Channel Networks and Watersheds

Streams naturally assemble themselves into surprisingly well-organized (quasi-fractal) networks. Figure 2 shows a typical channel network where many small streams converge to make progressively larger streams. The smallest streams in the network, which have no other streams flowing into them, are referred to as first order streams. When two first order streams meet, a second order stream is formed. When two second order streams meet, they form a third order stream, and so on. According to this conventional stream ordering system, first developed by Horton (1945) and refined by Strahler (1957), when a smaller order stream (e.g., first order) meets a larger order stream (e.g., second order), the resulting stream retains the order of the larger stream (in this case, second order).

Each stream has a watershed, also known as a ‘river basin’ or ‘catchment’ because it is the land that ‘catches’ precipitation and funnels it towards the stream. The watersheds of two first order streams are outlined with grey dashed lines in Figure 2. The watershed of a second order stream is outlined in black dashed lines and encompasses the two first order watersheds. The solid black outline in Figure 2 shows the watershed boundary for the fourth order watershed, which encompasses all other watersheds nested within it. The right side of Figure 2 shows the Mississippi River watershed highlighted in green, with the Missouri River watershed nested within it, highlighted in orange. By the time the Mississippi River reaches New Orleans, it is a tenth order stream (though only a few of its largest tributaries are shown in Figure 2), and drains more than one-third of the contiguous US.

The concept of connectivity between rivers and their watersheds will come up again towards the end of this module in the context of restoration. If a particular stretch of stream is impaired for one function or another (e.g., fish habitat has been degraded), in some cases it makes sense to ‘fix’ that specific stretch of river, while in other cases the impairment is simply a symptom of problems higher up in the watershed, so the ‘fix’ may need to be applied at that location in the watershed before human intervention or natural processes can begin to repair the impaired stream. Such is the way that watersheds and streams are connected.

Figure 2. Left panel shows a stream network colored and labeled according to stream order. Dashed grey, dashed black and solid black lines indicate watershed boundaries. Right panel shows entire Mississippi River watershed (green) with the Missouri River watershed (orange) nested within it.

Source: Images Patrick Belmont

Watersheds are Complex Systems

When you look around, you see that the world is full of systems…assemblages or combinations of things that form a functional unit. Some systems are human-made, others are made by nature. Some systems are simple, meaning the way they work is straightforward and the outputs from the system are easily predictable. Other systems are complex, meaning they often have many parts that interact, often in non-linear ways, making the outputs from those systems more difficult to predict.

For example, a coffee maker is a pretty simple system. You put in 8 cups of water and two cups of coffee grounds and (assuming you put them in the right places), you turn the machine on and get ~8 cups of coffee. If you change the amounts of either of the inputs, it is pretty easy to predict the impacts on the coffee you brew.

Watersheds are not such a simple system. They are incredibly complex. One example can be seen in how the relationship between rainfall and runoff changes throughout the year. In a simple system, you would expect a constant relationship between incoming rainfall and outgoing flow. For example, a 1-inch rain event should translate to a stormflow hydrograph that might last 2 days and peak at 1000 cfs. But this isn’t what we see. Figure 3 shows streamflow (blue line, values on the left axis) and precipitation (orange bars, values on the right axis) from March through September 2008 for the Maple River near Rapidan, Minnesota. Precipitation is relatively evenly distributed throughout the year. As you can see, in April and May, rainfall events that are 0.5 to 1 inch result in relatively high flows (1000 to 1500 cfs). However, in July, August, and September, similar rainfall events hardly elicit any flow response whatsoever! Why do we see such non-linear behavior?

Figure 3. Streamflow and precipitation measured for the Maple River near Rapidan, Minnesota.

Source: Data from the Minnesota Pollution Control Agency. Plot created by Patrick Belmont.

Watersheds comprise many interacting parts. Figure 4 (top panel) is one way to represent various ‘parts’ that might be considered to comprise the watershed. While this is clearly a very simple view of this complex system, it is useful to take a “crude look at the whole”, a term coined by Nobel Prize-winning Physicist Murray Gell-Mann, as a starting point. When one component of the system is systematically changed, it may have direct as well as indirect impacts that propagate through the system. For example, changes in precipitation, snowmelt regime, or water storage may change streamflow. This altered streamflow has direct effects on river channel morphology, sediment transport, riparian vegetation, water quality, nutrient processing, and biodiversity, as indicated by the yellow arrows in the middle panel of Figure 4. But there are other interactions within the system, feedbacks that are indicated by purple arrows in the bottom panel of Figure 4. So to predict impacts of the changes in flow on aquatic biodiversity you would have to take into account not only the direct effects (yellow arrow between flow and aquatic biodiversity, but also the indirect effects associated with changes in channel morphology. This concept is also relevant in the context of watershed ‘restoration’. If a particular stretch of stream is impaired for one function or another (e.g., fish habitat has been degraded), in some cases it makes sense to ‘fix’ that specific stretch of river, while in other cases the impairment is simply a symptom of problems higher up in the watershed, so the ‘fix’ may need to be applied at that distant location in the watershed before human intervention or natural processes can begin to repair the impaired stretch of stream.

These notions of complex feedbacks and cascading effects greatly complicate the process of predicting what impacts human activities or natural disturbances within a watershed might have downstream. We’ll come back to this theme of system dynamics and complexity throughout the course.

Figure 4. Top panel shows various components that can be considered part of a watershed system. The middle panel illustrates components of the system that might be directly impacted by a change in streamflow. The bottom panel highlights other linkages (feedbacks) within the system that may amplify or dampen effects of the streamflow changes on the various components.

Source: Patrick Belmont

Streams

Streams are the most obvious way that water is moved through a watershed because we see them all over. But there are many other means by which water moves, as discussed in module 2. Figure 5 illustrates the various stocks (places were water is stored, even if only temporarily) and fluxes (mechanisms by which water moves) of water that may exist within any given watershed. For example, one raindrop might fall onto vegetation (called interception) and subsequently be evaporated back up into the atmosphere. Another raindrop might fall onto the soil surface and then runoff the surface into the stream channel or it might infiltrate down into the soil. Once in the soil, the water might further percolate down into the groundwater, where the soil or rock is saturated with water. Alternatively, once in the soil, the water might travel downhill within the soil and runoff into the stream or it might be taken up by vegetation and transpired back into the atmosphere. Estimating and predicting which, and to what extent, water travels through these pathways is an active field of hydrologic research and is also vitally important for environmental management and policymaking, as certain pathways may be more or less prone to filtering or polluting water along its journey to the place where you might want to use it for drinking, irrigating, fishing, swimming or the myriad other purposes for which we need water.

Figure 5. The water cycle represented at the scale of an individual watershed, from incoming precipitation (top) to the outlet of the stream at the bottom of the watershed (lower right). Arrows represent fluxes of water that can be transferred through the various stocks (places where water is at least temporarily stored). Streams can receive water via direct inputs of precipitation, runoff from the surface or (soil) subsurface as well as seepage from groundwater.

Source: Patrick Belmont

Hydrograph

A hydrograph is a graph of discharge over time. The time period shown could be short, for example, the flow resulting from an individual rain storm, or it could be long, for example, a continuous record of flow over many decades. While numerous federal and state agencies, corporations, and individuals monitor discharge in streams throughout the country, the US Geological Survey is the chief entity charged with monitoring streamflow, maintaining over 9,000 stream gages, most of which record water discharge in 15 minute intervals and many of which also include water quality data. Visit the USGS Water Resource webpage (water.usgs.gov) and peruse the wealth of information compiled to assess water resources. Exercises utilizing these data are included below in module 3 as well as module 4.

The Figure 4 shows example hydrographs from the Logan River, near Logan, Utah for two different water years (2006 and 2012). The water year begins October 1 and ends September 30. Hydrologists often prefer to conduct analyses based on the water year rather than the calendar year to facilitate comparison of incoming precipitation and outgoing streamflow, and specifically to ensure that snow delivered in October, November, or December is accounted for in the same time period that it is likely to melt, which may be in spring or summer of the following calendar year.

The Logan River hydrograph shows a long (about 5 month) prominent peak in discharge, primarily driven by snowmelt, with many other smaller peaks superimposed (from accelerated snowmelt during warm periods or rain events). The hydrograph of the Logan River over a 50 year time period (Figure 6) shows the prominent peak from snowmelt each year, but provides little information about the smaller scale variability that is visible on the annual timescale. Note the non-linear y-axis of the plots. Such axes can be useful for visualizing detail in both high and low flow conditions, whereas the detail in low flows would not be visible on (typical) linear axes. The apparent shift in low flows circa 1970 on the Logan River was caused by removal of a water diversion upstream from the gauge. Note that there is a considerable amount of ‘noise’ (i.e., variability) in streamflow over the past 50 years. This variability is not random, but rather has some ‘structure’ to it, some of which is visibly obvious (annual peaks) and other portions that can only be quantified using advanced analytical or statistical techniques, which are beyond the scope of this course, but currently represent a vibrant facet of hydrologic research.

Figure 6. Hydrograph for the 2006 water year (October 1, 2005 through September 30, 2006) for the Logan River, near Logan Utah.

Source: USGS

Figure 7. Hydrograph for the 2012 water year on the Logan River.

Source: USGS

Figure 8. Hydrograph for approximately 50 years of flow on the Logan River, near Logan, Utah.

