Hydrology, Cryosphere & Earth Surface Opinion

How Can We Better Understand Low River Flows as Climate Changes?

When rivers run low, they threaten ecosystems, economies, and the communities who depend on them. Scientists need to determine how climate change alters this process, but to do so, they'll have to abandon a long-held assumption.

By , Erin Towler, and Ben Livneh

Rivers—the lifeblood of society—face unprecedented threats from the world’s changing climate. In particular, scientists expect that rivers in many regions will run lower than ever before and for longer spans of time.

In light of this, the study of low river flows, or low-streamflow hydrology, is critically important to society. A firm understanding of low-streamflow hydrology can help resource specialists manage, for example, municipal water supply, irrigation, industry allocations, river navigation, recreation, and wildlife conservation [Smakhtin, 2001].

Despite how low flow has direct ties to water scarcity and drought, relatively few studies evaluate how climate change will affect low flows. Low-flow studies have typically been grounded in the principle of stationarity—the idea that natural systems vary within a known, unchanging range of variability.

But this assumption no longer holds given that climate change has raised average temperatures to extreme highs not seen in millions of years [Milly et al., 2008]. We urgently need a better understanding of the changing behavior of low-flow conditions to inform sustainable water management and protect against potential risks and impacts.

The Challenge of Low Flow

The limited research on the phenomenon of low flow can be partially attributed to its unique challenges. First, there is the data challenge: Many basins lack hydrological monitoring stations, and human activities such as filling and drawing down reservoirs and extracting water for irrigation can affect data quality. Next, low flows are strongly influenced by groundwater interactions with surface water, for which data are difficult to obtain. Analyses therefore rely on models.

Finally, what someone defines as “low flow” will vary depending on water needs. For example, farmers, ranchers, and water resource planners may be concerned with average 7-day or 30-day minimum flow, whereas city planners and engineers may be interested in a particular exceedance probability threshold (e.g., the low flow occurred 25%, 10%, or 5% of the time).

Several tools can shed light on these issues from both scientific and policy-making perspectives. Each has shortcomings that we must work steadily to improve. But together they can advance the science of low flows so that water managers can make more informed decisions.

The Power of Hydrological Models

The first tool beyond the raw data scientists should turn to when studying low flows is hydrological models—computer simulations of water movement in the environment. These models incorporate characteristics of the land surface, such as the topography, vegetation, and soil. They also include the physical processes that govern the amount of water going in and out of the land, such as soil moisture, evaporation, plant transpiration, and runoff.

Such models already allow scientists to address changes in land use. But by combining these parameterizations with precipitation and temperature information, they can also assess low flow under changing climates.

These models can also partially alleviate issues with data availability. First, they work over areas with limited data coverage, interpolating over regions with missing data. They also offer a means to reconstruct data that are altered by humans via reservoirs and irrigation, predicting the “natural” flow by simulating streamflows solely on the basis of meteorology. Such predictions are important given that our inspection of more than 9000 U.S. Geological Survey (USGS) stream gauges in the continental United States revealed that more than three-fourths of them have been altered by reservoir storage.

Finally, if one wants to consider the impacts of changing land cover, physically based models offer the only way to estimate future low flows on the basis of future climate change projections.

Although a limited number of studies have successfully used hydrological models for low-flow research [e.g., Wilby and Harris, 2006; Demirel et al., 2013] and some have even included changing climate conditions [Vaze et al., 2010], the potential remains largely untapped.

Room for Improvement

Second, although scientists have done a lot of work to calibrate hydrologic models under average flow conditions or even in flood scenarios, they have done very little work to attempt to calibrate their models when rivers and streams run low. Model calibration is already limited by the availability of high-quality data and therefore involves some reconstruction of regions with missing data. This process, called regionalization, fills in empty regions with parameters from nearby basins or basins that have similar physiographic features (e.g., topography, vegetation, soils).

