utility as the basis for insurance (Ludwig and Monech, 2009).
The declining relevance of historical climate information to inform current and future planning presents a major challenge for water managers and policy makers. This shift signals the end of the static design paradigm for water resources systems, in favour of a dynamic response to changing conditions at all timescales (Brown, 2010). Such a dynamic approach would consist of periodically adjusting forecasts, building a flexible system focussed on robustness rather than optimisation, and taking a multidisciplinary approach to managing risk. A flexible, dynamic approach is required to minimise potential mismatches between water infrastructures and future climate (Matthews et al., 2011).
Climate change as water change
Despite the uncertainty and knowledge gaps, there is a significant and growing body of scientific evidence documenting the range of complex changes in the water cycle that can be expected in a changing climate. This evidence is useful to provide an overview of the broad range of changes that will present challenges for managing water systems in the future.
Climate change affects all aspects of the water cycle and water is the predominant means through which the impacts of climate change will be felt. Climate change is driving an ongoing intensification (increases in evapotranspiration and precipitation) of the water cycle4 (Huntington, 2006). In other words, in a warmer atmosphere, the water cycle is speeding up.
A certain amount of climate change is already unavoidable, regardless of future greenhouse gas emissions. Significant impacts have already occurred. The OECD Environmental Outlook to 2050 projects that without more ambitious mitigation policies, the mean global temperature could increase by 3 ºC to 6 ºC above pre-industrial levels by the end of the century (OECD, 2012). Significant action is required to limit mean increases to 2 ºC. Even if this target is achieved, it will only serve to slow the rate of climate change and there will still be considerable impacts.
Climate change impacts on freshwater include shifts in precipitation patterns, rising water temperature, deteriorating water quality, increases in evapotranspiration, and increases in the frequency and intensity of extreme events (Bates et al., 2008; IPCC, 2007). Impacts are expected to become more pronounced over time and the rate of change is expected to accelerate, with more severe impacts anticipated in the second half of the century. The range of impacts on water are summarised below.
Changing precipitation patterns and increased variability
Climate change is projected to shift the spatial and temporal distribution of precipitation, with some regions becoming wetter, others drier. In general, regions with high rainfall are projected to receive more precipitation, while arid and semi-arid regions are projected to become drier. Shifting precipitation patterns will affect runoff (see below), the rate of surface and groundwater recharge, and displace rainy seasons. More frequent and intense precipitation increases erosion and sediment loads in rivers, lakes and coastal zones with a negative effect on water quality. In arid and semi-arid regions, any reduction in rainfall has serious implications for rivers and lakes, even causing them to dry up, as seen in the case of Lake Chad (Ludwig and Moench, 2010).
Over the past several decades, increases in precipitation have been observed over land in high northern latitudes, while decreases have dominated in areas situated between 10 °S to 30 °N. Over the same period, land classified as very dry has more than doubled globally (Bates et al., 2008). Overall, the attribution of observed changes in global precipitation to climate change remains uncertain because precipitation is strongly influenced by large-scale patterns of natural variability.
Projections indicate increases in precipitation, average river runoff and water availability in high latitudes and in some parts of the tropics. Some dry regions at mid latitudes and in the dry tropics are projected to become drier (Bates et al., 2008). Many semi-arid and arid areas (e.g. Mediterranean basin, western USA, southern Africa, and north-eastern Brazil) are particularly vulnerable to the impacts of climate change and are projected to suffer a decrease in water resources (IPCC, 2007). Higher temperatures will also alter the proportion of precipitation falling as rain and snow. The form of precipitation is extremely important for snowpack-dominated regions (e.g. the Sierra Nevadas, the Andes and the Himalayas). These areas depend on seasonal snowpack to meet water demand in dry seasons. The destruction of natural storage in the form of snowpack means that alternative storage will be required to ensure that winter precipitation continues to be an economically valuable resource during the summer when demand is high.
Changes in precipitation manifest at regional or local scales and are generally poorly described by climate models. These changes are among the least well-understood and least predictable aspects of climate change while at the same time, they are among the most important for making adaptation decisions for water resources management.
