Reservoir Operational Performance

Streamflow in the Willamette River Basin (WRB) is regulated by 13 federally owned reservoirs. Together they operate as a system to provide flood regulation, power production, recreation, fish and wildlife conservation, irrigation, and water quality regulation. The reservoirs were constructed primarily to regulate flood flows, an objective that remains the highest priority for determining reservoir releases. The priority of other operating objectives varies by reservoir and circumstance.

The WW2100 reservoir team investigated whether reservoirs will continue to meet operational objectives, given expected hydrologic responses to climate change, population growth, and economic growth over the next century. We modeled operation of federal reservoirs by incorporating code from the US Army Corp of Engineers model called ResSim into Willamette Envision. ResSim determines water release quantities from reservoirs following a system of time and place specific operating rules.

In this analysis, we evaluated (1) the reliability of hydrologic modeling within Willamette Envision by comparing modeled reservoir inflows to observed historical data, and (2) the potential effects of climate warming on reservoir operational performance, by comparing outcomes for different WW2100 futures scenarios. Our analysis indicates that basinwide, Willamette Envision tends to underestimate inflows during the winter months but reliably produces inflows during the July to November time period. Our comparison of WW2100 futures scenarios suggests that climate warming will have a limited effect on the ability of the system to meet spring and summer flow targets. 

 

Reservoir Modeling in Brief

Reservoir modeling within Willamette Envision duplicates the rules of the US Army Corps of Engineers’ 2012 ResSim model for the Willamette River Basin (WRB). Each reservoir has operational zones, which are based on pool elevation on a particular date (Fig. 1) and rules sets associated with each zone. These rule sets are prioritized within each zone and across reservoirs, such that the model calculates release quantities and locations (e.g., Powerhouse, Regulating Outlet, Spillway) at each time step to meet the highest priority rule. Example rules are those that establish minimum or maximum flows or water elevations, such as the minimum flow targets established by the 2009 Biological Opinion (NMFS, 2008) for April to October, which vary based on the type of water year. These rules are often evaluated at important downstream locations, called control points, which are used to establish releases at reservoirs upstream. Many reservoirs can be influenced by a single control point (e.g., Salem), and many control points can be influenced by a single reservoir. For more information about reservoir modeling in Willamette Envision, refer to the supplemental materials from Jaeger et al. (2016).

Diagram of reservoir operation zones.

Figure 1. Reservoir operating zones (Lookout Point). The conservation curve (green) is the primary rule curve. The zone above the rule curve is the Flood Control Zone, while below the curve is the Conservation Zone. If the pool elevation is above the elevation demarcating the primary flood control zone (horizontal dark blue line), the zone is labeled Top of Dam Zone. If the pool elevation falls below the red line, the zone is labeled the Buffer Zone. An alternative flood control zone exists in this case, when the pool elevation is above the rule curve but below the secondary flood control line (in light blue).

Evaluation of Predicted Reservoir Inflows

We compared simulated reservoir inflows to those observed historically to evaluate the reliability of the hydrologic model in replicating the key hydrologic processes of the catchment. The three measures and their interpretation are summarized below.

  • Root mean square error (RMSE) is an absolute measure of model fit, meaning that the units are directly interpretable, such that higher values represent higher errors. RMSE is calculated as the square root of the variance of the residual. There is no absolute value for a reliable RMSE, as it must be interpreted with respect to the range of predicted values and practical application.
  • Normalized RMSE (NRMSE) is a nondimensional form of the RMSE, normalized to the mean of the observed data, which is useful for comparing between basins and across seasons with different ranges of flows. Expressed as a percentage, the NRMSE is interpreted as the error relative to the mean.
  • Nash Sutcliff Efficiency (NSE) measures the signal-to-noise ratio of hydrologic models by comparing the magnitude of model residuals to the variance in the observed data. A value of 1 represents perfect fit between observed and simulated inflows, a value of 0 represents a model that is equally well represented by the mean, and a value less than 0 indicates a model of questionable value. The reliability of NSE is compromised when extreme values are present in the dataset.

