Hyporheic Zone Restoration 2: Results of the Bold Experiment
Streambed temperature difference map of the Kingfisher restoration site. PP4 - PP6 are plunge-pool structures, and PW is a pocket water structure. Size of blue (upwelling) and red (downwelling) dots is proportional to strength of water movement.
…Continued from “Can the Hyporheic Zone Be Restored?…
In the 1966 science fiction movie The Fantastic Voyage, a submarine and its crew are shrunk to microscopic size and injected into the bloodstream of an injured scientist in order to save his life by travelling deep into his brain to remove an inoperable blood clot. I fantasized about this movie many times when I was designing the measurements I would use to study the effectiveness of the engineered streambed, the new hyporheic zone, at Thornton Creek. The hyporheic zone is a difficult place to study. We can’t directly see into it, and if we try to expose its secrets by digging up or poking our equipment into this world beneath the streambed, we risk modifying it or destroying its integrity. Most of the techniques being used to study the physical aspects of the hyporheic zone, such as the way it cools and cleans the surface water, are quite new and not well tested. And, like all scientific studies in the natural environment, it is impossible to control everything that influences the stream. The weather, the activities of people living nearby, and even random events such as a tree falling into the creek can make the results of painstaking measurements difficult to interpret.
In the spring of 2015, less than a year after construction of the restored hyporheic zone, the engineered streambed, at Thornton Creek in Seattle, I began a series of measurements to answer two questions. First, how well is this new hyporheic zone working, at least physically? And, how long is it likely continue to function? I will first summarize what I found, after three years of measurements. Then, I will describe in greater detail how the measurements were made. In the next article, I will discuss other exciting studies of the success of this restoration project, focusing on water chemistry and biology.
August 10 - 19, 2016 upwelling (blue dots) and downwelling (red dots) rates. Larger dots mean larger rates of flow. The creek flows from upper left to lower right, and north is up. The largest downwelling rate was 3.4 meters/day (11.1 feet per pay). Only one of the wells measured upwelling (0.31 meters/day or 1 foot/day), located at the far downstream end, where all of the hyporheic water had to suddenly re-emerge. Colored diamonds are floodplain wells. Red diamonds mean subsurface water is flowing away from the creek; blue diamonds indicate water flowing towards the creek.
This is what I learned:
The local rate of water movement into or out of the streambed in the vicinity of the plunge-pool structures is very large, averaging 89 times the pre-construction rate, and 17 times larger than maximum rates measured in a pristine stream in a national forest in Idaho. This maximum rate was a whopping 8.1 meters/day (26.6 feet/day). Imagine a column 1 square foot at the bottom and 26.2 feet high moving into the streambed, every day! The average rate was 1.7 meters per day (5.5 feet per day).
The relative strengths of upwelling and downwelling throughout the restored streambed in the constructed channel were larger than in a nearby unrestored part of the creek (the control), and there was more variation over short distances in the restored section. More variation means more complex habitat – a good thing for fish!
If you were to cut a cross section of the creek and its floodplain, the area occupied by the hyporheic zone increased from essentially zero in the unrestored control, to 1.93 times as large as the cross section of the flowing surface water, after restoration.
Because of mixing with emerging hyporheic water, surface water temperatures in the restored channel actually dropped 1.5 °C (2.7 °F) while passing through the segment occupied by a step-pool structure, during a summer day, in full sunlight. Without this hyporheic mixing, the surface water would just continually heat up as it flows downstream. Warm water is harmful to salmon and trout.
The streambed gravel mixture did not return to the embedded, sand-clogged condition that had existed before restoration, indicating that the new engineered streambed is capable of being cleaned by natural processes of sediment movement.
So, in a nutshell, these results, taken together, show that the engineered streambed achieved the goal of a functional, and lasting, hyporheic zone.
What I actually did
To measure the local rate of water movement into or out of the streambed, I designed and installed special wells, called piezometers, that measured the water temperature at precise depths in the gravel below the streambed. Each day, during dry weather, the surface water goes through a cyclic temperature change, warming from morning to afternoon, then cooling during the night. As this solar heat is carried down into the streambed, the temperature cycle becomes dampened, and shifted in time. That is, the amplitude, or difference between the daily peak and the daily minimum, gets smaller, and the daily peak occurs later in time. These two differences allowed me to compute the vertical rate of water movement in the streambed.
I found that this vertical rate of water movement was quite variable from one point to another, even over short distances. Moreover, I found that the vertical water movement rate changes dramatically from year to year, and season to season, sometimes even from week to week. This was quite unexpected, and I was reluctant to believe it at first. But, another scientist studying chemical changes in the water confirmed this. He put food coloring into one of my special wells, to see where the water moving into the streambed at this well would come out downstream, and found out that the place where it came out downstream shifted quite a bit, several feet, over just a few weeks.