Source: USGS

Examples of Logan River Hydrographs 2008-2009

Figure 11. 2008 Logan River Hydrograph

Source: USGS

Figure 12. 2009 Logan River Hydrograph

Source: USGS

River Flow Regimes

The temporal patterns of high and low flows are referred to collectively as a river’s flow regime. The flow regime plays a key role in regulating geomorphic processes that shape river channels and floodplains, ecological processes that govern the life history of aquatic organisms, and is a major determinant of the biodiversity found in river ecosystems. There are five components that characterize the flow regime:

1. Magnitude: the total amount of flow at any given time
2. Frequency: how often flow exceeds or is below a given magnitude
3. Duration: how long flow exceeds or is below a given magnitude
4. Predictability: regularity of occurrence of different flow events
5. Rate of change or flashiness: how quickly flow changes from one magnitude to another

River in regions with similar climate, geology, and topography tend to have similar flow regimes. For example, rivers draining high mountains, such as the Logan River, tend to have relatively infrequent, high magnitude, long duration, and predictable flood events that have a slow rate of change (Figure 6 on the previous page). Rivers in many tropical climates have similar flow regime characteristics as mountain rivers, due to predictable rainy and dry seasons. In contrast, rivers in arid regions are often characterized by high magnitude, short duration floods of low predictability and high flashiness (e.g., Figure 11 on the next page).

Within regions of similar climate, local factors such as soil type, soil depth, vegetation cover, and watershed size influence the natural flow regime. For example, watersheds with deep, permeable soils will be able to absorb more precipitation than watersheds with thin, impermeable soils, and will thus tend to have less flashy floods of lower magnitude and longer duration. Large rivers tend to be less flashy than small streams, which respond more quickly to individual precipitation events. Thus, natural flow regimes can be somewhat variable between nearby watersheds. Also, although general patterns in flow regime can be determined from watershed characteristics, yearly variation in precipitation patterns means that many years of flow monitoring will be required to fully characterize the flow regime of individual rivers.

Temporary vs. Perennial Streams 

Most large rivers are perennial, meaning they maintain flow throughout the year. However, many headwater streams or streams in arid regions sometimes run dry. A stream is considered temporary if surface flow ceases during dry periods. Temporary streams are often classified further as intermittent and ephemeral. An intermittent stream becomes seasonally dry when the groundwater table drops below the elevation of the streambed during dry periods. A spatially intermittent stream may maintain flow over some sections or surface water in deep pools even during dry periods due to locally elevated water tables or perched aquifers. An ephemeral stream only flows in direct response to precipitation such as thunderstorms. Thus, the flow variability of an intermittent stream is much more predictable than in an ephemeral stream.

In many parts of the world, such as the desert southwest, temporary streams may comprise a majority of the river network, >80% in some areas. However, even in wet regions, temporary streams at the head of river networks can account for >50% of the total stream network. Thus, river networks can be considered dynamic systems, with total miles of surface flow expanding and contracting in response to precipitation events.

Figure 15. The San Rafael River watershed, located in southeastern Utah. Left figure shows only perennial streams. Right figure shows perennial, intermittent, and ephemeral streams. Note the large proportion of the network comprised of temporary stream channels.

Source: Brian Laub

Why would we still call a channel that goes dry for much of the year a stream? In other words, how can we distinguish between a temporary stream and an upland terrestrial ecosystem? In short, a stream has characteristic hydrological, geomorphological, and ecological processes. However, as with many topics in environmental science, the distinction between stream channels and uplands and between perennial streams and temporary streams is often fuzzy and scale-dependent. Individual stream channels may hold water for decades and then become dry during exceptional droughts that occur infrequently (once every 50-100 years). Similarly, small gullies on hillsides may flow only a few days of the year and may transport sediment but not be resident to aquatic life. Are such systems part of the river network?

Flow Duration Curve

While it can be very informative to study hydrographs and the other flow metrics described above, often an important question often asked about rivers is ‘what percentage of time does flow exceed (or not exceed) a given value (e.g., 100 cfs)?’ It might be important to answer that question to determine the percentage of time when the flow is too low to support a particular fish species. Or it may be important to know what percentage of time the river exceeds a certain value known to cause flood damage. The proportion of time any given flow is exceeded can be determined by generating a flow duration curve. Figure 21 shows the flow duration curve for the hydrograph shown in Figure 21 (2006 water year) as well as the three subsequent years. You can immediately see that the mid and lower flows (exceeded about 40% (or 0.4) of the year) are relatively similar in each year, but the larger flows exhibit quite a bit of variability. In 2007 the highest flow of the year was only a bit over 400 cfs, while it was over 1500 cfs in 2006. The flow that was exceeded 20% of the time (0.2 on the x-axis) was approximately 450 cfs in 2005, but only 200 cfs in 2007.

Figure 22. Flow duration curves for the Logan River, in northern Utah for four different years. As the Logan River is a snow-melt dominated system, observed differences in high flows each year (left side of plot) correlate to differences in snowpack in each of the years.

Source: Patrick Belmont

Note that this plot provides detailed information on different parts of the flow duration curve depending on whether you use linear or log scales for the x or y axes (see example from the Stilliguamish River, Washington below in Figures 22-25).

Figure 23. Stilliguamish River, Washington

Source: USGS

Figure 24. Stilliguamish River, Washington

Source: USGS

Figure 25.Stilliguamish River, Washington

Source: USGS

Figure 26. Stillaguamish River, Washington

Source: USGS

Figure 27. Root River, Minnesota

Source: USGS

Flow duration curves can be made for a given river over two different time periods to illustrate if/how the range of flows has changed over time. For example, Figure 27 shows flow duration curves for the Le Sueur River in southern Minnesota for two different time periods (1950-1970 in blue, 1990-2010 in red). Note that in these plots the fraction of year exceeded is labeled as ‘exceedance probability’. These two terms are interchangeable, both being computed as:

$$Ep=\frac{R}{\left(n+1\right)}$$

Where E_p is the exceedance probability or the fraction of the year that a given flow is exceeded, R is the rank, and n is the total number of values (365 if you are using daily-averaged flow values for a non-leap year). High flows (toward the left side of each plot) and low flows (toward the right side of each plot) appear not to have changed in the Elk and Whetstone rivers. In the Blue Earth River, low flows (exceeded more than 85% of the time) have not changed much, but mid-range and high flows all appear to have increased. In the Le Sueur River, the full range of flows appears to have increased. Note that the y-axis is plotted on a log scale, so even the modest difference between the two curves represents a significant increase in high flows (e.g., those that are only exceeded 5-10% of the time). The Root River, in southeastern Minnesota, has experienced significant increases in high and low flows within the past two decades, see example above.

Figure 28. Flow duration curves are shown for two different time periods in the Le Sueur River, southern Minnesota. Changes in flows are attributed to changes in artificial drainage of agricultural fields, as well as changes in precipitation and crop type. See Schottler et al., (2014) Hydrological Processes for more information.

Source: Patrick Belmont

Number of Channels and Sinuosity

While the variety of river types is best thought of as a continuum, rather than a bunch of discrete boxes, it is often useful in science to create a taxonomy to classify items for the purpose of description and communication. Figure 30 illustrates some of the most common characteristics by which rivers can be classified (see Brierley and Fryirs, 2005 or Montgomery and Buffington, 1997 for detailed discussions of channel classification). At the most basic level it is useful to classify rivers according to the number of channels they contain, from single-threaded to braided (with more than three interweaving channels that are frequently reorganized) to anastomosing (which typically have somewhat stable, vegetated islands between channel threads), to discontinuous streams that have un-channelized reaches). Wandering rivers are those that alternate between single-threaded and slightly braided reaches. Another useful metric, particularly for single-threaded channels is sinuosity, which is calculated as the length along the river divided by the straight-line distance along the river valley. Rivers can have sinuosity ranging from one up to three (i.e., the river length is three times longer than the valley). Bends in rivers are called meanders. Meanders can exhibit a variety of forms with some in nature being remarkably regular (see the Fall River in Rocky Mountain National Park in Google Earth) and others being irregular or tortuous (frequently folding back on itself).

Figure 30. The number of channels and degree/type of sinuosity is two common metrics for characterizing rivers. The number of channels includes: single, wandering, braided, anastomosing, and discontinuous. The Degrees of Sinuosity are straight, low sinuosity, and sinuous. The types of sinuosity are regular, irregular, and tortuous. 

Source: Patrick Belmont, after Brierley and Fryirs, 2005.

Stream Power

While there is currently no generalizable equation or universal law describing what a river channel should look like, a vast array of field data and modeling has culminated in some useful generalities. Stream power, defined as the product of water density (about 1000 kg/m³), gravitational acceleration (9.8 m/s²), discharge (m³/s), and channel slope (m/m), is one useful predictor of channel form and dynamics because it quantifies the amount of ‘work’ that can be done by a stream, such as moving sediment on the bed or in the banks of the river (i.e., erosion or sediment transport). For example, braided rivers tend to have more stream power than single threaded meandering rivers because their channel slope tends to be higher as they often flow closer to mountains (on steeper topography).

Sizes of a River Channel

River channels are self-formed. Typically they are only partially filled, the water level is well below the tops of the banks. Sometimes, they are overfilled and water spills out onto the floodplain. These simple observations lead to the fundamental question, ‘what sets the size of a river channel?’ Figure 30 conceptually illustrates the rationale supporting the empirical finding that an ‘effective discharge’, which occurs frequently enough and has sufficient power to do work, ultimately dictates the size of the channel. Specifically, the brown curve illustrates that the frequency distribution of discharge in a river is typically right (positively) skewed, meaning that relatively low discharges are quite common and increasingly higher discharges occur with diminishing frequency. There is some discharge below which sediment does not move on the river bed because there is insufficient power to move the sand or gravel, as indicated by the light orange line starting at some moderate discharge and increasing in a non-linear manner at progressively higher discharge. Multiplying the brown and light orange lines together yields the darker orange line, which has a peak at some relatively high discharge value. This ‘effective discharge’ tends to occur when the river is approximately full up to its naturally formed banks. Even very large floods, which greatly exceed the capacity of the channel, do not necessarily add a proportionate amount of power to the channel because much of the additional water (and therefore the energy to transport sediment) is dissipated on the adjacent floodplain.

