Sediment Transport in 2020: What Caused the September Event?
Sediment export from Island Park Reservoir was 134% of average in 2020, due primarily to the highest reservoir draft since 2016.
Thanks to a springtime freshet and low aquatic vegetation growth, a net of 1,170 tons of sediment were removed from the Island Park to Pinehaven reach in 2020.
A sediment event at the reservoir September 8-25 produced the highest turbidity we have measured there but accounted for only 7% of the total annual sediment load.
The event exported 152 tons of sediment above background levels. Fate of that sediment was:
22 tons deposited between the dam and Pinehaven
58 tons deposited in Ashton Reservoir
22 tons deposited between Ashton and St. Anthony
50 tons were still in suspension at the Parker (Red Road) bridge
I analyze two theories to explain the September event: 1) an earthquake swarm September 10-16 and 2) a rare northeast wind event September 7-8.
Background on Suspended Sediment
The title of my recent water quality blog was "High Turbidity Headlines 2020." Turbidity in the river at Island Park Dam, Pinehaven, and everywhere in between was above-average during the spring, summer, and fall of 2020. Meanwhile, turbidity in other river reaches was generally below average this year. As explained below, turbidity and suspended sediment are related, so sediment concentrations in 2020 were higher in locations affected by Island Park Reservoir and lower elsewhere.
For the purposes of water quality, the vast majority of biological activity, sediment transport/deposition, and concerns over dissolved oxygen and water temperature occur during the spring, summer, and fall, generally April 1 to October 31. This is the period during which we deploy all of our water-quality sondes; only sondes at Island Park Dam, Buffalo River, and Marysville can stay in the river all year. So, the "annual" or "year" figures in this blog refer to those over the April 1 - October 31 period. The following map shows the locations of our water quality instrument network, for reference.
What is river sediment?
River scientists and managers used the term "sediment" to refer to any and all mobile, non-woody material in the stream channel and floodplain. Sediment size ranges from the finest clay particles all the way up to large cobbles and boulders. As long as it is capable of being mobilized by the river at some flow level, then it is considered sediment. Many anglers tend to think of sediment as only very fine material, but fine material is only one component of total sediment in the river and floodplain. Although sediment as primarily mineral (inorganic) in nature (sand, gravel, etc.), sediment can also contain biological material such as decomposing plant matter. The organic fraction of sediment can play very important roles in aquatic food webs, including providing food for insects, which in turn provide fish food and enhance the fishing experience.
When sediment is transported in a river system, it is moving either as bedload or suspended load. Bedload consists of sediment particles that are too heavy to be suspended in the water column but rather move downstream by essentially rolling along the stream bottom. Bedload transport occurs only when stream power is high enough, due to a combination of water velocity and gradient, and so is most common in high-gradient stream reaches when streamflow is high. Suspended load is what it sounds like: particles light enough to be suspended in the water column are transported downstream in the current. The size of particles that can remain in suspension is determined by the amount of turbulence in the river. Higher turbulence will suspend larger particles. Although rivers always carry some amount of suspended load, suspended load is generally higher when flows are higher, not only because turbulence is higher but also because high flows in natural river systems are usually associated rain or snowmelt, which directly introduces sediment into the stream channel due to overland flow of water.
In most reaches of the Henry's Fork and its tributaries, stream power is low most of the year because of relatively low peak flows and low gradients. As a comparison, a natural peak flow event in the Henry's Fork upstream of Island Park Reservoir is roughly 2-4 times the magnitude of the river's typical low flow (called baseflow), and the gradient of the river is only around 3 feet per mile (0.06%). A natural peak flow event on the South Fork Snake River is around 8-15 times baseflow, and the gradient is 11 feet per mile (0.2%). Thus, bedload transport does not occur very often in most reaches of the Henry's Fork, and when it does, the largest particles that move are gravel and very small cobble. Those of you who have fished the river for a long time know that the same riffles and rocks are in the same places in the river they have been for decades. Compare that to the South Fork, say in the area around Lorenzo, where the entire river may be in a completely different place from one year to the next due to bedload transport. Bedload transport is important in the Henry's Fork Watershed primarily in Fall River, Teton River, and the Henry's Fork downstream of St. Anthony. Most sediment transport in most reaches of the Henry's Fork occurs as suspended load. Thus, for the remainder of this blog, "sediment" will refer to suspended sediment, which consists of relatively fine particles.