The piezometer wells were too expensive and difficult to install everywhere. So, to get a more complete picture, throughout the whole study area, of vertical water movement in the streambed, I had to find a simpler technique. What I did was to take 60 short pieces of flexible plastic tubing and push them 10 centimeters (4 inches) into the streambed in an array that covered more than half of the restored area. Then, we used a precision thermometer to measure the temperature difference between the bottom of each tube, in the hyporheic zone, and the surface water. Since we only had one precision thermometer, we had to cycle through the whole array, one tube at a time, over a period of about an hour. The size of this temperature difference allowed me to know whether the water was upwelling or downwelling at that point. If hyporheic water is upwelling, cold water is moving towards the streambed surface, and the temperature difference between it and the surface water will be larger. Conversely, if warm surface water is moving down into the streambed at this point, warmer temperatures are pushed deeper, and the temperature difference at this tube will be smaller. So this technique allows me to map upwelling and downwelling zones, and to know their relative strengths, but not to know the exact rate of water movement. Nevertheless, I was surprised at how diverse the streambed was in terms of upwelling and downwelling.
We also applied this technique to an unrestored control area upstream. In the control area, the upwelling and downwelling rates were not as strong, or as diverse as they were in the restored reach. This was proof that the restored reach is providing improved conditions for fish and other things living in the stream.
Both of the techniques I have just described are quite new, and have not been used very often to study the hyporheic zone. To date, the piezometer well study has really only been used by river scientists twice, to my knowledge. Most of the scientists studying the hyporheic zone have used tracer studies. So I did this as well.
In a tracer study, a chemical is slowly dripped into the stream water, at a constant rate, over the course of a few hours. The concentration of this chemical is measured every few seconds immediately downstream from this “injection point,” and also at a distant point downstream, in this case about 80 meters (260 feet) away. The tracer we used was sodium chloride solution, ordinary salt water. As the pulse tracer is carried downstream by the flow, it tends to spread out, becoming less intense, but also taking longer to go by. This is because some of the tracer is going into the hyporheic zone, where it moves more slowly. Because it is slowed down, it will continue to re-emerge from the hyporheic zone long after the source drip is turned off and the bulk of the surface water affected has flowed downstream. By comparing the tracer concentration over time with a model that predicts what this concentration would be for different sizes of hyporheic flow cross section, I was able to estimate this cross-section size. I found that it was 1.93 times, nearly double, the size of the cross-section of the surface water flow. Before the restoration, there had been essentially no hyporheic cross-section.
I was also interested in flow of subsurface water beneath the floodplain. So, I installed some monitoring wells in the floodplain, on both sides creek, upstream and downstream of one of the plunge pool structures. By measuring the water level in these wells, and comparing them with the water surface elevation in the creek channel, I could infer the direction and strength of floodplain subsurface water flow. By taking well measurements during dry weather throughout the year, I could see how this subsurface flow changes by season. I already knew that cool groundwater was emerging from the valley sides on the south and west edges of the new floodplain. So it was no surprise that the floodplain wells showed subsurface water flow in this direction. But I also observed that water was flowing outward into the floodplain’s just upstream from the step-pool structure, and then flowing back towards the stream channel on the downstream side. This was expected. The hyporheic flow is actually three-dimensional, going up and down, out and back, and moving down the valley. If this had been a natural floodplain, there would have been buried creek channels, abandoned long ago as the creek changed course. These relic channels would be conduits where hyporheic water could easily flow out into the floodplain and back. But our constructed floodplain does not have these, and the lateral hyporheic flow is limited to short distances around these restoration structures.
Finally, I wanted to address the criticism that the engineered streambed, with its clean, fresh gravel, would quickly silt in and lose its hyporheic connection. To study this, I took samples of the streambed, both the coarsened surface layer and the subsurface layer beneath. To do this sampling, we had a tall, three-sided box made of plywood with a plastic skirt around bottom edge to seal it to the streambed. By placing this box on the streambed with the open side facing downstream, we could carefully dig out a portion of streambed without the current washing the material away as we dug. These samples were dried and then sieved to determine the median size of the grains of rock making up each layer, and the percentage of sand. By comparing these measurements over time, I could see whether the streambed was indeed silting up, increasing its sand content and decreasing its median size. What I found was that the streambed sand content and median size does change over time, but not in this simple way. In some cases, the sand content decreases. We also observed cases where the median size increased, rather than decreased. This told me that the streambed is not simply silting in, but is changing in a more dynamic way, much as it does in a natural river. In natural rivers, the streambed will erode, or scour, and then fill back with new material transported in from upstream, over the course of a flood. This is how streambeds renew themselves, flushing out accumulated fine sand and silt. Thornton Creek is apparently now doing this, which is good news, because this is a crucial aspect of sustainable salmon habitat.
My study only went on for three years; it will be interesting to see if somebody can check this again in a decade to make sure that the streambed is still maintaining itself!