WESTERN GROUNDWATER CONGRESS: Quantifying surface water depletion from groundwater pumping

Understanding and conveying groundwater’s role in a changing world

Surface water commonly is hydraulically connected to ground water, but the interactions are difficult to observe and measure and have largely been ignored in water-management considerations and policies.  However, the Sustainable Groundwater Management Act (SGMA), passed in 2014, is California’s first statewide law that explicitly reflects the fact that surface water and groundwater are frequently interconnected and that groundwater management can impact groundwater-dependent ecosystems, surface water flows, and the beneficial uses of those flows.  The challenge of quantifying these interactions has led to the development of several techniques. 

At the 2019 Western Groundwater Congress, Gilbert Barth, PhD is an Associate and Senior Hydrologist with S.S. Papadopulos & Associates in their Boulder office, gave an overview of groundwater-surface water interactions and discussed some of the tools and techniques that he has helped develop. Dr. Barth provides quantitative assessments of groundwater resources to address questions associated with water planning, and specializes in model development and calibration with a focus on quantifying changes between surface water and groundwater systems.  He’s developed and applied models throughout the Western US for regional, interstate, and international deliberations.


It’s important to understand what motivates us, Dr. Barth began.  “Even if you are talking about the diffusion equation and quantifying things, you still need to start off by understanding how much change is too much, how much flow is too little, and how we will measure the things that are significant to us and you need to go through that process fairly systematically, otherwise you’ll probably answer the wrong question or you’ll just get the wrong answer.”

Establishing relationships is also critical, because options need to be explored in a way that’s not adversarial.  “This isn’t equations, this isn’t quantifying, and it may be at the front and the very tail end of your process, but I still think if you want to have stream depletion results that you are actually able to make use of and someone actually implements, these are the things that you need to keep in mind.”

It’s also important to try, whenever possible, to squash myths, and in the stream depletion context, there are several of them that Dr. Barth likes to target:

    • Aquifer volume does not equal supply if you’re talking stream depletion.
    • Groundwater is not a savings account; there is a cost taking groundwater out of storage, especially in a stream depletion context.
    • Monitoring triggers will not necessarily prevent impacts.

The world is changing as is our ability to perceive those changes, he said, noting that this isn’t a presentation on those changes, that’s just how it is.  Western supply is decreasing and demand is increasing, and understanding how groundwater is affected and when and for how long stream depletion will occur in the context of those changes is important.  And while those changes can prove challenging, they can also prove fruitful for setting up analyses to gain insight into the system.

Groundwater is the stream depletion driver,” Dr. Barth said.  “I say that from a groundwater-centric point of view, that’s what I was trained in, that’s what I believe in, but it is the foundation.  When you come to me in October and tell me your stream is dry, I’ll tell you why.  It’s because there’s something wrong with the groundwater; there’s nothing wrong with your stream.”

What do we need to know in order to quantify stream depletion?  “We can study the physics, we can look at the geometry of the system, we can characterize this system, we can even get into aspects of non-stationarity especially in today’s world, we can characterize our direct anthropogenic effects, and we can do a better job of characterizing pumping,” he said.  “So we have a lot of tools, and I like to think that if we start to build some of those pieces together, we can come up with a couple of steps that are critical for doing the quantification of stream depletion.”

Steps to quantifying stream depletion due to groundwater pumpingThe first step is to build the tools, which involves identifying and quantifying the concerns, characterizing the system, and developing the tools and models.  Dr. Barth noted that those tools and models are rarely completely accurate, so it’s very important to understand the limitations.

The second step is to then explore the surface water-groundwater interaction and specifically look at some of the differencing.  “Is it a change that you forecast in the future?  Is it a change that you saw in the past and you want to understand better?  Either way, the fact that you’re doing differencing can end up being a great benefit in terms of the possibility of eliminating some of the bias that could be in the system.”

Once you have the tools, you can explore with differencing and start to assess onsets and gaining insights into persistence, but the end goal is to effectively communicate the information.  “There are plenty of times that you can go out there and crunch some equations or develop some models and not really do anything with it, but if you’ve gone through these steps and you have done a good job of identifying the concerns and you can communicate the information in the end, then you’ll be able to warn people or advise people about what is the outcome that will be in their future if they choose a certain particular route or if conditions keep trending a certain way,” he said.