Sediment dynamics in the Henry's Fork
In general, suspended sediment dynamics in the Henry's Fork conform to the usual laws of river sediment dynamics: sediment is mobilized from the stream bottom when streamflow increases, is moved downstream in suspension during high flows, and then is deposited wherever it happens to be when streamflow drops to the point that the sediment particles can no longer stay in suspension. So, if the river were not regulated by dams and diversions, sediment would be mobilized and transported during the spring freshet, deposited on the stream bottom as flows recede in early summer, and pretty much stay where it was deposited until the following spring.
However, the Henry's Fork is highly regulated, which turns sediment dynamics on its ear. Island Park Reservoir captures all bedload and most suspended load during runoff, and that material is stored on the bottom of the reservoir. Thus, the river downstream carries much less sediment during the spring freshet than it would if the dam weren't there. Sediment is then transported out of the reservoir during mid- to late-summer, when flows are high to deliver irrigation water, reservoir drawdown exposes sediment to the effects of wind, rain and erosion in the old river channel, and when outflow occurs through the bottom-withdrawal gates rather than through the power plant. The bottom-withdrawal gates sit at the lowest elevation in the reservoir, in the old river channel where sediment accumulates. The power plant intake is located in a little bay off to the side of the gates and is higher up in the water column.
During the mid- to late summer when sediment delivery out of the reservoir is at its peak, the river channel downstream, especially in Harriman State Park, is filled with aquatic vegetation (what we call macrophytes). These macrophytes slow the water velocity and allow the suspended material to settle out. Thus, most sediment delivered from Island Park Reservoir during the summer and fall ends up being deposited in the river between the dam and Pinehaven. This material can be remobilized and carried out of that reach only after the macrophytes have died off in the winter. Of course, that is a time of low flows, so stream power isn't sufficient to mobilize that material until the spring. There is only a relatively short time period when sediment can be mobilized and transported out of that reach, generally from late March until early July. That is also a time when water is being stored and retained in the reservoir for release later in irrigation season, which is why a managed freshet such as the one delivered in late April and early May of this year is so important to moving sediment out of the Henry's Fork between Island Park and Pinehaven. It is also why one of our water management goals is to keep irrigation-season flows as low as possible, to limit the amount of sediment transported out of the reservoir in the first place.
There is also a major feedback loop at work in sediment dynamics downstream of Island Park Dam. More sediment deposition creates more nutrients and substrate for macrophyte growth, and more abundant macrophtye growth traps more sediment, etc. That cycle will continue to re-enforce itself over periods of years when the reservoir is drafted heavily, resulting in net deposition and storage of sediment in the Harriman Reach. This occurred during the early to mid-2000s and again during 2013-2016. Breaking the cycle and pushing it back the other way requires a period of years with low reservoir draft and managed springtime freshets, as we had between the summer of 2017 and spring of 2020.
Why is sediment of such concern?
Obviously, seasonal sediment mobilization, transport, and deposition are natural features of rivers and necessary for maintenance of physical habitat and food webs. But in the Henry's Fork, where sediment dynamics are driven by dam management and not by natural processes, fine sediment can build up in the river over periods of years, as described above. Over time, fine sediment exported from Island Park Reservoir and deposited in the Harriman Reach degrades habitat for aquatic insects, which are arguably the most important feature of the fishery in that reach. Most trout spawning occurs in the Buffalo River and immediately downstream of the dam, where sediment deposition is not nearly as problematic as it is in the Ranch. We know from previous fish population research that spawning is not a limiting factor in the Henry's Fork. Rather, winter survival of juveniles is limiting, and that is directly related to streamflow in Box Canyon--the original reason HFF advocated for limiting draft of the reservoir.
The bottom line is that fine sediment deposition has a negative effect on aquatic invertebrates in the river between Last Chance and Pinehaven. My blog from the spring presents data clearly linking water management, sediment, and aquatic insects.
How do we measure suspended sediment?