Scientists often use regionalization to estimate how often rivers flood but do not have an analogous approach to calibrate predictions of how often they run dry. Most efforts, including those of the USGS, are focused on average conditions. But we think that a comparable approach could be developed for low-flow characteristics, which would be a great improvement.

Accounting for a Changing Climate

The impact of climate on low river flows is visible both today and in the historical record on time scales as small as decades [Lins and Slack, 1999]. Further, low flows are influenced by recurring climate cycles due to fluctuations in the ocean-atmospheric system, such as the El Niño–Southern Oscillation and the Pacific Decadal Oscillation.

Therefore, an integral part of investigating nonstationarity in low-flow features also lies in identifying the influence of climate variability at decadal to multidecadal scales, as previous studies have indicated [Stahl and Hisdal, 2004]. However, for many stream gauge locations, for example, in the United States and elsewhere, data are available for only a limited number of years, which may not span the recurrence interval and duration of such fluctuations.

Locations of stream systems that may soon experience significantly low flows. Color bubbles indicate location of the stations and loss-of-flow estimates in cubic feet per second (cfs) per day per year. The size of the bubble is proportional to the magnitude of trend value. The image was generated using data from the U.S. Geological Survey's Hydro-Climatic Data Network. Credit: Maryam Pournasiri Poshtiri and Indrani Pal
Locations of stream systems that may soon experience significantly low flows. Color bubbles indicate location of the stations and loss-of-flow estimates in cubic feet per second per day per year. The size of the bubble is proportional to the magnitude of trend value. The image was generated using data from the U.S. Geological Survey’s Hydro-Climatic Data Network. Credit: Maryam Pournasiri Poshtiri and Indrani Pal

Once studies specifically address how climate influences low river flow on broad to local scales, the research community needs to develop a way to incorporate those findings into how it designs and reports indices of low flow. This step is critical because current estimates of low-flow characteristics do not reflect these shifting risks—rather, they are “stationary.”

For instance, one of the most common low-flow indices in the United States is the minimum flow that occurs, on average, every 10 years, as measured over a 7-day period (10q7). Originally developed to regulate stream pollution in the 1970s, 10q7 has since been used widely in hydrology.

However, because of periodic shifts in the ocean-atmosphere system and because the Earth’s climate is entering new territory, the current practice of estimating a single 10q7 value may be inappropriate because it does not accurately reflect the shifting risk.

One tool that can be applied to this issue is extreme value theory [Coles, 2001], a branch of statistics uniquely geared toward modeling minimum (or maximum) values. Extreme value theory is well suited to studying extremes in hydrology [Katz et al., 2002] because it develops techniques to model extreme events rather than average conditions.

Scientists have already used extreme value theory to look at changing low flows in Europe [Feyen and Dankers, 2009]. The tool also offers a means to include trends or climate information such as El Niño to quantify how minimum flows vary from year to year. Incorporating trends or climate indices has not been done in the study of low flows but offers a promising new way to account for nonstationarity.

Connecting Scientists with Policy Makers

To truly advance low-flow science in a changing climate, research directions and outcomes must also reflect the needs and concerns of decision makers. The successful use of climate knowledge requires some iteration between the scientists who produce knowledge and those who must use it to make decisions and set policies [Dilling and Lemos, 2011].

Several research initiatives in this vein are building the capacity to prepare for and adapt to climate variability and change, such as the Western Water Assessment, one of the National Oceanic and Atmospheric Administration’s Regional Integrated Sciences and Assessments. This and other key efforts aim to develop closer partnerships between scientists and engineers [Tye et al., 2015] and are prime examples that can guide future low-flow research.

We challenge the scientific community, together with engineers, regional water managers, local drought planners, and state agencies, to more actively engage in low-flow discourse. Only together can we advance the science of low flows, develop tools to manage the changing risks of low flows, and achieve sustainable water management in a changing climate.


The National Center of Atmospheric Research is funded by the National Science Foundation, and support for E.T. was provided by NSF award 1048829. We extend our thanks to Maryam Pournasiri Poshtiri for helping us create the map.