Water is a cycle,” he said.  “There are no magical barriers; surface water and groundwater is connected.  I don’t care what some law in some state tells you or used to tell you or what’s on the books.  There are all connected so let’s just work from that standpoint.  I want to dive into groundwater pumping specifically because with groundwater pumping, you can break that out into storage, capture, and stream and then into the stream depletion component, and these are important to consider because when we’re talking about these kinds of systems, it can play into how much we have in terms of the magnitude and the persistence of the effects.”

Pumping removes water from aquifer storage initially, and then very quickly starts removing it from the increased recharge or decreased discharge which together comprise the capture for the system, he said.

If we start with a system without any pumping and there is water moving to the stream, after a while of pumping, we will see differences in our hydraulic gradients and there will be changes in terms of the increased recharge (more water entering the system) or the decreased discharge (less water leaving the system), and those combinations of one through three will be what’s coming out of the well,” he said.  “This is important to sit and rest with for a moment.  It sounds fundamental.  But it does play a role, and it’s amazing to me how often people will not really take into account that this little drawdown cone has to be filled if you want to get back to your previous condition.”

flow between surface water and groundwaterThere are three basic functions that happen between the groundwater and the surface water system:

  • Gaining stream: Water is discharging from the aquifer to the stream because the water level in the adjacent aquifer is higher than in the stream.
  • Losing stream: The hydraulic gradient is reversed and water is flowing from the stream into the aquifer.
  • Disconnected stream: There is no longer water being exchanged between the aquifer and the stream. Barth noted that despite being called disconnected, there are still fluxes moving from the surface water to the groundwater system.

He likened a disconnected stream to a teenager.  During early childhood parenting, there is interaction between the child and the parent; the parental stream teaches the child aquifer and vice versa, it goes both directions.  But a disconnected stream is like the teenager; you can call it disconnected, but keep in mind that this is still the water cycle continuum, he said.

estimating capture and balancing the water budgetCapture can be estimated by looking at the water budget and quantifying the different components, such as springs, stream/drain/ditch seepage, groundwater uptake, evaporation, recharge, and interbasin or under flows,  There are a number of ways to estimate capture, from simple to complex; a lot will depend on the outcome of the first step.   

When we were looking at our objectives, did we figure out that we needed to have huge long range envelopes of time that we were evaluating, or were we concerned with something on the next year, the next irrigation season?” he said.  “It might be that we need to do the full model or it might be that we can simply look at an analog or solution rate in the proximity of the stream and only really care about what’s happening in the immediate adjacent aquifer and the stream itself.”


aquifer size, storage, and captureThe diagram shows a large aquifer with plenty of storage.  There is vegetation that is drawing from the groundwater so there is ET happening, the stream is interacting with the aquifer, so what happens if we start pumping? 

There will be minor water level changes that will affect a number of things because when we start pumping, the water comes from storage and it comes from capture,” Dr. Barth said.  “It’s not that you have some disconnected system; it’s all a continuum.  The point is that when we’re talking about stream depletion assessments, we need to keep that in mind that it’s not this huge savings account or this vast volume that you can just keep sucking dry.  We need to realize that capture happens and so therefore it affects streams at virtually any level of pumping.”

I like to preach it’s a credit card,” he continued.  “As soon as you start doing that pumping, you are going to have some sort of annual fee and you’re going to have some sort of equivalent of an interest fee.  It’s also worthy of considering that as you start this process and you affected the capture, that if you were to continue pumping, then you have to consider that in order to undo that capture that you’ve caused, you have to fill back that whole aquifer.”

persistence and metric selectionDr. Barth pointed to a 2009 study by Bredehoeft and Durbin that is a good example of persistence and metric selection.  The paper looked at a hypothetical aquifer 50 miles long by 25 miles wide with two streams feeding in at the up gradient end and a spring coming out of a down gradient end, and they considered what would happen if a well was put in the basin. 

If somebody proposes pumping from there and other people get upset that they’re going to hurt the spring, the pumper could say, ‘we won’t hurt the spring, we’ll monitor the spring, and as soon as we get to dropping 10% of the flow, we’ll turn off all pumping,” he said.  “So we start pumping and we effect storage in a large part of the basin and after 50 years, the flow has dropped 10% from its historical 100 cfs, so now it’s down to 90 cfs.  The interesting thing is looking at those calculations, it’s another 25 years until the spring got down to the lowest spring flow levels, so if they hadn’t gone through this process, somebody could be standing on the surface, scratching their heads for 25 years wondering why the spring flow is still going down, when to us as a room full of people who analyze aquifers and analyze stream depletion, it’s perfectly obvious.  This is a hideous configuration of geometry that plays havoc on attempting to monitor it, if this is the way you’re going to monitor and make your decisions.”

