Effects of Island Park Reservoir on Sediment and Phosphorus Transport

This week I attended Idaho Department of Environmental Quality's annual Water Quality Monitoring workshop in Boise.  This was the workshop's 25th year, although it was my first time attending. HFF was invited to participate because of the large amount of water quality data we are now collecting and analyzing. This week's blog consists of my presentation at that workshop.

Here is a link to the powerpoint slides. The following text explains each slide.

  1. Photo of Island Park Reservoir. My presentation co-author is Ken Lindke, one of my best former statistics students, who helped me with data analysis.
  2. U.S. Bureau of Reclamation (USBR) contributed laboratory analysis in 2013. A large number of HFF staff, interns, and volunteers assisted with field data collection in both 2013 and 2014. Thank you to those of you who have helped fund the water quality project.
  3. Photo on the outline slide is looking downstream from the dam, across the Island Park hydroelectric plant.
  4. Remember that 2013 was a very dry year. At the beginning of irrigation season in 2013, water managers were predicting that Island Park Reservoir could be drawn down to as little as 10% of capacity, prompting concerns that sediment could be mobilized from the reservoir bottom and delivered downstream. In 1992, when the reservoir was drawn down to minimum pool, between 50,000 and 100,000 tons of suspended sediment were delivered into the river downstream.
  5. Both HFF and USBR wanted to monitor sediment delivery in 2013, should the reservoir get low enough that large amounts of sediment might be mobilized and delivered. This was the motivation for our first objective. Phosphorus is an important nutrient in aquatic systems and is generally what we refer to as the “limiting” nutrient, meaning that when more phosphorus is added to the system, it becomes more productive. Since phosphorus is often transported on sediment particles, suspended sediment (SS) and total phosphorus (TP) concentrations are often highly correlated. In 2013, the only location upstream of the reservoir that we sampled was in the mainstem Henry’s Fork, and we know that a large percentage of both flow and sediment input to the reservoir comes from the tributaries to the west end of the reservoir, most notably Sheridan Creek.  So, we added Sheridan Creek as a sampling site in 2014 so we could quantify inputs of SS and TP as well as outputs. Long-term objectives are to understand the processes that determine loads of SS and TP into and out of the reservoir and the consequences of SS and TP export on the aquatic system downstream.
  6. The Henry’s Fork watershed lies at the upper end of the Eastern Snake River Plain and transitions into the Yellowstone Plateau. The absence of the apostrophe in “Henrys” conforms to the official USGS nomenclature. I’ve always used the apostrophe for grammatical reasons, but I must admit that I’m starting to omit it more and more, primarily because I frequently write for scientific publication, where precision dictates that only official names be used.
  7. All readers of this blog are familiar with the study area and know these sites well. Sheridan Creek was selected as a representative of all of the streams that drain the south side of the Centennial Range and flow into the reservoir. The reason it is important to include the input from these streams is because the Centennial Range is the major geologic source of phosphorus in the upper Henry’s Fork watershed. The volcanic rocks that dominate most of the upper watershed contain very little phosphorus.
  8. As mentioned above, in 2013, we sampled only at Flatrock (actually just a little ways downstream at Coffee Pot rapids) and at the dam. We measured turbidity, SS, and TP at different locations across the channel and found that location accounted for less than 1% of variability in the measurements. So, in 2014, we measured only one point in the center of the stream channel. We also reduced sampling from four days per week to two days per week, because our sampling error was so low at 4 days/week that we could still attain very high precision at 2 days/week and afford to expand the number of sites and constituents.  In 2014, we added orthophosphate (OP) as a constituent because it is a measure of inorganic forms of phosphorus, which are the forms that can be absorbed and used by plants.
  9. Water is stored in Island Park Reservoir primarily during the winter and early spring and then delivered for irrigation during July and August. Reservoir capacity is 135,000 acre-feet. On average, the reservoir is drawn down to around 60,000 acre-feet (45% of capacity), but in below-average water years, it is drawn down below this volume.
  10. In general, there is a mild positive correlation between SS and TP concentrations and river flow upstream of the reservoir but not downstream of the dam, where there is no correlation at all. In unregulated streams, nutrient and sediment concentrations are generally highest when streamflow is highest, but where flow is regulated by a large dam and reservoir, these correlations disappear because streamflow is disconnected from physical and biological processes that occur in the watershed.
  11. In 2014, sediment was stored in the reservoir during snowmelt runoff, in May, and then was released throughout the summer and fall. Over the period from mid-May to the end of the calendar year, less sediment was released than came into the reservoir, for a net storage of about 105 tons.
  12. In 2013, when the reservoir was drawn down to about 35,000 acre-feet (25% of capacity), there was a clear trend toward increasing sediment concentration as the reservoir was drawn down. In 2014, when the reservoir stayed above 60,000 acre-feet, there was no relationship at all between the reservoir volume and suspended sediment concentration.