As changes in climate alter precipitation patterns or as land and water management modulates the proportion of precipitation that becomes streamflow, the frequency curve in Figure 30 may change and thus change the effective discharge as well as the geometry of the channel. In this way, rivers are dynamic features in the landscape, growing bigger when more water is flowing through the landscape and smaller during extended drier periods.

Figure 31. Conceptual figure illustrating the concept of effective discharge, which occurs frequently enough and moves a sufficient amount of sediment such that it often determines the size (width and depth) of a river channel.

Source: Patrick Belmont, after Leopold and Wolman, 1960.

Forecasting and Predictions

Meteorologists have made excellent progress in the past few decades to improve our abilities to forecast when rain events might occur over the next week or so, which facilitates the short-term forecasting of floods and droughts discussed in the paragraph above. However, complex atmospheric dynamics prevent forecasts beyond more than a few weeks in advance. Nevertheless, we must have some basis for making decisions about development, infrastructure and agriculture (e.g., How big should we build a culvert under a road? What size retention basin is needed next to a new housing development? Which agricultural fields will require artificial drainage and which will require irrigation?).

For these longer-term predictions we can use statistics to determine how likely it is that a given location will experience, for example, more than 10 cm of rain in a day, or less than 5 cm of rain during a given month. Many million- and billion-dollar decisions about development and infrastructure are based on such predictions.

To make these predictions, hydrologists synthesize historical data and use a statistics-based approach to determine the likelihood that a given event might occur. While Figure 1 highlights the ‘messiness’ of precipitation events over time, reorganization of the data provides useful information. Figure 2 shows a histogram of the precipitation data presented in Figure 1. A histogram is a plot showing the number of events that fall within chosen data groups or “bins” (shown on the x axis). From these data you can quickly determine that Kingston, NY experiences no rain about 2 out of every 3 days (731 out of the total 1096 days in this record). Only 10 days in the record had rainfall that exceeded 6 cm, so from these data alone you would expect such large rainfall events to happen 10 days out of 1096, or about 1% of the time. On 210 days during this time period the amount of rainfall was between the minimum measurable (typically 0.025 cm or 0.01 inch) and 1 cm (0.4 inches).

Hydrologists tend to use the term ‘forecast’ when referring to a future projection for which we have a lot of information (and therefore relatively high certainty of when an event might occur and what magnitude it might be). In contrast, hydrologists use the term ‘prediction’ for future projections for which less information is available, and therefore uncertainty is greater.

Figure 2. Histogram showing the number of events (specifically, total rainfall measured for each day from October 1, 2010 to September 30, 2013) in each of the bins indicated on the x axis. Bin labels are simplified for illustration purposes. For example, the bin labeled 0-1 includes all 210 days where precipitation was between 0.025 cm (minimum measurable) and 0.999 cm.

Click for a text description of Figure 2

Daily precipitation categorized by precipitation amount.

Daily Precipitation (cm)	Number of events
0	731
0-1	210
1-2	70
2-3	43
3-4	14
4-5	8
5-6	10
6-7	2
7-8	3
8-9	2
9-10	1
10-11	1
11-12	0
>12	1

Source: Patrick Belmont, data from NOAA National Climatic Data Center, as used in Figure 1 on the previous page.

Predicting Frequency and Magnitude

We can do similar calculations to estimate the frequency and magnitude of floods. For example, Figure 3 shows the annual maximum series of flows for the Lehigh River at Bethlehem, PA from 1910 to 2013. The annual maximum series is simply the highest flow value recorded for each year (typically for the water year, October 1 – September 30 as discussed in module 3). The largest flow on record (92,000 cfs) occurred in 1942 (on May 23, to be exact). Note that these are daily flow values which average flow over the course of the entire day, so the actual peak that occurred on May 23 was probably slightly higher. Interestingly, the year that had the smallest peak discharge on record was April 6, 1941 (only 8210 cfs, less than 10% of the whopper flood that came through the following year). But note that we are still talking about the highest flow for that particular year, which is still considerably higher than the average annual flow which is about 1,300 cfs. How can we use this information to predict what might happen in the future to support decisions about development, flood risk, or factors that may influence biota in the river or floodplain (e.g., factors limiting the return of certain fish)?

Similar to the precipitation example above, it helps to reorganize the data. The histogram shown in Figure 4 illustrates the frequency of events in 19 different “bins”. For example, the Lehigh River at Bethlehem, PA has never experienced an annual peak flow less than 5,000 cfs, so there is no orange bar in that bin. But it has experienced 4 peak flows that fall between 5,000 and 10,000 cfs. It has experienced a total of 55 peak flows that fall within the range of 10,000 to 25,000 cfs. These are the most common peak flows for the Lehigh River. And you can see the two outliers at the high end (92,000 cfs in 1942 and 91,300 cfs in 1955).

Typically, hydrologists will approximate the distribution of events that you see in Figure 4 as a probability density function, represented by the smooth, dashed-grey curve superimposed on the plot. This allows them to integrate under the curve above a value of interest to determine the probability that a flood of that magnitude (or larger) will occur. For example, integrating to find the area under the grey dashed curve above 60,000 cfs in Figure 4 would tell you the probability in any given year (in the future) that a flood of that magnitude (or larger) will occur. From that information, you can build your bridge to the appropriate size, or set your flood insurance rates accordingly, etc. Calculating probabilities associated with floods of a given magnitude is discussed in greater detail in the exercise associated with this module.

The same general principles apply to studying droughts. However, as we discuss below, droughts are more difficult to define and quantify because they build up over time and vary immensely over any given area. Nevertheless, similar plots of drought frequency, severity and duration can be developed for droughts, similar to Figure 4, to make sense of all the messy variability we observe in water deficiencies. With this brief introduction hopefully you can appreciate the great challenges in predicting the occurrence and severity of floods and droughts, and you can begin to see the implications for socio-economic systems and ecosystems.

Figure 4. Orange bars show the histogram of the annual peak flows shown in Figure 3 for the Lehigh River. Grey dashed line is a theoretical distribution (a probability density function, or PDF) approximating the complete distribution represented by the histogram of historical data. The PDF can be used to make predictions of flood frequency.

Source: Patrick Belmont, data from US Geological Survey

Hydrologic Versus Hydro-Geomorphic Perspectives

Scientists tend to think about floods in two different (but related) ways, one being strictly hydrologic and the other requiring an evaluation of the floodplain topography and how different flow regimes might impact the channel and surrounding areas. In module 3, we explored how hydrologic analyses (analyzing the patterns such as the frequency, duration, and magnitude of flood events) could be used to characterize the river flow regime (e.g., how often does the river exceed 800,000 cfs?). However, a hydrologic analysis does not provide information about the extent or duration of flooding across the landscape (e.g., which parts of the natural floodplain or streets will be flooded?). To predict how much of the floodplain might be inundated by a given flow, we need to consider the channel and floodplain topography (a hydro-geomorphic analysis). For example, a river system with low channel banks and a broad, flat floodplain will experience more frequent flooding of greater extent than a river system with tall banks and a narrow floodplain, given the same flow regime.

Because the size of a river channel can change over time, the relationship between the hydrologic flood frequency and hydro-geomorphic mapping of the area inundated may also change, as discussed in the later module section on hydrologic non-stationarity. For example, the Minnesota River (a major tributary of the Mississippi River) has widened by nearly 50% in the past 3-4 decades. Therefore a flood that may have inundated a significant amount of floodplain 50 years ago may now be entirely conveyed within the channel itself. Thinking back to the example of the Lehigh River in Figures 4 and 5 it is very likely that the 1941 flood (the lowest on record) did not fill the channel and inundate the floodplain. 1943 and 1944 had moderately high peak flows, but may also not have gotten out of the channel because the massive 1942 flood would have widened and deepened the channel. Over the following years, the channel would likely have narrowed again, in response to relatively smaller floods. Flood frequency analysis discussed below and in the exercise associated with this module, is strictly a hydrologic analysis. A hydro-geomorphic analysis is needed to estimate the risk of flood damage. Both types of analyses may be important for engineering plans and ecological studies. High-resolution topography data (elevation data with a vertical precision of about 15 cm and horizontal resolution of about 1 m, also known as ‘lidar’ for Light Detection and Ranging) is revolutionizing the way we make flood inundation predictions. Lidar data contains very detailed information about the ground surface, as well as vegetation on the floodplain, which exerts a strong influence on the velocity and depth of the water. Many states are revising their flood risk maps using this new high-resolution data.

Societal and Economic Implications of Floods

Floods are consistently ranked among the most costly natural disasters around the world, with many billions of dollars in damages reported annually. For example, the Centre for Research on the Epidemiology of Disasters International Disaster Database [12] (EM-DAT) reports that floods accounted for four of the ten most deadly natural disasters in 2013, with confirmed global fatalities exceeding 9500 people (EM-DAT, (2013)). The same 2013 report documents $54 billion in damages directly related to river floods.

During late spring and summer of 1993 heavy rains all throughout the Midwestern US resulted in flooding along the upper Mississippi and Missouri river systems. The floods were the most costly in US history, causing about $15 billion in damages and forcing about 75,000 people from their homes. Heavy rains from Tropical Storm Irene in August 2011 caused approximately $10 billion in damages throughout the Caribbean and eastern United States (including flooding Kingston, NY during the wet period indicated in Figure 1). Notably, the numbers of fatalities associated with these extreme events were relatively low (~50 deaths each) due to remarkably accurate flood forecasting, highly effective emergency response systems and regulations that limit development in flood-prone areas.

