Future Hydroclimatology and theResearch Challenges of a Post-Stationary World


Stationarity is dead

(Milly et al. 2008).

This provocative statement issued recently in Science directly challenges the basic assumption underlying the way surface water resources in much of the developed world have been managed for decades. Milly et al. (2008) claim that anthropogenically-induced climate change is the reason that stationarity has died and “cannot be revived.” Although they acknowledge that the validity of the assumption has been questioned regularly in the past, Milly et al. highlight a pressing need to address this issue due to a convergence of observations and research findings that demonstrates the urgency of the influence of climate change and variability on surface water processes. Specifically, they note that projected changes in future runoff “are large enough to push hydroclimate beyond the range of historical behaviors” (p 573). With this imperative in mind, in this essay, after first addressing the call to move beyond the stationarity assumption, I present a series of questions and suggestions on how hydroclimatic research might be integrated into a future water resources agenda for geographers that addresses a “post-stationary” world, especially with respect to hydrologic extremes.

Beyond Stationarity

When a hydroclimatic time series is said to be stationary, its statistical properties (e.g., mean, variance, skewness, etc.) are all assumed to be constant over time. In practice this means that the probabilities derived from, say, a time series of annual stream flows or instantaneous flood peaks from a gauged record will be reliable estimators of the variability of those processes outside of this record – either in the past or in years to come. Those who study hydroclimatic processes as they vary over long time periods are quick to point out that in the physical world, the means and variances of hydroclimatic variables do indeed change over time due to climate variability, geomorphic change, land use alterations, and a variety of other factors. Hence for such researchers “stationarity has always been dead.” Yet the stationarity assumption has prevailed in water resources research, practical applications, and engineering design because of its operational utility and the lack of alternative methods to address the mathematical complexity of modeling nonstationary processes. If, as Milly et al. propose, we have come to the end of an era of natural hydroclimatic change and variability that is “sufficiently small to allow stationarity-based design,” a critical research need in upcoming decades will be to find innovative ways to grapple with analyzing, managing, and adapting to the water resources of a post-stationary world.

Post-Stationary Hydroclimatology and Geographic Research

The subfield of hydroclimatology has long been an active area of research for water resource geographers (see Mather 1991, Shelton 2009). Studies of surface water processes from a climatic, geomorphic, biogeographic, and cryosphere-based approach have engaged physical geographers for decades, as have studies of water resources from the perspective of policy, risk, and culture. A perusal of recent professional meeting presentations and published work by geographers reveals ongoing efforts that cover a wide array of hydroclimatic and water-related research topics including: local and global water balance components; soil moisture variations; streamflow variability; snowpack and snow cover extent; changes in timing of spring snowmelt; synoptic circulation patterns for moisture delivery; land-atmosphere feedbacks; paleoclimatology and paleohydrology; causes and variability of extreme precipitation; floods, droughts and other water-related hazards; water policy issues; and the impact of water supply variations on past and present societies. Unifying many of these efforts is the emerging issue of water-related vulnerability and adaptation to a changing climate that links the biogeophysical and social science traditions within water resource geography in new and profound ways.

How might these avenues of current hydro-climatic research be re-envisioned to meet the complex needs of a “post-stationary” world? Following are three of the most critical areas in need of fresh insights and expertise from hydroclimatologists as we face a future when hydrologic processes are expected to extend well beyond the range of historical behaviors:

  • 1Hydroclimatic change, including rising temperatures and shifts in the seasonal timing of snowmelt as they affect local and regional water balances and water supplies;
  • 2Hydrologic extremes, including droughts and floods, which have been projected to increase in magnitude and/or frequency in response to an intensification of the hydrologic cycle under global climate change; and
  • 3Cross-disciplinary and integrated assess-ments of how hydroclimatic changes have – and will – affect relationships between geophysical, socio-economic and ecological systems across multiple spatial and temporal scales.

A list of key questions for advancing a vibrant hydroclimatology research agenda in each of these three areas follows. A unifying theme emerges from the perspective of the essential geographic themes of region and scale, and how sensitivity to them is essential to meet the research challenges of a post-stationary world.

Hydroclimatic Change

  • Which geographic regions are most vulnerable to changes in the extent and timing of seasonal hydroclimatic events (extreme summer heat, winter snowfall, spring snowmelt, etc.) and what impacts will such changes have on local and regional water issues?
  • How can modeled projections of future precipitation and other changing moisture-related components of the global energy balance (Intergovernmental Panel on Climate Change 2007) be applied effectively to regional and watershed-scale areas to address biogeophysical and socio-economic water issues of the future?
  • What are the most effective communication mechanisms and scales for translating the science about hydroclimatic change into formats that will foster effective adaptive management to expected changes via stakeholder interactions, policies, decision-making, and action across diverse societal settings, cultures, and geographic regions?

