Hydrologic responses to climate change: considering geographic context and alternative hypotheses

Received 6 January 2011 Accepted 14 January 2011 One of the most significant consequences of climate warming is the likely change in streamflow as a result of warming air temperatures. Hydrologists have responded to the challenge of understanding these effects. Many recent studies quantify historical trends in streamflow and usually attribute these trends to climate warming, via altered evapotranspiration and snowpack (Figure 1.a). However, without questioning the fundamental reality of a warming climate, hydrologists should also consider biotic and social processes whose omission may produce misleading interpretations about climate change effects on hydrology. The aim of this commentary is to raise awareness of ecological and social processes that may confound the interpretation of climate effects on hydrology, to review how the geographic context of streamflow records affects interpretation of the climate signal, and to suggest a ‘checklist’ of working hypotheses that can be used to structure studies of streamflow responses to climate change. A wide variety of trends in streamflow have been detected and attributed to climate change and variability, but a few themes dominate the literature. The most common studies report earlier snowmelt, a shift to earlier streamflow timing, altered spring maximum flows, and/or intensified summer drought (Adam et al., 2009; Barnett et al., 2008; Brabets and Walvoord, 2009; Burn et al., 2010; Cuo et al., 2009; Hamlet et al., 2007; Hodgekins et al., 2003; Hodgekins and Dudley, 2006; Huntington et al., 2004; Jefferson et al., 2008; Knowles et al., 2006; Lee et al., 2004; Mote et al., 2003; Shepherd et al., 2010; Stewart et al., 2005; Stewart, 2009; Wilson et al., 2010; Xu et al., 2009). These studies focus on mountainous regions or near-polar latitudes of the Northern Hemisphere, and the relationships among warming, snowmelt, and streamflow vary with geographic location, elevation, and latitude. Another frequent finding is a trend of increased streamflow (annual, winter, and/or spring) associated with increased precipitation or temperature, or both (Andreadis and Lettenmaier, 2006; Birsan et al., 2005; Chen et al., 2006; Gautam et al., 2010; Johnston and Schmagin, 2008; Lins and Slack, 1999; Liu et al., 2010; Milliman et al., 2008; Peterson et al., 2002; St. George, 2007; Wilson et al., 2010; Xu et al., 2009; Zhang et al., 2001; Zhang and Schilling, 2006). The flow quantiles affected vary, with some studies reporting increased low flows (Liu et al., 2010) while others project increased flood risk (Allamano et al., 2009, but see Wilby et al., 2008). The climate-streamflow trend literature also contains considerable discussion of methods. Most studies use the Mann-Kendall non-parametric test (Hirsch and Slack, 1984; Helsel and Hirsch, 2002). Moreover, there is broad recognition that trends can be confounded with long-term climate cycles (Burn, 2008; Huntington et al., 2004; Lee et al., 2004; Marengo, 2009; St. George, 2007; Weider and Boutt, 2010; Woo et al., 2006) and that trends are sensitive to the start date of the record (e.g. Wilby et al., 2008). These latter issues are not addressed in this commentary.