Figure 6. Flooding along the Mississippi River near St. Louis, Missouri on July 9, 1993.

Source: Photo by Andrea Booher, Federal Emergency Management Agency.

Rivers that cannot transport their sediment load (sand and gravel) are particularly susceptible to flooding because sediment settles out in the river bed, causing the river channel to become shallower relative to its banks, thus increasing the chances of flooding. The Yellow River in China is one example of such a river. While the Yellow River has played a pivotal role in the Chinese economy for thousands of years, sedimentation has repeatedly caused the river channel bed and banks to actually build up higher than the surrounding floodplain. This is an especially dangerous situation that can cause the river to catastrophically flood, breach its banks, abandon its channel altogether, and ultimately form a new channel elsewhere within the floodplain, a process known as an avulsion. The Yellow River has caused many devastating floods, including a flood in 1332-1333 that killed an estimated 7 million people. Another Yellow River flood in September of 1887 inundated an estimated 130,000 km² (50,000 square miles, an area approximately the size of Alabama!) and killed an estimated million people. Yet another flood in 1931 is estimated to have killed 1-4 million. Such catastrophic disasters have earned the Yellow River its nickname, ‘China’s Sorrow.’

Humans have made extraordinary efforts to reduce flood damages. In some cases, these efforts involve limiting development in flood-prone areas. In other cases, these efforts involve building structures meant to control the floodwaters. Flood control structures include dams and retention basins that store water and/or building levees, dikes, and floodwalls that attempt to keep floodwaters confined. Some of the most extensive flood control systems in the world include the floodway diversions on the Red River, which runs between Minnesota and the Dakotas and crosses the US-Canada border into Manitoba.

While we typically think of floods as dangerous and costly natural hazards, they can also provide benefits to society. For example, floods naturally deliver fresh, nutrient-rich sediments to their floodplains, which have historically benefited farmers in many places throughout the world. Yearly floods of the Nile River allowed the early Egyptian people to grow crops, which helped them thrive as a civilization for thousands of years. However, the severity of the floods was unpredictable and floods that were too large caused significant damage. Therefore, in the mid-1900s the Egyptians constructed a flood-control dam on the Nile River. The dam eliminated both the risks and benefits of annual flooding and therefore agricultural practices have had to adapt by using irrigation and petroleum-based fertilizers to replace the water and nutrients that are no longer delivered to the floodplain by the river.

Flood control is not always feasible, given the unpredictable nature of these events as well as geographic or economic constraints. Nor is flood control necessarily desirable in many situations, given the potential environmental benefits for the river and floodplain discussed at the beginning of this section and discussed in greater detail towards the end of this section. In such cases, efforts can be made to reduce economic losses from floods. For example, in many places regulations limit the construction of permanent buildings on floodplains. Emergency response programs, such as the National Weather Service and Federal Emergency Management Agency (FEMA) help flood victims by improving methods to warn and evacuate people from flood-prone areas and to provide relief aid. In an alternative approach, communities in Tonle Sap, Cambodia have constructed their houses on floats and stilts to deal with the annual flooding of 8-9 m (26-30 ft) from the Mekong River.

Figure 7. Houses on Tonle Sap lake in Cambodia built on stilts to accommodate annual flooding of the lake by the Mekong River.

Source: Colocho, shared under Creative Commons Attribution-Share Alike 2.5 Generic license.

In many places, flood insurance can be purchased to help cover costs associated with residential and commercial flood damages. However, the private insurance industry is somewhat limited because the number of potential claimants far exceeds the number of people who wish to ensure their property against flooding. As a result, the US Congress created the National Flood Insurance Program in 1968. The program is an effort to provide flood insurance to protect homeowners, renters, and business owners as well as an effort to encourage communities to adopt flood risk management policies established by FEMA.

Measuring the Severity of Drought

Many different indices have been developed over the past several decades to indicate the occurrence and severity of drought. The simplest index relates precipitation amounts during a specific period of time to the historical average during that same time period. For example, precipitation for the month of June 2014 was 15% below the historical average for Wenatchee, Washington. While this statement conveys some useful information, it is not possible to determine whether or not that 15% deficit qualifies for any of the definitions of drought. The number of days with no precipitation is another simple index, but again must be considered in the context of historical data or water demand, and there is no standard definition for what number of days without precipitation would necessarily qualify under any of the four types of drought. Also, if an area receives a very small amount of precipitation (< 0.1 cm) during an otherwise unusually dry time period, a strict interpretation of this index would ‘reset the clock’, but in reality, the severity of the water deficit remains essentially unchanged. Complex phenomena, such as drought, require somewhat complex metrics to be measured in a meaningful way.

The Standardized Precipitation Index (SPI) is a slightly more complex measure of precipitation deficit that compares measured precipitation to the median historical precipitation over multiple timescales, ranging from one month to 24 months. As dry or wet conditions become more severe, SPI becomes more negative or positive, respectively. Several different indices of varying complexity have been developed to assess drought based on both water supply and demand using multiple environmental criteria. The most common index used to define and monitor drought is the Palmer Drought Severity Index (PDSI), which attempts to measure the duration and intensity of long-term, spatially extensive drought, based on precipitation, temperature, and available water content data. PDSI ranges from values exceeding 4.0, which are considered extremely wet, to values below -4.0, which are considered extreme drought (see Figure 12). Weekly maps of PDSI for the entire US (current and historical) can be viewed on the web page maintained by the National Weather Service Climate Prediction Center [15].

Figure 12. Palmer Drought Severity Index map of the conterminous US, segmented by climate divisions. This example is drawn from the week ending August 30, 2014.

Source: Climate Prediction Center, NOAA [16]

Related indices are the Palmer Z Index, which attempts to measure short-term drought on a monthly timescale, the Palmer Crop Moisture Index, which attempts to measure short-term drought and quantify impacts on agricultural productivity, the Palmer Hydrological Drought Index, which attempts to estimate the long-term effects of drought on reservoir levels and groundwater levels. An immense compilation of current and historical drought information for the entire US is freely available on the US Drought Monitor web page [14], maintained by the University of Nebraska National Drought Mitigation Center.

Increasingly, government and industry groups are using ‘cloud seeding’ techniques to induce precipitation and reduce the severity of a drought. One of the potentially limiting steps in the formation of precipitation is the presence of tiny particles (nuclei) on which water can condense and coalesce to form raindrops or ice crystals large enough to begin falling through the air. Cloud seeding is the practice of injecting nucleating agents, such as silver iodide (AgI), into clouds in an attempt to form precipitation. The effectiveness of these approaches is questionable, but under the right conditions, cloud seeding may increase the probability of rain and therefore it is practiced in some semi-arid regions, including the western US. However, questions remain regarding environmental and human health impacts as well as concerns regarding ‘stealing’ atmospheric moisture from would-be recipients downwind.

Floods and Droughts Impact Ecosystems

How do floods and droughts impact ecosystems?

Variation in river flow (i.e., the river flow regime – see Module 3) exerts a strong influence on river and riparian ecosystem function. In particular, floods and droughts control the creation and maintenance of river and floodplain habitats and the sustainability of the high biodiversity observed along river systems. The temporal pattern of floods interacts with channel and floodplain topography to create a highly heterogeneous landscape of depressions, oxbows, gravel bars, and terraces (Figure 13). The hydro-geomorphic diversity means that the inundation frequency varies strongly over short distances on river floodplains, and creates habitats for a diverse suite of organisms adapted to a wide range of flooding frequencies.

Figure 14. The Tagliamento River and floodplain in northeastern Italy. Note the diverse array of hydrogeomorphic surfaces, including vegetated islands, gravel bars, off-channel pools and non-flowing channels. The different surfaces will become inundated at different magnitude floods and thus will be flooded at different frequencies over time. The diverse array of flooding frequencies provides habitat for a diverse group of aquatic and riparian organisms.

Source: Brian Laub, aerial imagery from Esri

Both riparian and aquatic organisms have adapted to take advantage of flood-drought cycles in river ecosystems. For example, many fish species time spawning runs to coincide with predictable floods, because this allows large adult fish to access small streams that provide optimal habitat for egg development and growth and survival of young fish. In the Amazon River, many fish species can almost be considered forest-dwelling fish, because they feed directly on leaves, fruits, seeds, and insects that fall into the river when it floods surrounding forests during the annual rainy season. Trees of these seasonally flooded forests have in turn developed fruits and seeds that mature during the flooding season and that can survive fish digestive systems in order to take advantage of the seed dispersal ability of mobile fish species. In the western U.S., cottonwood trees time the release of seeds to coincide with the recession of flood peaks in order to access fresh sediment deposits with elevated water tables that provide ideal habitats for germination.

It may be less obvious that droughts could be beneficial for aquatic and riparian biota, but when coupled with periodic flooding, droughts play an important role in the survival of many river organisms. During droughts, resources such as organic material and nutrients can accumulate on floodplain surfaces, and when a flood does occur there is a pulse of greater resource availability than would occur under regular flooding, and this period of high resource availability can ensure the quick growth and survival of organisms, including young fish. In addition, periodic drying of rivers and floodplain wetlands eliminates competitors and predators for organisms that can quickly colonize areas when water returns. Such areas of refuge from predators are critical for the persistence of many aquatic organisms and would not exist without periods of drought.

The importance of floods and droughts to the integrity of river-floodplain ecosystems is apparent when alterations to the natural flow regime occur. Riverine organisms are often closely adapted to the local magnitude, frequency, duration, and predictability of extreme events, such that alteration of any one component can threaten species persistence. For example, recruitment of cottonwood trees along many dammed rivers in the western U.S. has essentially ceased, because the dams prevent flooding and creation of germination sites during the spring when cottonwood trees release their seeds. Excessive drought is also highly detrimental to river systems. One of the most famous examples of drought impacts is seen in the Colorado River delta in Mexico, which was once a highly productive floodplain forest and swamp, but due to prolonged drought conditions in the river basin and water infrastructure development, is now a dry desert.