The only direct way to measure suspended sediment is to collect a water sample, filter the suspended material, and weigh it. We have been collecting weekly water samples at various locations around the watershed since 2013. To date, we have been sending these to a commercial lab in Pocatello, which reports results back to us 10-14 days after we collect the sample. The sediment concentration is reported to us as a mass per unit volume of water, in the case milligrams per liter (mg/L). This process is expensive, both in staff time and cash, and it takes too long for us to detect and measure events in real time.
In 2014, we began installing our network of water quality sondes, which record information every 15 minutes. Unfortunately, the sondes can't directly measure suspended sediment, but they can measure turbidity, which is defined in terms of light penetration through the water. Fortunately, turbidity and suspended sediment concentrations are closely related. Since 2014, we have collected enough suspended sediment samples in the field that we can pair observations of suspended sediment with turbidity measurements collected at the same time. We then fit a statistical model to the data, which allows us to calculate an approximate sediment concentration from the turbidity data recorded by the sonds and transmitted to our water quality website in real time. Some examples of these suspended sediment-turbidity relationships are shown in the graphic below. Obviously, the relationships aren't perfect, but they are sufficiently good to calculate sediment concentrations without having to collect physical water samples 24/7. We can even quantify the uncertainty in these relationships and use that to put statistical confidence intervals around our calculations.
While sediment concentrations reflect turbidity, which in and of itself has a negative effect on fishing via reduced visibility, it is total sediment load that determines ecological effects. Sediment load is concentration (mass/volume) multiplied by streamflow (volume/time). Thus, load is measured as mass/time. We usually measure suspended sediment load in tons/day, which is what I use in this blog. Multiplication by streamflow is why total suspended load out of Island Park Reservoir is highest during irrigation season. That's when streamflow is highest. Sediment concentrations are often higher during the fall, when the reservoir is drawn down, but total load is lower then because streamflow is lower. Typical mid-summer sediment concentration at Island Park is around 6 mg/L. At a typical irrigation season flow of 1,200 cfs, this is 19 tons/day. In the fall, when streamflow is 400 cfs, a concentration of 18 mg/L gives the same sediment load. Until this September's event, we had never recorded a concentration higher than 16 mg/L, yet we routinely see loads in the range of 10-20 tons/day during irrigation season because flow is high, even though sediment concentration is relatively low.
Watershed-wide Sediment trends in 2020
Peak sediment concentrations--on the order of 20-25 tons/day--were observed at Pinehaven, Marysville, and St. Anthony during the spring and at Island Park Dam during mid-September. By far the two locations with the lowest concentrations were Flat Rock, which carries very little sediment even during runoff, and Ashton Dam. Unlike Island Park, sediment trapped in Ashton Reservoir is rarely released downstream for a variety of reasons. Ashton is a run-of-river power reservoir and is not drafted except very infrequently for repairs and maintenance. As a result, sediment concentration downstream of Ashton Reservoir is pretty constant throughout the year.
Graph of suspended sediment concentration.
The graph of suspended sediment load, shown below, looks quite different. Although the scale makes it difficult to see seasonal changes at most locations, I have used the same scale for all six locations to illustrate that load generally increases with distance downstream of the headwaters, as streamflow increases. Island Park, Pinehaven, Marysville, and St. Anthony all had sediment concentrations in the range of 20-25 mg/L at some point in the summer, but loads were much greater at St. Anthony because sprintime streamflow there is so much higher. You can see peaks in sediment load at Pinehaven, Marysville, and St. Anthony during each of the spring rain events we experienced. These peaks are very small, if they show up at all, at Island Park and Ashton because the reservoirs trap sediment associated with these natural runoff events. Note also that sediment load at Island Park during peak irrigation delivery in late July was about the same as during the September event, for the reason explained above.
Graph of suspended sediment load.
Total seasonal (April 1 - October 31) loads at each major river location are given in the table below. The entry for the Buffalo River and other tributaries between Island Park Dam and Pinehaven was estimated from turbidity/suspended sediment measured in the Buffalo River and from streamflow measured in the Buffalo River, with a multiplication factor to account for additional inflow from springs and small tributaries between the Buffalo River and Pinehaven (e.g., Blue Springs Creek and Thurmon Creek). We gaged those smaller tributaries a few years ago and found that their total flow was around 10% of that of the Buffalo River. All of those tributaries are spring-fed and so have similar sediment and flow characteristics to the Buffalo River.