Hydrologic Extremes

  • How will the projected “intensified” global hydrologic cycle (Trenberth et al. 2003) manifest itself regionally, and how will we know that observed hydroclimatic extremes in specific geographic areas have been affected, if and when they are?
  • Will the droughts and floods of the future be distinctly different in magnitude and frequency from those of the present (or past) under this intensified regime?
  • If projected hydroclimatic intensification changes the nature of extremes in given regions, how can probability estimates of extreme events be developed when assumptions of stationarity or linearity may no longer apply?
  • In what ways can extended paleo-records of reconstructed droughts, floods, and stream flow be used to provide evidence of extremes that have occurred prior to the gauged record in specific watersheds, thereby offering an expanded range of possible scenarios for future projections? How can the issue of nonstationarity be addressed when paleo-data are appended to gauged data to produce long time-series? How can extreme-value statistics derived from these long time-series be augmented with climate information and used in innovative ways to reduce uncertainties in the future?
  • Which global geographic regions are currently most vulnerable to floods and droughts and will this vulnerability increase or decrease as climate changes? Which additional regions might begin to experience extreme events more often in response to latitudinally shifting extra-tropical or tropical storm tracks? Which regions are most at risk or least resilient to flooding and inundation from rising sea level, including small islands where local changes may be harbingers of more widespread, global impacts? Will local communities need to rezone their floodplains to become more resilient to an uncertain hydroclimate and, if so, what might the floodplain maps of the future look like?

Integrated Assessments

  • Milly et al. note that “In a nonstationary world, continuity of observation is critical” (p 574). How will lengthy and continuous observing networks be maintained, especially when resources to do so are limited in many of the world's most climatically sensitive regions? As advances in remote sensing allow increasingly sophisticated observations over broad areas of the globe at multiple scales (National Research Council 2008), how can persons and networks on the ground be integrated into data collection to address validation? In particular, how might the engagement of “citizen scientists” aid in observing and monitoring the effects of hydroclimate change? (See, for example the USA National Phenology Network http://www.usanpn.org/).
  • At the watershed scale, how can new, bi-directional, translational science approaches be used to move hydroclimatic research “from the laboratory into the field” to address stakeholder and watershed-based needs, concomitant with moving the field experience of water managers “from the watershed to the laboratory.”
  • At the regional scale, what methods, data sets, and cross-disciplinary approaches will most effectively communicate complex climate-sensitive issues of concern to water resource decision-makers, emergency managers, and policy planners (e.g., Jacobs et al. 2005)? What will comprise functional data sets in various global scenarios, and how can they be shared?
  • How can an integrated geographic understanding of the biogeophysical and socio-economic attributes of watersheds and larger areas be applied toward the building of multiple scenarios for assessing the impacts of future climate and adapting to it?
  • Across all scales, what are the complex and interacting water-related mechanisms and processes that result in the emergence, sustainability, or collapse of socio-ecological systems (Costanza et al. 2007)? How can this be integrated into our models, while recalling that culture itself has always been dynamic and implicitly nonstationary?

Research Needs for the Future

To address these questions, realigned priorities, new approaches, and improved tools and data sets will become increasingly important (e.g., Gupta 2000, Logan and Helsabeck 2009) and innovative statistical techniques for modeling nonstationary behaviors in hydroclimatic processes will be required (Griffis and Stedinger 2007, Milly et al. 2008). Downscaling methods will need to be advanced and the limitations, accuracy, and precision of their results clearly communicated, especially at the watershed scale (Pulwarty 2003). “Scaling up” from local data and the identification of process-based linkages between local stream flow and regional and global circulation, will become as important as scaling down “from globally forced regional models” (Pulwarty 2003, Hirschboeck 2003). Innovations will be needed in the quest to define teleconnections and linkages between regional variations in stream flow, snow pack, or drought and indices of large-scale atmospheric and oceanic circulation patterns (see McCabe and Dettinger 2002, McCabe et al. 2004, Kingston et al. 2006, Redmond and Koch 1991). This latter effort is critical for addressing nonstationarity by obtaining a better understanding of low-frequency variations in hydroclimatic time series. Another more elusive goal, and one of great importance, is that of reliable long-term climate forecasts. These would be issued for use in water resource management by a future National Climate Service (see Miles et al. 2006). For any of the approaches noted above, developing problem-specific and regionally tailored atmospheric circulation indices may prove especially useful.