Non-Stationary Hydrology

When ‘normal’ changes: non-stationary hydrology

As we have discussed at length above, floods and droughts can have significant impacts on society and the environment. However, as also discussed above, we can characterize the frequency, duration and magnitude of these rare or extreme events and both human and natural systems have mechanisms to deal with them. For example, we can estimate the magnitude of a 100-year flood and build a bridge sufficiently large to pass the flood with minimal risk of damage to the bridge or changes in water depth (i.e., from water backing up behind the bridge and therefore flooding additional land upstream). Similarly, we can compute the width of a floodplain that would be inundated from that 100-year flood and choose to not build within that flood corridor, or choose to require those who do build within the flood corridor to purchase flood insurance to cover costs of potential damages.

Similarly, as we discussed in module 3, river channels naturally adapt their width and depth to accommodate common floods. Thus, society and the environment naturally develop some amount of resiliency to historical climate extremes. However, all of the prediction methods we have discussed so far in this module rely on the assumption that the future will look statistically similar to the past (i.e., the distribution of events will not change and occurrence of events in the future will be consistent with the probabilities computed from the histograms or probability density functions such as those shown in Figures 3 and 4). This assumption is known as the Stationarity Assumption in hydrology. Specifically, stationarity implies that while there is considerable variability in precipitation and streamflow, that variability is bouncing around a relatively constant average value and has a relatively constant spread, as shown in the hypothetical plot of peak streamflow over time in the left panel of Figure 17. From left to right, the mean (µ) and standard deviation (s, i.e., the spread of the data around the mean) don't change. However, as climate changes, the magnitudes, durations, and frequencies of floods and droughts may occur that are outside the historical range of observations, resulting in a change in the average magnitude of floods (Figure 17, middle panel) or a change in the variability of floods (Figure 17, right panel). In either of these situations, the statistics from the left side of the graph don't provide a good basis for making predictions about the right side of the graph (or predicting what will happen in the future!). Understanding and accounting for such non-stationary patterns in precipitation and streamflow are among the greatest challenges in hydrology today because we need to make accurate future predictions for many decisions about flood and drought risk, infrastructure design (roads, bridges, culverts, ditches, parking lots, detention basins, sewers, etc.!) and water availability. So this is a hot topic in the field and many new techniques are emerging!

Non-stationarity will be common in the future as regional climates systematically change. According to the Intergovernmental Panel on Climate Change, climatic warming will increase the risk of both floods and droughts (Table SPM2 in IPCC, 2007; see also IPCC, 2014). The multitude of factors that combine to ultimately cause floods and droughts are exceptionally difficult to predict over the next few decades. Nevertheless, there is a high level of agreement among the competing IPCC climate simulation models regarding the general trends of several metrics. For example, precipitation intensity increases are expected in most places and especially at mid- and high-latitudes where mean precipitation also increases. Summer droughts are also expected to increase over low and mid-latitude continental interiors. Snowpack is expected to decline overall as more precipitation will fall as rain rather than snow, especially in areas with temperatures near 0°C in fall and spring. Relatedly, snowmelt is projected to occur earlier. The combination of less snowpack and earlier snowmelt increases risk of summer to fall drought in snowmelt dependent regions, such as the western US. Some regions outside the US are highly dependent on meltwater from glaciers for water supply. Accelerated melting due to climatic warming increases the risk of flooding downstream in the near-term. In the long-term, glaciers in many of these areas will shrink and ultimately cease to exist, posing a serious threat to the downstream water supply. This poses a serious risk to the hundreds of millions of people in China and India who depend on glacial meltwater from the Hindu Kush-Himalayas. In closing, there is a good reason to expect floods and droughts to become more severe in the coming decades, increasing the urgency for improved predictions, mitigation efforts, and adaptation strategies.

Nutrient Supply to Floodplains

Flooding is a natural process that replenishes soil and nutrients to floodplains. Of course, floodplains are ideal sites for agriculture – they are flat, water is accessible, and – at least prior to modification of the system by levees or dams – the soils are among the most fertile on Earth due to recurring flooding that deposits nutrient-rich fine-grained sediments. Historically, these are the sites of major agricultural and population centers, including the “Fertile Crescent” along the Tigris-Euphrates floodplain (now largely barren due to long-term effects of irrigation-based agriculture and flood prevention), and the Nile Floodplain (see The Nile River and Aswan Dam below). Likewise, most major modern agricultural production is localized to floodplains - including the Central Valley of California, the Susquehanna River Valley, the upper Tigris-Euphrates basin, the Nile Valley, and the floodplains of the Mississippi and Missouri Rivers.

Prevention of flooding through the combination of dams (which control river discharge) and levees (which artificially channelize flow and shunt it downstream so that it cannot spill onto the floodplain) is a strategy to limit the loss of crops and property, and allow development in otherwise flood-prone areas. This approach, while generally effective in limiting short-term losses, affects soil fertility, groundwater systems, and the health of downstream waterways in the longer-term. For example, flood prevention eliminates a major source of recharge to aquifers in valley-fill sediments that lie below the floodplain. Recurring floods also serve to flush salts that accumulate naturally in soils due to evaporation and transpiration (i.e. water is transported to the atmosphere by these processes, but even small amounts of dissolved salts remain in the soil). Reduction or elimination of this flushing can lead to soil salinization, with negative effects on soil fertility.

Perhaps most notably, by eliminating or limiting the replenishment of nutrients to the floodplain, imported fertilizer is required to grow crops. Excess fertilizer application, in turn, leads to runoff enriched in Nitrogen and Phosphates that affects aquatic species and can cause eutrophication of lakes and estuaries downstream. This is a longstanding problem that leads to algal blooms at river mouths, consumption of Oxygen by organic matter (dead algae), and ultimately to “dead zones” in these regions that affect fisheries (Figure 2).

Figure 2: Dissolved oxygen levels in the Gulf of Mexico in Summer, 2011. Note clearly demarcated “dead zone” extending along the Texas-Louisiana coastline.

Source: NOAA (Public Domain) [19]

The Nile River and Aswan Dam

Large rivers are difficult to control. The Nile River, so important to Egypt's populace, is no exception. But since the late 1960s, the Nile River has been under the control of humans because of the construction of the Aswan High Dam. Part of the rationale for this dam was to manage the natural cycles of flood and drought to produce dependable water supplies for farming and other uses. The consequences of cyclic climate variations on a decadal scale were buffered by the large storage capacity of the Nile Valley behind the High Dam, which is nearly six trillion cubic feet (157 km3) of water! This is about four times the amount of water stored behind Hoover Dam (USA, Lake Mead) and Three Gorges Dam (China) (Chao et al., 2008). In addition, the Aswan High Dam initially produced a significant amount of electrical power (about 50%, now less than 15% of Egypt's needs) that allowed electrification of "rural" Egypt.

Figure 3 shows the narrow Nile River Valley slicing northward through the Egyptian desert. The narrow green band of the Nile River Valley represents farmland irrigated by waters of the Nile River. Prior to the completion of Aswan High Dam, the Nile River would flood its valley annually during the rainy season in its higher altitude headwaters (Ethiopia, Sudan, Kenya, Uganda), bringing nutrient-rich silt to fields and renewing fertility. In addition, a substantial volume of sediment was carried down the Nile River Channel to its large delta, building out the delta into the Mediterranean Sea, providing additional fertile land for farming. This no longer happens because the Aswan High Dam effectively (an unintended consequence) traps sediment carried from the highlands behind it. Now, the delta region, which subsides naturally as the result of compaction of sediment (newly deposited sediments have water contents of 70% or more that are reduced by compaction by overburden), is diminishing in size because rates of coastal erosion exceed supply of sediment. Currently, almost 1/3 of the Nile Delta’s land area sits within a meter of sea level. Subsidence rates vary across the delta, but in some areas, the land surface is sinking as fast as 1 cm/yr. Control of the Nile’s flow has also lead to water quality problems. Because once-regular floods no longer flush salts, sewage, fertilizers, and waste from the delta, surface waters are polluted and those living near the Mediterranean coast increasingly rely on groundwater to meet demand for drinking water and domestic use. Extraction of groundwater, coupled with land subsidence, has led to saltwater intrusion in the aquifer as far as 30 km inland.

There have been other unintended consequences of the Aswan High Dam including the spread of disease (Schistosomiasis), a decrease in water quality and increase in algal blooms resulting from fertilization of farm fields and irrigation runoff, flooding of historical sites, and displacement of people from the regions flooded by the reservoir.

Figure 3. Egypt and the northern Red Sea (Google Earth image) and the Nile River Valley coursing through the desert region. Much of the green swath in the image is the fertile Nile valley with agriculture. The Aswan Dam and its reservoir (darker blue), however, can be seen in the lower part of the image.

Source: Google Earth

Figure 4. Satellite image showing the northern Nile River and Nile Delta on the southeastern margin of the Mediterranean Sea, as well as the Gulf of Suez, Gulf of Elat and northernmost Red Sea in the Egyptian Desert.

Source: NASA / MODIS

Sediment Trapping

Because large reservoirs behind major dams are areas where water flow velocity is slowed (also often called “slackwater”), sediments are deposited where rivers enter the water body (Figure 5). Sedimentation in reservoirs behind dams has several consequences. Sediment deposition reduces reservoir water storage capacity and therefore limits the useful lifetime of the dam for flood control, water supply, and hydropower generation. Recent detailed studies of storage capacity and sedimentation rates for reservoirs in the U.S. suggest that average annual storage losses range from less than 0.5% to more than 2% (see supplemental reading: Graf et al., 2010; “Sedimentation and sustainability of western American Reservoirs, Water Resources Research”). The highest rates of storage loss are occurring in the American West, and the lowest in the Northeast.