Aside from the obvious observation that loads get larger with distance downstream, the table immediately shows that sediment loads at the maintstem locations least affected by Island Park reservoir had below-average loads in 2020. Remember that total water-year natural flow was 92% of average, and sediment loads at Flat Rock, St. Anthony, and Parker reflect that. However, sediment loads at Island Park, Pinehaven and Marysville ("Ashton Reservoir inflow" in the table above) were all around 135% of average. This is because of a relatively short averaging period (2014-2019) and the fact that three of the six years in that period of record had very low draft of Island Park Reservoir and low irrigation-season flows. This was the first year since 2016 that the reservoir has been drafted lower than 70% of average and that peak outflow exceeded 1,300 cfs. The Buffalo River is a bit of an outlier here--sediment load there was above average, despite below-average streamflow. Most of the additional sediment in the Buffalo River in 2020 was due to a large spike in sediment concentration, not flow, in early May and so could have been related to operations at the Buffalo River hydroelectric plant. Our sonde sits just upstream of the hydro dam and would reflect operations that mobilized sediment in the forebay.
The table above allows calculation of a sediment balance for the Island Park to Pinehaven reach. Input to the IP to Pinehaven reach is sediment load at Island Park Dam plus the contribution of the Buffalo River and other tributaries. During 2020, total input to the reach was 2,510 tons, while outflow at Pinehaven was 3,680 tons. The difference of 1,170 tons is sediment removed from the intervening river reach. Around 167 tons, 14% of the annual total, were scoured and removed just during the 6-day freshet operation in late April and early May, showing the effectiveness of that operation. Updated data and calculations showed that the net export during the freshet was smaller than I had reported in the spring. The sediment budget for the whole year can be seen in the graph below. The highest daily net sediment export from the reach (difference between blue curve and solid red curve), around 65 tons, occurred on the first day of the freshet operation. Note that sediment sediment outflow was roughly equal to inflow from late July to late October, with the exception of the September sediment event, when more sediment was delivered from the reservoir than flowed past Pinehaven. More details on that below.
At least over the short period of record shown in the table, mean sediment removal from the Island Park to Pinehaven reach is around 770 tons/year. At that rate, 65-130 years would be required to mobilize and transport the 50,000-100,000 tons that were deposited during the 1992 event. However, application of mean sediment concentrations to a longer period of flow (remember that streamflow is more important in determining sediment load out of Island Park than concentration) puts the annual estimate at more like 1,500 tons per year. At that rate, 33-66 years would be required to move all of the 1992 sediment out. Either way, our data show that the 1992 sediment is being slowly moved out of the river, that the rate of transport can be accelerated by delivering a springtime freshet, and that in all likelihood, there is still legacy sediment from 1992 in the river.
September sediment event
The graphs of sediment concentration clearly show an unusual event that originated at Island Park Dam and propagated all the way to the bottom of the watershed through mid- to late-September and even into October.
Timing and duration
I defined the beginning and end of the event based on deviation from the background trends in suspended sediment concentrations during September and October. The beginning of the event at each location was defined as the first day of increased concentration following the initial increase at Island Park Dam. The end was defined as the first day of decreased concentration after the peak at the given location. I used concentration rather than load to define the event, because load was changing independently of concentration throughout the event due to changes in outflow from Island Park Reservoir and diversions in the lower watershed that were being made throughout the event as irrigation demand was dropping. Once the beginning and end of the event were defined at each location along the river, I interpolated sediment concentration between the first and last day of the event to estimate the background trend in sediment concentration that was occurring independent of the sediment event. The graph below clearly identifies the event and shows the background sediment concentration trends at each location.
Next, I calculated total load and estimated background load from the concentrations, the difference being sediment load attributable to the sediment event, over and above background sediment load. Those are shown in the next graph.
Keeping in mind that the shaded area in the graphs above is the total sediment load due to the event, it is fairly easy to see that the amount of sediment left in the river at our Parker-Salem sonde location was quite a bit lower than what was exported from the reservoir to begin with. Thus, sediment was being deposited in various river reaches as it moved down the river. It is also easy to see that several days were required for the sediment to move from Island Park to other locations downstream and than the sharp initial peak and relatively steep decay at Island Park were flattened ("attenuated") as the sediment moved downstream. The peak at Island Park occurred on the second day of the event, whereas it occurred over a week into the event at Parker. The duration of the event also increased as the sediment moved downstream.