Final note is that spring recovery takes about 500 years,” he added.


Jocko River, Montana

The first example is from the Jocko River Valley in Montana.  The Jocko River flows northward through the valley; it’s surrounded by mountains on all sides, so it is similar to the visual on the slide in the previous section. 

They wanted to know in this valley what kind of capture are we going to have from wells and what kind of persistence of pumping, so we went through our process,” Dr. Barth sproportion of pumping from captureaid.  “We did a good job of sorting out the concerns.  We built and tested the model and we were happy that we understood the limitations of the model.”

So in the model we start putting in a well in many locations, and we just kept running the model again and again and comparing it to the base run and we were able to come up with a series of capture maps,” he said, presenting a one year pumping map and a ten year pumping map.  With the one year capture map, after one year, the values that are closer to 1, so of what comes out of the well, almost 100% has been captured from the river; the brown area has much lower values.  After ten years, virtually all the pumping that is happening is being captured from the stream.

After ten years, there’s another temporal component to consider, he said.  “If you were to pump for ten years and then just say, I’ve started to cause an impact, I’ll walk away, we can expect that it’d be about ten years that your impact would persist, so another ten years you don’t get to pump anything, but somehow you’re affecting streamflow.  Now granted, it would taper with time, but it’s one of those things that I like to convey to people when I’m talking with them about trying to find options for systems.”

Scott River Valley

The Scott River Valley is an alluvial valley with mountains on the perimeter.  There were low flows that affect habitat and a proposal for adjustments that might be able to help the fish, so they used a similar approach.

We found out what the concerns are, we collected the data, we built the model, and then we went ahead and ran it,” said Dr. Barth.  “Once again, we ran it many, many times, looking at where the impact of the wells are throughout the valley and with that kind of information, we can explore specifics, but we can also look at it on a bigger context.”

designing depletion or accretionThere were many different issues that were explored.  As an example, on the slide there are three wells indicated, just three of the many locations of wells.  Location A is located on a tributary, and the graph shows that after pumping for two years, almost 100% of what’s being pumped out of that well is coming from the tributary.  At location B which is next to the mainstem of the Scott River, after two years, about 80% is coming out of the river.  Location C is sort of split between a tributary and the Scott River.

The final product was 250,000 pages in a PDF file.  “The point is that if you sit down in front of a group of people, you don’t bring the stack of paper, but you bring parts of that and you can start to leaf through and figure out what combinations might be helpful if I’m pumping or if I’m injecting and how can I somehow help the flow downstream which might be indicative of what the fish will benefit from in the long-term,” he said.

A hybrid approach

The last example is a hybrid approach that was relatively simple.  It used a combination of a numerical model and an analytical model to the assess stream depletion associated with water transfers.  The folks wanted to look at things like, for instance, having a set of wells and saying instead of pumping these wells, can we pump those wells? 

They had a lot of questions about how much of the water transferred from here to that well and they had their formulas to go through that, but in the end, they gave us a number and said, we’re shutting down those wells and turning off that pumping rate, we’re going to start pumping here, and so we know that the model has been tested and works pretty well, and we have an analytical solution outside of the model domain that we can apply,” said Dr. Barth.  “There is a numerical grid domain that’s outlined here, and if you happen to have those three points within the domain that are your moved from location, then our interface launches MODFLOW, runs the simulations, accrues the difference and so we have a time series of accretions associated with pumping turned off.”

In this particular example, there is one moved-to well which has a distance from the river, the model does the calculations; in this case, since it was outside of the numerical grid, they implemented Glover Balmer and went through the process of generating those depletions and the results were then combined with the moved-from portion.

The point is that in the context of this session, it’s a tool that can allow you to rapidly move through a series of steps that gives you an opportunity to have discussions about what are these impacts and do we need to make adjustments,” said Dr. Barth.


Mr. Barth then summarized.  “You should plan things out carefully, you should consider what your question really is, and can you characterize the question accurately somehow,” he said.  “Design the system so that you have tools and understand the limitations of the tools – I made sure not to tell you once in this talk that a tool is accurate, but if you understand the limitations, you can still get a heck of a lot done.  And then a big part of implementing is communicating clearly and making sure people understand that aquifers are not a savings account.”

Print Friendly, PDF & Email