For example, almost 20 million tons of sediment are deposited annually in reservoirs along the Mississippi River (UNESCO, 2011). China’s Three Gorges Dam alone (one of several along the Yangtze River) traps 34 million tons of sediment per year, or 31% of the river’s sediment load (Hu et al., 2009). Globally, the amount of sediment trapped behind dams is estimated to be 73 km3 (Syvitsky & Kettner, 2011). This sediment accumulation slowly reduces reservoir capacity behind dams and is one factor that limits their useful life expectancy. Recent studies of sediment accumulation suggest that the life expectancy of Lake Powell is ~300-700 yr, and that of the Three Gorges Reservoir in China is ~150 yr.

Figure 5. Aerial photo (taken in 2004) of delta and sediments deposited into Upper Las Vegas Bay by Las Vegas Wash where it flows into Lake Mead. The deposition of sediments has already filled in much of the bay, and as a result, the boat ramp and marina (part of the Lake Mead National Recreation Area) have been abandoned.

Image from US Geological Survey [20]

Figure 6. Sediment discharge in the Mississippi River and major tributaries before (left) and after (right) large-scale damming of the river system.

US Geological Survey Circular 1133 (1995)

The concomitant reduction in sediment delivery to downstream areas also has several consequences (Figures 5-6). Ultimately the decreased sediment supply to the river mouth translates to net erosion of beaches and loss of land in coastal regions, as natural coastal erosion by currents and subsidence caused by compaction of delta sediments is not offset by delivery of sediment. For example, prior to construction of the Aswan High Dam began in 1960, the annual sediment flux to the Nile Delta was ~100 million tons. This sediment supply was enough to offset erosion and natural subsidence.

Other Impacts

In this section, we will consider other impacts of dams.

Three Gorges: A “Mega-Dam” and Its Impacts

The Yangtze River is the longest river in Asia and is the world’s 3rd longest (only the Nile and Amazon are longer). It flows for ~6300 km from its headwaters on the Tibetan Plateau to its delta at Shanghai, where it discharges to the East China Sea. The Yangtze watershed encompasses approximately 1/5 of China’s land area. The river serves the water demand of millions of people and the delta alone supports almost 20% of China’s GDP. However, the Yangtze is also notorious for its frequent and devastating floods. Floods in the twentieth century alone led to the loss of an estimated 300,000 lives, including 145,000 drowning deaths in a 1931 flood, and 30,000 deaths in 1954 from flooding and diseases that followed. In addition to loss of life, these floods inundated hundreds of thousands of acres of productive farmland and caused billions of dollars of damage.

To protect over 15 million people in Shanghai and the lower Yangtze floodplains, and control flooding of almost 15,000 square km of land, construction of the Three Gorges Dam began in 1994. The dam, constructed at a cost of between ~$28-60 billion dollars (exact cost is not known because the project has been funded by a combination of government subsidy and private investment), is nearly 200 m high, spans more than 2 km across the river, and was engineered to withstand a magnitude 6.0 earthquake (Figure 11). The long, narrow Three Gorges Reservoir extends ~600 km upstream and has a capacity of almost 40 billion m3 of water (equivalent to about 32 million acre-feet) (Figure 12). At the time of construction, the dam was the largest hydroelectric power plant on Earth, with a generating capacity of over 20,000 megawatts - more than 20 times that of Hoover Dam, equivalent to 18 nuclear power plants, and enough to supply almost 10% of China’s power demand. According to the Chinese government, if this amount of electricity were generated using coal-fired power plants instead, 100 million tons of additional carbon dioxide would be released into the atmosphere. The dam also increases the navigability of the Yangtze, allowing large freighters to transport goods far into China’s interior.

Despite the obvious benefits of the dam for the economy and generation of renewable energy, the Three Gorges Dam has been mired in controversy since its inception. Concerns about the dam include an array of environmental impacts, the forced relocation of over a million residents, initiation of large landslides and earthquakes by the rising reservoir, and flooding of important historic and cultural sites in the gorge upstream of the dam. The chief environmental issues center on impacts to river ecology and already threatened species increased chances for waterborne diseases, and water quality degradation associated with the slowed flow of the river in the backwaters of the dam, in tributaries, and in downstream regions. Indeed, in the wake of pollution concerns, during construction, an additional $4.8 billion was budgeted for new treatment plants and garbage disposal sites along the river’s upstream reaches.

Figure 11. The Three Gorges Dam in the final phases of its construction.

Source: "ThreeGorgesDam-China2009" by Le Grand Portage Derivative work: Rehman - Licensed under Creative Commons Attribution 2.0 via Wikimedia Commons

Figure 12. Image of the Three Gorges Reservoir, taken by astronauts on the International Space Station on April 15, 2009, when the reservoir was filled to near its capacity (maximum water level was reached for the first time in October of 2010)

Source: NASA

Enhancement of Populations of Anadromous Fishes

Dam removal has become increasingly popular, particularly as regards enhancement of populations of anadromous fishes--salmon and steelhead in the western U.S. and shad, herring and other species in the eastern U.S. Figure 13 illustrates the dams that have been removed in the lower 48 states (Figure 14 expands the eastern U.S.) from 1936 to 2013 on the basis of data tabulated by American Rivers [26]. These dams have been removed largely to improve the ecological conditions of river systems and to allow migratory fish to pass unimpeded for spawning in the upper reaches of rivers. Dam removal is commonly cited as a way of increasing stocks of imperiled fishes (Chesapeake Quarterly: Those Dammed Old Rivers [27]), allowing them to spawn in rivers that have been inaccessible to fish migration for as much as a century. Others argue that dam removal from coastal rivers is only part of the solution for fisheries improvements. Although numbers of anadromous fish may increase, the size of individual fish may not as long as fishing pressure remains the same (Oregon State University: Department of Fisheries and Wildlife [28]). This has been argued in the case of Elwha Dam removal in the Olympic National Park watershed (Crosscut: Elwha dams: Will bringing down NW dams really help salmon? [29]). In some cases, the benefits will accrue to native Americans who formerly depended on fishing for their nutrition and livelihood.

Figure 13. Map of the lower 48 states showing locations of dams removed from 1936 to 2013. At American Rivers [30], the locations are clickable to show information on each dam removed.

Source: A [31]merican Rivers [30]. Used with permission. All rights reserved.

Figure 14. Blow up of eastern U.S. portion of the dam removal map.

Source: American Rivers [30]. Used with permission. All rights reserved.

Esthetic Improvements

Arguments have also presented for esthetic improvements as the result of dam removal.  An example of this approach is the argument for the removal of Hetch Hetchy (O'Shaughnessy) Dam on the Tuolomne River in the Sierra Nevada Mts. of California (Restore Hetch Hetchy [32]).  This dam and its reservoir flooded a canyon much like Yosemite just to its south beginning in 1923, even though Hetch Hetchy was included in Yosemite National Park in 1890 by President W.H. Harrison.  President Wilson, in 1913, signed the Raker Act that allowed San Francisco to dam the valley. Proponents of this dam removal argue for ecological improvements as well as access to once-spectacular scenery eliminated as the result of the water project. Hetch Hetchy provides some 20 percent of hydroelectric power generation to San Francisco as well as significant water supplies.  Nonetheless, proponents of the dam removal argue that impacts on water or electrical power availability would be minor (Hetch Hetchy Today [33]).

Issues Accompanying Dam Removal

One of the issues accompanying dam removal is the potential impact of the large volume of sediment that accumulated behind some dams over time. This so-called "legacy sediment" is commonly very fine-grained and contains stored nutrients and organic matter, among other possible pollutants. When dams are destroyed, efforts must be made to avoid a large flux of this sediment downstream as the newly released river cuts down to its natural base level. This requires careful engineering and significant funding. Merritts and Walter (2010) have suggested that most rivers and streams in eastern Pennsylvania and Maryland run through floodplains and levees of legacy sediment, not natural river valleys, created by the plethora of small dams built to impound water for hydropower in the past. And, even now, some large dams serve as a buffer against sediment transport that would create broad mudflats and high turbidity in coastal bays such as the Chesapeake Bay. Conowingo Dam in Maryland is one such structure that is estimated to trap as much as 50% of the sediment carried by the Susquehanna River before it empties into the Chesapeake Bay. There are concerns that this dam only has a bit more than a decade of useful life before it is filled to the brim with sediment, thus imperiling Chesapeake ecosystems (Langland, USGS, 2012). For an example of the engineering and costs of removing a dam, check out the San Clemente Dam Removal & Carmel River Reroute Project [34] in central California.

Figure 13. Map of the lower 48 states showing locations of dams removed from 1936 to 2013. At American Rivers [35], the locations are clickable to show information on each dam removed.

Source: AmericanRivers.org Used with permission. All rights reserved.

Figure 14. Blow up of eastern U.S. portion of the dam removal map.

Source: AmericanRivers.org Used with permission. All rights reserved.

Many dam removal projects are proposed, but await funding from federal sources—Congress must appropriate funds. Projects on the Snake River and the Klamath River in the west remain controversial, but at least in the case of the Klamath, the Department of the Interior has recommended removal to allow salmon to reinhabit this river. Figure 9 illustrates the consequences of waiting for funding of dam removal using the Olmstead Dam example.

Figure 15. The costs of waiting to remove and replace dams once approved using the Olmstead Dam and Locks system on the Ohio River. Lambrecht, Bill (16 September 2013). "Costly Dam, Illinois dam costing billions more than expected as delays mount". St. Louis Post-Dispatch. Retrieved 9 March 2014.