The table below lists the start, end, and duration of the event as it moved downstream. Two days were required for the event to be detectable at Pinehaven and Marysville, and another two days were required for the event to be detectable downstream of Ashton Reservoir. On the back end, sediment concentrations at Island Park Dam returned to background levels in 17 days, but that number increased to 26 days at the bottom of the watershed. Thus, the effects of this event were still present in the river 31 days after it began!
Sediment load and fate
The load numbers in the table above correspond to the shaded areas in the previous graph. These are the total loads at each location above background levels, that is, the load due strictly to the event. Load at Island Park Dam was 152 tons, about 7% of the total annual load. The load would have been much higher had this event happened earlier in the summer, but outflow from the reservoir was reduced from 680 cfs on September 8 to 490 cfs on September 9 and again to around 330 cfs on September 22. Event load was 130 tons at Pinehaven, indicating that 22 tons were deposited in the river between the dam and Pinehaven, most likely in the Harriman reach. Load at Marysville was a little higher than at Pinehaven, due to some combination of imprecision in the calculations and to the fact that background sediment concentrations were increasing at Marysville throughout the event. In any case, there is little evidence that much of the sediment was deposited in the reach between Pinehaven and Marysville. Load at Ashton Dam was much lower than at Marysville, showing net storage of 64 tons in Ashton Reservoir. If we assume that retention there was proportional to the difference between the 144 tons at Marysville and 130 tons at Pinehaven, the amount of sediment originating at Island Park that was stored in Ashton Reservoir was 58 tons. Another 22 tons were stored in the reach between Ashton and St. Anthony, a river reach that, like Harriman, supports a high density of macrophyte growth during the late summer. The remaining 50 tons was still in suspension at Parker. So, the data strongly indicate that sediment was deposited in river reaches with high macrophyte abundance and in Ashton Reservoir but nowhere else.
Although 22 tons of the sediment from this event were deposited in the Harriman Reach, a net of 1,170 tons were removed from the reach over the year as a whole. One way to look at it is that 1,192 tons would have been removed without this event. But another way to look at it is that removing 167 tons with the freshet event in the spring was an insurance against deposition later in the summer, regardless of whether the sediment came from irrigation-season draft of the reservoir or an unusual event. Either way, 22 tons is a small fraction of the overall sediment budget for the Harriman Reach. We will be able to detect any long-term effects from our annual sampling of aquatic invertebrates next March.
In the short term, however, we decided to repeat some sampling of habitat conditions in Harriman that was done by Zach Kuzniar in 2013 and 2014 as part of his Master's thesis research at Grand Valley State University. We used the same methods and the same GPS-located sites at the same time of year--the third week in September. The three most important variables we measured were macrophyte cover, stream substrate size, and river depth. Macrophyte cover averaged 68% in 2013, 90% in 2014, and 64% in 2020. Had macrophyte cover been higher this year, more sediment from this event would have been trapped in the Harriman reach. Given the feedback loop I described earlier, water management over the past few years in the form of springtime freshet flows, low reservoir draft, and relatively low irrigation-season flow at Island Park Dam is partly responsible for lower macrophyte cover this year, which, in turn, minimized retention of sediment during the event.
Not surprisingly, we observed a very small but statistically significant reduction in substrate size. Median particle size at the 90 random sample locations averaged 2.70 mm in 2013, 2.70 mm in 2014, and 2.12 mm in 2020. Qualitatively, the median particle size in all three years fell within the category of "fine gravel." However, the most surprising result we found was that after accounting for the effect of flow (we sampled at lower flows in 2020 than in either 2013 or 2014), the average depth of the river was 2.2 inches greater this year than in 2013. At a mean depth of 20.9 inches in 2013, 2.2 inches is 11%. Depth was greater in 2014, even after accounting for flow, due to the displacement effect of the heavy macrophyte growth that year. But, a 2.2 inch improvement in depth since 2013 at the same macrophyte cover indicates overall scour of the stream channel since then, more evid