  13. This graph shows reservoir volume and turbidity, which is a surrogate for suspended sediment. It is obvious that the amount of sediment delivery and the magnitude of drawdown in 2013 were nowhere near those of the 1992 event.
  14. Phosphorus loads showed a similar pattern—storage of phosphorus in May and release in the middle of the summer. However, much more phosphorus was released than flowed into the reservoir, indicating that biochemical processes in the reservoir converted phosphorus to a soluble form that was readily transported during the middle of the summer. Another possibility is that additional phosphorus was being added to the reservoir through groundwater, which we did not measure. Houses around the reservoir, all of which have septic tanks, provide a potential source of additional phosphorus to the reservoir.
  15. In both Sheridan Creek and the Henry’s Fork upstream of the reservoir, phosphorus concentrations were positively correlated with sediment concentrations, indicating that phosphorus did enter the reservoir on sediment particles. Note, however, the absence of this relationship downstream of the reservoir, indicating that phosphorus was being transported in the river downstream primarily in solution, rather than on sediment particles. The horizontal line at 0.1 mg/L indicates the maximum TP concentration that is generally considered desirable in rivers and streams. Higher concentrations can lead to undesirable, excessive production of aquatic plants and/or algae.
  16. Additional evidence for these conclusions is provided by the fraction of total phosphorus (TP) in the inorganic form (orthophosphate, or OP). Almost none of the phosphorus inputs, from Sheridan Creek and the upper Henry’s Fork, was from OP, whereas the highest concentrations of TP downstream were attained when OP comprised the majority of the phosphorus. This suggests that decomposition of organic matter in the reservoir in mid-summer was releasing inorganic phosphorus (OP) into the water column, so it could be exported from the reservoir in solution, independent of release of suspended sediment. Most of this organic matter was probably algae, which reaches peak growth in the reservoir in late summer. Decomposition of algae is accelerated late in the summer, when the reservoir stratifies, and oxygen concentrations become very low near the reservoir bottom. In a nutshell, poor water quality in the reservoir late in the summer converts phosphorus to a dissolved form that can be readily exported downstream.
  17. The numbers speak for themselves; suspended sediment loads in late summer 2013 and 2014 were nowhere near what they were during the 1992 event. However, both SS and TP loads were higher in 2013 than in 2014, primarily because a lot more water was delivered from the reservoir in 2013 than in 2014. Load is concentration multiplied by flow rate, so even if concentrations are similar, a higher flow rate will yield a higher load. But, we also know that suspended sediment concentrations were higher in 2013 because the reservoir was drawn down to a lower level, so SS load was higher in 2013 for both reasons. Higher phosphorus loads in 2013 were primarily a function of higher outflow, not higher concentrations.
  18. The large amount of phosphorus exported from the reservoir fuels the growth of macrophytes downstream. We now know from other studies that macrophytes are the primary driver of physical habitat conditions in the Harriman reach, contributing depth, cover, high dissolved oxygen levels, and habitat for insects and other invertebrates.   
  19. These data were collected by Zach Kuzniar, as part of his graduate work on habitat use by adult rainbow trout in the Harriman reach.  The box plots summarize measurements taken at 180 random points in the Harriman reach over the summers of 2013 and 2014. Note that macrophyte cover peaked in August in 2013 but not until September in 2014. In 2013, the weather was hot and sunny until September, which was relatively cool and rainy. On the other hand, weather was relatively cool and rainy most of the summer in 2014, but September and October were warm and sunny. Thus, the macrophytes grew more early in the summer in 2013 but were delayed in reaching peak growth until September in 2014.
  20. This is a time series plot of dissolved oxygen (DO), as recorded by our continuous instruments at four sites in the watershed. For reference, DO concentrations of greater than 6 mg/L are needed for trout survival. Optimal concentration for trout are 8-12 mg/L. Note that DO immediately downstream of Island Park Dam was below 6 mg/L for most of the summer and declines throughout the mid- to late-summer. This reflects low DO levels in the reservoir during the time when OP concentrations were high, providing yet more evidence that decay of organic matter in the reservoir is the reason for the high OP concentrations and overall high export of phosphorus from the reservoir during late summer. The other thing to notice in this graph is the large amplitude of daily cycles in DO at Pinehaven. This reflects very high productivity in the aquatic system, driven by macrophytes. These daily cycles in DO, produced because the plants produce oxygen only during daylight hours, correspond to the cycles in water depth I described last week.
  21. These conclusions are based on only two summers of data; to construct long-term models of sediment and phosphorus inflow to, storage in, and outflow from Island Park Reservoir, we will need at least a decade of data, which we are now positioned to collect. For now, we can attribute the incredible productivity—including the size of our wild rainbow trout—of the Henry’s Fork downstream of Island Park Reservoir to a combination of natural geologic and hydrologic processes that deliver sediment and phosphorus to the reservoir, storage of sediment and phosphorus there, and algae growth and low oxygen levels in the reservoir that convert phosphorus to a form that can be exported downstream and readily used by macrophytes.
  22. The end.