Click Here for Text Alternative of Olmsted dam spending graph

In 1988, Congress authorized spending 775 million dollars for a 7-year project to build Olmsted Locks and Dam. But the cost has more than quadrupled to 3.1 billion dollars, and 25 years later, the project is barely half done.

 

Year	Funds in millions
1991	5
1993	60
1995	35
1997	70
1999	55
2001	60
2003	70
2005	75
2007	115
2009	115 + 5 in stimulus funds
2011	140
2013	145
2014	163

Source: Army Corp of Engineers

Human Intervention in the Global Water Cycle

On the other hand, as discussed above in Ponding the Waters: Impacts of Dams, the effects of large-scale alteration of river systems and the hydrologic cycle have become increasingly clear in the past few decades. The scale of human intervention in the global water cycle is also becoming apparent, including restriction of river flows such that they no longer reach the ocean in many years, associated two- or three-fold increases in the residence time of runoff, decreased sediment delivery to the oceans, and a long-term measurable effect on global sea level caused by the impoundment of thousands of cubic km of water (Vörösmarty et al., 2004). As a result, new large dam projects have been heavily scrutinized and faced political and environmental opposition. At the same time, the efficiency and economics of energy production, and the net offset of greenhouse gas emissions from increased hydropower generation have been increasingly questioned (World Commission on Dams, 2000). One way to minimize environmental impacts is to design “run of the river” systems, in which no reservoir is created and instead the natural flow of the river in its channel is harnessed to generate power. However, these systems have several drawbacks: they rely on natural flows, so the power generating capacity fluctuates dramatically as a function of seasonal rainfall patterns and climate change; and there are no added benefits of flood control or water supply.

Dams and Economic Development

Despite the controversy, in many developing nations, major dam projects remain important engines for economic development and hold substantial potential for renewable energy generation. As of 2012, hydroelectric power constituted as much as 16.5% of global electricity production (and 75% of estimated renewable energy generation) (REN21, 2013). Of this, ~23% is in China, ~12% in Brazil, ~10% in Canada, and ~7.5% in the United States; combined, these four nations generate over half of the world's hydropower!

Moreover, globally, estimates suggest that up to two-thirds of economically viable dam sites have yet to be exploited. Undeveloped sites are especially abundant in Latin and South America (79% of renewable water remains unused), Africa (96%), India and China (48% is unused in Asia) (UNEP, 2013). Rapidly growing energy demand in India, China, and the Amazon Basin have driven the construction of hundreds of large dams as of 2002 (Figures 10-11; Table 1). This development may be a harbinger of things to come on the African continent. Africa has the second-highest population (after Asia), and the fastest-growing (See Module 1.3); it also has the lowest per capita energy use (UNEP, 2013). Looking to the future as demand for energy, water, and food in developing nations continue to grow – both per capita and in total as populations swell - it seems inevitable that demand for large dams will persist well into the 21st century.

Figure 16. Number of large dams by geographic region, as of 1998 (World Commission on Dams, 2000).

Click Here for Text alternative of dams by geographic region graph

Region	Number of dams
China	22000
Asia	9000
North and Central America	8000
Western Europe	4000
Africa	1000
Eastern Europe	1000
South America	800
Austral-Asia	500

Source: World Commission on Dams

Figure 17. Map showing existing and planned hydropower projects in the headwaters of the Amazon

Source: Finer M, Jenkins CN (2012) Proliferation of Hydroelectric Dams in the Andean Amazon and Implications for Andes-Amazon Connectivity. PLoS ONE 7(4): e35126. doi:10.1371/journal.pone.0035126

Goals and Objectives

Goals

• Explain the distribution and dynamics of water at the surface and in the subsurface of the Earth
• Interpret graphical representations of scientific data

Learning Objectives

In completing this lesson, you will:

• Identify the properties of artesian wells and describe the conditions under which they form
• Explain the difference between porosity and permeability
• List and describe the properties of aquifers that control the movement and storage of groundwater
• Explain the role of fractures in determining the transmission properties of aquifers
• Use Darcy's Law to explain the roles of aquifer properties and driving forces in governing the rate of groundwater flow

Aquifers Explained

Aquifers come in many shapes, sizes, and “flavors”. For example, some aquifer systems span hundreds – or even thousands – of kilometers across several states or nations and may include multiple individual layers of rock or sediment that total thousands of meters in thickness. Other aquifers may be restricted to a few kilometers within a stream valley, and be only a few meters thick.

Nonetheless, there are some important common features of different kinds of aquifers. In this section of the module, we provide a basic overview of aquifers: What is an aquifer? What are the different types of aquifers, and what is their anatomy?

Types of Aquifers

In more detail, there are three main classifications of aquifers, defined by their geometry and relationship to topography and the subsurface geology (Figures 6-9). The simple aquifer shown in Figure 6 is termed an unconfined aquifer because the aquifer formation extends essentially to the land surface. As a result, the aquifer is in pressure communication with the atmosphere. Unconfined aquifers are also known as water table aquifers because the water table marks the top of the groundwater system.

A second common type of aquifer is a confined aquifer, which is isolated from pressure communication with overlying or underlying geologic formations – and with the land surface and atmosphere – by one or more confining layers or confining units. Confined aquifers differ from unconfined aquifers in two fundamental and important ways. First, confined aquifers are typically under considerable pressure, which may be derived from recharge at a higher elevation or from the weight of the overlying rock and soil (known as the overburden). In some cases, the pressure is high enough that wells drilled into the aquifer are free-flowing. This condition requires that the water pressure in the aquifer is sufficient to drive water up the wellbore and above the land surface, and such wells are called artesian wells (Figure 7). Second, confined aquifers typically remain saturated over their entire thickness, even as water is removed by pumping wells. Water extracted from the aquifer comes only from the depressurization of the aquifer – a combination of depressurization and expansion of the water itself, and relaxation of the aquifer formation upon reduction in pressure (Figure 8).

Figure 6. Schematic cross-sectional diagram showing a layered system with an upper unconfined aquifer above a confining unit, and underlain by a confined aquifer. Note the water level in the two wells: In the unconfined aquifer, the water level in the well is the same as the height of the water table. In the confined aquifer, the water level is higher than the top of the aquifer – indicating that the aquifer is fully saturated and that the water is under pressure.

Source: USGS Water Science Photo Gallery [39]

Figure 7. Artesian well in Southern Georgia

Source: USGS Water Science Photo Gallery [39]

Figure 8. Schematic illustrating the effects of pumping from a well in an unconfined aquifer (left) and a confined aquifer (right). When unconfined aquifers are pumped, the water withdrawal leads to a drop in the water table, and the pore spaces become unsaturated. Pumping in confined aquifers decreases the water pressure, but the pore space remains fully saturated.

Source: W.M. Alley et al., 1999, USGS Circular 1186: Sustainability of Groundwater Resources [36]

The third main type of aquifer is a perched aquifer (Figure 6). Perched aquifers occur above discontinuous aquitards, which allow groundwater to “mound” above them. Thee aquifers are perched, in that they sit above the regional water table, and within the regional vadose zone (i.e. there is an unsaturated zone below the perched aquifer). The dimensions of perched aquifers are typically small (dictated by climate conditions and the size of aquitard layers), and the volume of water they contain is sensitive to climate conditions and therefore highly variable in time.

Figure 9. Schematic cross-section showing the occurrence of perched aquifers above an unconfined aquifer.

Source: D.T. Snyder, U.S. Geological Survey Scientific Investigations Report 2008–5059, Estimated Depth to Ground Water and Configuration of the Water Table in the Portland, Oregon Area [40]

Aquifer Properties

In the previous section, you’ve learned about the different types of aquifers, and the basic characteristics that define an aquifer – namely the ability to store and transmit water. But what, exactly, about a rock or sediment beneath the ground determines whether the rock can hold water, or whether water can percolate through it? In the following section, we will explore this question in more detail, to define the important individual properties of the rock.

Regional Aquifer Systems: Examples

Ground water flow systems extend over a wide range of scales, from small perched aquifers that may supply water for a single-family, to regional rock formations that span thousands of km and cross several states (Figure 18). These regional systems supply water for irrigation and domestic uses in many areas, especially in semi-arid and arid parts of the American West and coastal population centers along the East coast (remember Module 1, figures 10-12?). These regional systems commonly consist of several layered sedimentary formations and may extend to several kilometers in depth. The U.S. Geological Survey has compiled detailed studies of regional aquifer systems across the U.S., with useful information about climate, recharge, subsurface geology, use, and problems related to water quality or quantity (a list and links for each of the principal regional aquifers in the U.S can be found at USGS Groundwater Information [46]. A detailed atlas with information about the major aquifer systems in particular regions of the U.S. can be viewed at USGS Ground Water Atlas of the United States [47]. In this module, we will focus on a few example regional aquifer systems of particular relevance to the Northeastern and mid-Atlantic U.S. and the Central Valley of CA.

Figure 18. Principal aquifers of the conterminous United States.

Source: U.S. Geological Survey, Water Resources [48]

Darcy’s Experiments and Darcy’s Law

In 1855, Henri Darcy, a French hydraulic engineer (Figure 24), oversaw a series of experiments aimed to understand the rates of water flow through sand layers, and their relationship to pressure loss along the flow paths. Darcy’s experiments consisted of a vertical steel column, with a water inlet at one end and an outlet at the other. The water pressure was controlled at the inlet and outlet ends of the column using reservoirs with constant water levels (Figure 25) (denoted h₁ and h₂). The experiments included a series of tests with different packings of river sand, and a suite of tests using the same sand pack and column, but for which the inlet and outlet pressures were varied.  For part of our in-class activity this week, we will perform our own set of “Darcy Tube” experiments, and also work with the original dataset generated by Darcy in his experiments.