Coles, S. (2001), An Introduction to Statistical Modeling of Extreme Values, 208 pp., Springer, London.

Demirel, M. C., M. J. Booij, and A. Y. Hoekstra (2013), Effect of different uncertainty sources on the skill of 10 day ensemble low flow forecasts for two hydrological models, Water Resour. Res., 49(7), 4035–4053, doi:10.1002/wrcr.20294.

Dilling, L., and M. C. Lemos (2011), Creating usable science: Opportunities and constraints for climate knowledge use and their implications for science policy, Global Environ. Change, 21(2), 680–689.

Feyen, L., and R. Dankers (2009), Impact of global warming on streamflow drought in Europe, J. Geophys. Res., 114, D17116, doi:10.1029/2008JD011438.

Katz, R. W., M. B. Parlange, and P. Naveau (2002), Statistics of extremes in hydrology, Adv. Water Resour., 25, 1287–1304.

Lins, H. F., and J.  R. Slack (1999), Streamflow trends in the United States, Geophys. Res. Lett., 26(2), 227–230.

Milly, P. C. D., J. Betancourt, M. Falkenmark, R. M. Hirsch, Z. W. Kundzewicz, D. P. Lettenmaier, and R. J. Stouffer (2008), Climate change: Stationarity is dead: Whither water management?, Science, 319, 573–574.

Rosenberg, E. A., E. A. Clark, A. C. Steinemann, and D. P. Lettenmaier (2012), On the contribution of groundwater storage to interannual streamflow anomalies in the Colorado River basin, Hydrol. Earth Syst. Sci. Discuss.9(11), 1475–1491.

Smakhtin, V. (2001), Low flow hydrology: A review, J. Hydrol., 240(3–4), 147–186.

Stahl, K., and H. Hisdal (2004), Hydroclimatology, in Hydrological Drought: Processes and Estimation Methods for Streamflow and Groundwater, Dev. Water Sci., vol. 48, edited by L. M. Tallaksen and H. Van Lanen, pp. 19–51, Elsevier, Amsterdam.

Towler, E., M. Roberts, B. Rajagopalan, and R. Sojda (2013), Incorporating probabilistic seasonal climate forecasts into river management using a risk-based framework, Water Resour. Res., 49, 4997–5008, doi:10.1002/wrcr.20378.

Tye, M. R., G. Holland, and J. M. Done (2015), Rethinking failure: Time for closer engineer-scientist collaborations on design, Proc. Inst. Civ. Eng. Forensic Eng., doi:10.1680/feng.14.00004.

Vaze, J., D. A. Post, F. H. S. Chiew, J. M. Perraud, N. R. Viney, and J. Teng (2010), Climate non-stationarity—Validity of calibrated rainfall–runoff models for use in climate change studies, J. Hydrol.394(3), 447–457.

Wilby, R. L., and I. Harris (2006), A framework for assessing uncertainties in climate change impacts: Low-flow scenarios for the River Thames, UK, Water Resour. Res., 42, W02419, doi:10.1029/2005WR004065.

—Indrani Pal, Department of Civil Engineering, College of Engineering and Applied Science, University of Colorado, Denver; email: [email protected]; Erin Towler, National Center of Atmospheric Research, Boulder, Colo.; and Ben Livneh, Cooperative Institute for Research in Environmental Sciences, University of Colorado, Boulder

Citation: Pal, I., E. Towler, and B. Livneh (2015), How can we better understand low river flows as climate changes?, Eos, 96, doi:10.1029/2015EO033875. Published on 6 August 2015.

© 2015. The authors. CC BY-NC 3.0
  • Sara

    which hydrological model can be used to estimate low flow for ungauged basins?

  • Gerald Kauffman

    I agree, extreme value theory is a useful tool in evaluating the effects of low flow (drought) hydrology of river systems. We used EVT to satisfactory effect to evaluate the frequency and intensity of extreme drought in the Delaware Basin from 1600-2002 (Kauffman and Vonck, 2011).