A new awareness among water managers about the impact of climatic change on water supplies has highlighted a need for expanded data sets that capture a much wider range of hydroclimatic and streamflow variability – and their driving mechanisms – than is available in systematically gauged records. Such long-period records will be essential for demand-side analyses, as well as for future scenario modeling. Researchers are already active in developing these data sets for use in water management operations and decision-making via stochastic model runs, compilation of historical meteorological and climatic records (e.g., Mock 2003), reconstructing long records of precipitation, drought and stream flow using tree rings (e.g., MacDonald 2007, Woodhouse and Lukas 2006, Woodhouse et al. 2006), and defining paleo-stage indicators of past extreme floods (see House et al. 2002). Paleo-data studies can also address the role of extreme events in shaping past human-environmental interactions (e.g., Magilligan and Goldstein 2001, Therrell et al. 2004).

As more managers recognize the need for a systematic integration of long-term data into their water management operations for informing water allocation, the next 20 years should be a fruitful field for both modelers and paleo-researchers. In addition, models that integrate surface hydrology and demand, dynamic human geographic data and scenario-driven atmospheric circulations of changing climate will provide another important avenue of research.

Effective cross-disciplinary communication about water issues that can capture all of the contingencies described above will require a new generation of visualizations for integrated assessments, planning, adaptation, and creative outreach across diverse societies and cultures. New maps and other visualizations that can handle multiple dimensions of complexity – including nonstationarity and nonlinearity – will also be needed to generate and articulate theories about the interacting water-related mechanisms and processes that result in the emergence, sustainability, or collapse of geophysical, ecological, and socio-economic systems, both regionally and globally.

Water resources geography's traditional strengths in hydroclimatology and surface water processes at the watershed and regional scale have already laid an excellent foundation for a vibrant research agenda for the next 20 years, but the post-stationary future of hydroclimatology will require real innovation in research approaches. It will be especially important for geographers to continue to carve out unique niches and areas of expertise within the vast climate-change research arena. Climate-based initiatives that address water resources in the context of meteorological and climatic hazards and human-environment interactions such as Weather and Society – Integrated Studies (WAS*IS), http://www.sip.ucar.edu/wasis/, and the National Oceanic and Atmospheric Administration's (NOAA) Regional Integrated Sciences and Assessment program (RISA), http://www.climate.noaa.gov/cpo_pa/risa/, are excellent forums in which to foster stakeholder interactions and opportunities for translational science (e.g., Bales et al. 2004). At the same time, there are many “basic science” and theoretical research questions in need of fresh and creative re-thinking, ranging from how to assign probabilities to a nonstationary stream flow time series, to how to model nonlinearities in hydroclimatic processes, to how to accomplish long-range water resource planning when faced with the specter of abrupt hydroclimatic change – or even “climate surprises” (Overpeck 1996).

It is particularly important to note that the assumption of future nonstationarity precludes an expectation that a single research approach or solution to changing conditions will be all that is required. Effective management of water resources in a post-stationary world must be able to constantly adapt and adjust. Enhanced monitoring of hydroclimatic changes will need to be paired with iterative interactions with stakeholders as old analogs fail and new surprises emerge. This will require a research agenda that has the ability to be timely, flexible, and nimble enough to respond quickly to continually evolving and newly emerging needs.

Closing Remarks

In a changing world that is expected to face a range of future hydroclimatic processes for which our current approaches are ill-equipped, we need to take advantage of all types and manner of hydroclimatic data, invent novel and creative ways to analyze it, and develop powerful and practical ways of communicating and visualizing the results. For this, temporal depth is the necessary companion to spatial detail in any geographic analysis. Integrated assessments that link geophysical, biological, and social sciences across multiple temporal and spatial scales are a necessity for navigating the increasingly complex and interrelated local, regional, and global environments of a post-stationary world. As suggested by Costanza et al. (2007, p 526), “The insight, data and models generated from the close collaboration of environmental historians, archeologists, ecologists, modelers and many others [i.e., geographers] will allow the construction and testing of new ideas about humans' relationship with the rest of nature.”

Author Bio and Contact Information

Katherine K. Hirschboeck is an Associate Professor of Climatology at the University of Arizona in the Laboratory of Tree-Ring Research. She also chairs the University of Arizona's Global Change Graduate Interdisciplinary Program. Her research and teaching address climatology, dendroclimatology, hydroclimatology, and the climatology of extreme events in the past and present – especially the analysis of flood-producing atmospheric processes and tree-growth responses to anomalous atmospheric circulation patterns. Dr. Hirschboeck also holds joint faculty appointments in the departments of Hydrology & Water Resources, Geography & Regional Development, and Atmospheric Sciences. She can be reached at: The Laboratory of Tree-Ring Research, University of Arizona, Tucson AZ 85721, email: katie@ltrr.arizona.edu.


Inspiration and support for this essay was provided by the Climate Assessment for the Southwest (CLIMAS) (NOAA Cooperative Agreement no. NA07OAR4310382). Special thanks go to Dr. Mike Crimmins of the University of Arizona who provided valuable insights about adaptive management needs.