Figure 24. Portrait of Henri Darcy

Source: Glenn Brown [51], Biosystems and Agricultural Engineering, Oklahoma State University.

Darcy’s findings laid the foundation for the modern science of hydrogeology by quantifying the relationships between volumetric groundwater flow rate, driving forces, and aquifer properties. Specifically, Darcy’s experiments revealed proportionalities between the flux of water, Q, through the laboratory “aquifer” and different characteristics of the experimental system (refer to Figure 25 above):

(1) Q was directly proportional to the difference in water levels from inlet to outlet, h₁ - h₂ =  Δh:

               Q  ∝  Δh     

(2) Q was directly proportional to the cross sectional area of the tube:
 
               Q  ∝   A  

(3) Q was inversely proportional to the length of the column:

               Q ∝ 1/Δ L

Combining these proportionalities leads to Darcy’s Law, the empirical law that describes groundwater flow:

               Q = KA(Δh/ΔL)

where K is a constant of proportionality that defines the water flux for a given hydraulic gradient (Δh/ΔL). The above equation can also be recast in terms of the water volume flux per unit area, Q/A (also called "Darcy flux" or "Darcy velocity" with units of length per time):

               Q/A = K(Δh/ΔL)

Note that in Darcy’s Law, the other terms all describe the driving forces or geometry of the experimental system; none of them would change if the sand pack in the tube were changed to gravel, silt, or another material. This is where the constant of proportionality, K, comes in: it describes the ability of the material in the column to transmit water (sound familiar?). Indeed, the constant of proportionality and hydraulic conductivity (also K) are one and the same! For example, consider the same experimental geometry, but with silt or clay in the tube rather than sand. The flow rate would be lower, and this would be described in Darcy’s Law by a smaller value of K.

Goals and Objectives

Goals

• Describe the two-way relationship between water resources and human society
• Explain the distribution and dynamics of water at the surface and in the subsurface of the Earth
• Thoughtfully evaluate information and policy statements regarding the current and future predicted state of water resources
• Interpret graphical representations of scientific data

Learning Objectives

In completing this lesson, you will:

• Apply the concept of hydraulic head to draw flowlines on maps and cross-sections
• Interpret the current and historical balance between groundwater recharge and water extraction from well hydrographs
• Propose a course of action to address overdraft in an aquifer

Driving Forces for Groundwater Flow

The driving forces that control groundwater flow are a bit more complicated than those controlling flow in rivers and streams. As you learned in Module 3, surface water flows downhill due to gravity, and the flow direction is defined by the topography. Water flows downhill because gravity is a form of potential energy – and the water, or anything that falls or rolls downward – flows in response to differences in potential energy (from high to low).

In contrast to surface water, groundwater is separated from the atmosphere, and as a result, it can be under considerable pressure. Therefore, the potential energy that drives groundwater movement includes both pressure and gravity. In this section, you will learn about these driving forces, how we define them, and how they translate to the direction and rate of groundwater movement in the subsurface.

Effects of Pumping Wells 

Groundwater is accessed by either pumping from wells drilled into the aquifer (Figure 33), or by developing natural springs where the potentiometric surface intersects the land surface (Big Spring in Bellefonte, PA is one example of a relatively large spring that is used for municipal supply). Although springs are relatively inexpensive to develop, they are not always present, nor are the flow rates always sufficient to support demand. As a result, most groundwater extraction occurs by pumping wells, or in many cases “fields” of wells concentrated in a small area.

Figure 33. Example of groundwater extraction well in the Garden of Cambremer, France

Source: Philippe Alès (Own work) CC-BY-SA-3.0 [53], via Wikimedia Commons (top)

 

Figure 34. Example of an irrigation well in Walker Lake, NV

Source: USGS Nevada Water Science Center, photo by Lindsay R. Burt (bottom)

Cones of Depression: Pumping at a well, or at a wellfield, pulls water toward the well from all directions – in other words, it induces radial flow (around the radius of the well). In doing so, pumping causes a reduction in hydraulic head, known as drawdown. This drawdown generates a cone- or funnel-shaped depression called a cone of depression (Figure 35). The reduction in hydraulic head drives groundwater flow to the well (in the down-gradient direction), as shown in the example from Long Island in Figure 36.

Figure 35. Schematic cross-section showing the effects of pumping and generation of a cone of depression in an unconfined aquifer between a lake and a stream. Pumping and the ensuing cone of depression reverse the hydraulic gradient from pre-development (A) and induce flow from the stream into the groundwater system and to the well(B)

Click for more information

Two similar images. Image A is under natural conditions while image B shows the effect of pumping. Both diagrams show the same piece of land with a lake on the right, a stream on the left, a ground-water divide at a water table, with a surface-water divide right above it, and a confining unit beneath the groundwater flow. Diagram A shows an equal groundwater flow direction to the stream and the lake split at the ground-water divide. Diagram B has a high capacity pumping well close to the lake causing a cone of depression. Water flows from the ground-water divide to the stream as before but also to the well. Water is also pumped from the lake to the well instead of flowing out to the lake.

Source: USGS: Groundwater in the Great Lakes Basin: the case of southeastern Wisconsin

Figure 36. Potentiometric surface for southwestern Long Island, New York, ca. 1903 (pre-development), 1936 (near the peak in pumpage for combined municipal and industrial uses), and 1965, well after the municipal supply was switched to the upstate supply system described in Liquid Assets, and after reduction in withdrawals for industrial use.

Source: USGS [54]

Both the width and the depth of the cone of depression scale with the rate of pumping, the aquifer permeability, and storativity, and the duration of pumping. In general, larger cones of depression result from larger pumping rates, higher permeability or lower storativity, and longer elapsed time.  If cones of depression from two separate pumping wells grow large enough to overlap, it is known as well interference. When well interference occurs, the respective drawdowns are added together. The result is that drawdown is accelerated when multiple cones of depression interact. This is generally not desirable, and is one important consideration in the design, permitting, and operation of wells.

Not only does the cone of depression draw water to the well, but if the pumping rate is large enough or pumping is sustained for a long time, it can reverse the natural hydraulic gradient hundreds of meters to several tens of km away from the well(s). In some cases, this may result in interception of groundwater that would normally feed a stream or river as baseflow, and even in the interception of streamflow itself by inducing infiltration in the stream bed or banks (Figure 35B). In other cases, large cones of depression (up to a few miles wide!) associated with industrial or municipal well fields may reverse regional topographically-driven hydraulic gradients and lead to problems like saltwater intrusion (Figure 35B).

Groundwater Budgets

Now that you are an expert on how groundwater flows, we will apply that knowledge to the important problem of groundwater budgets. As groundwater flows through and exits an aquifer, for example at springs or at extraction wells, those losses of water may be balanced by recharge that percolates from the land surface. As you’ll investigate in the following section, and through a case study of the famous Ogallala aquifer in the American Midwest, understanding the budget of inflows and outflows to an aquifer is critical to evaluating the sustainability of groundwater use.

The High Plains Aquifer

The High Plains Aquifer system consists of Tertiary sedimentary rock, dominantly sandstone and gravel (Figure 45), eroded from the ancient Rocky Mountains and deposited in the Tertiary period (from about 31 to 5 million years ago). The Ogallala Formation is the primary aquifer unit in the system. The aquifer underlies almost 175,000 mi2 and spans eight states, with most of its area in Nebraska, Texas, and Kansas. This region is among the largest and most productive croplands in the U.S. and is the source of almost 20% of our corn, wheat, and cotton production, as well as a significant portion of our soybeans, sorghum, and alfalfa. It is also host to almost 20% of the cattle raised in the US. Because the climate is semi-arid, with mean annual precipitation ranging from 12 inches in the West to 33 inches in the East, growing economically viable crops requires substantial irrigation. If you have ever flown over the Midwest on a clear day you may have seen circular “patches” of irrigated land - the hallmark of center-pivot irrigation systems (Figure 46).

Although farming has been a major part of the economy in the region since the late 1800s, in the 1960’s new technology in electrical pumps allowed access to deeper groundwater and ushered in an era of rapidly expanding irrigated acreage (Figure 47). Accompanying this expansion, aquifer-wide groundwater withdrawals increased from a few million acre-feet (M-AF) per year to almost 20 M-AF. Water level declines began in the 1950s, with the onset of intensive groundwater extraction for irrigation-based agriculture. The total overdraft of the aquifer is almost 250 million acre-feet, from pre-development (ca. 1950) to 2011 – this is almost 17 times the annual flow of the Colorado River.

Figure 45. Outcrop of the Ogallala Formation in Texas.

Source: USGS High Plains Water Level Monitoring Study [58]

Figure 46. Satellite image of center pivot irrigation systems in Nebraska. Each circle is ~1/4 mile in diameter.

Image generated in Google Earth.

Figure 47. Map showing single-pivot irrigation systems in Nebraska in 1972 (upper map) and 1984 (lower map).

Source: USGS National Groundwater Atlas [59]

Figure 48. Map showing decreases in groundwater level from pre-development through 2011.

Source: USGS High Level Monitoring Study

As a consequence of sustained overdraft for several decades (see Figure 42 above), water levels in the aquifer have dropped substantially, by more than 100 feet in many areas (Figure 48). Over half of this decline has occurred since 2000. Water level declines are not evenly distributed, however. The highest rates of decline are focused in the southern reaches of the aquifer system, where recharge rates are low and irrigation demand is highest. In the northern portions of the aquifer system, water level declines are considerably smaller – and in some cases, the water table has actually risen – primarily due to locally higher recharge focused along the Platte River (Figure 48).