Analysis of Reservoir Operational Performance

The 13 federal reservoirs in the WRB are operated with a primary objective of flood regulation, while power production, recreation, fish and wildlife conservation, irrigation, and water quality regulation provide ancillary benefits. In this analysis, we focused on the potential effect of climate change on two high priority operational objectives: flood regulation and fish and wildlife conservation (low flow targets). We evaluated the ability of reservoirs to meet these operating objectives based on the concept of operational reliability, where reliability is defined (e.g., Hashimoto, 1982; Ray et al., 2010) as the likelihood of achieving a flow target or level of flood protection.

Selected Findings from Reservoir Performance Analysis

Observed vs. Simulated Reservoir Inflows

  • The simulated historical hydrograph of all reservoir inflows combined (Fig. 2) illustrates that the model tends to underestimate inflows during the winter months but reliably produces basinwide inflows during the July to November time period.
  • The relative model fit across reservoirs varies with the metric used to evaluate fit.

    • Based on root mean square error (RMSE; Table 1), model fit is lowest at Green Peter and Detroit reservoirs, which are among the largest reservoirs in the system.

    • Based on normalized root mean square error (NRMSE; Table 2), model fit is lowest at Blue River, Fall Creek, and Fern Ridge, likely as a result of lower flows into those reservoirs.

    • Based on Nash Sutcliff efficiency (NSE; Table 3), model fit is lowest at Blue River reservoir.

  • Model fit is generally higher for the dry summer months than for the wet winter months.

    • Based on RMSE (Table 1), model error is lower in dry summer months than wet winter months.

    • Based on NRMSE (Table 2), model fit is generally higher in dry summer months than wet winter months. Note that because this metric is normalized by mean flows, it will inherently be higher for summer months than winter months, since mean flows are higher in winter than summer.

    • Based on NSE (Table 3), model fit is generally higher for summer than winter months, with the exceptions of Cottage Grove and Fern Ridge.

  • There is no strong evidence of any trend in model fit across the tributaries with the expected degree of groundwater contributions.
  • Note that Foster and Lookout Point reservoirs were omitted from this analysis because they are located downstream of and are under the influence of reservoir operations at Green Peter and Hills Creek, respectively, complicating comparisons of historical and simulated inflows.

Total daily-averaged reservoir Inflow - observed vs. simulated 1980-2009

Figure 2. Total daily averaged reservoir inflow - observed vs. simulated for the period 1980-2009. Results were generated by taking a geometric average of daily streamflows to account for a log-normal streamflow distribution. The simulated data is from the simulated historical climate scenario (called HistoricRef), a WW2100 scenario that modeled the historical period using simulated historical weather data.

 

Table 1. Root mean square error (cms) of observed reservoir inflows vs. modeled reservoir inflows for the HistoricRef scenario, 1980-2009.

  Annual Nov-May Jun-Oct
Blue River 3.74 4.8 1.21
Cottage Grove 1.2 1.49 0.59
Cougar 4.21 5.1 2.5
Detroit 8.41 10.22 4.89
Dorena 3.82 4.9 1.2
Fall Creek 3.85 4.91 1.4
Fern Ridge 2.97 3.79 1.08
Green Peter 8.27 10.57 2.87
Hills Creek 4.06 5.11 1.76

 

Table 2. Normalized root mean square error of observed reservoir inflows vs. modeled reservoir inflows for the HistoricRef scenario, 1980-2009.

  Annual Nov-May Jun-Oct
Blue River 41% 33% 65%
Cottage Grove 24% 19% 63%
Cougar 22% 19% 27%
Detroit 16% 15% 19%
Dorena 27% 22% 44%
Fall Creek 34% 28% 52%
Fern Ridge 31% 24% 122%
Green Peter 24% 20% 34%
Hills Creek 15% 14% 13%

 

Table 3. Nash Sutcliffe efficiency of observed reservoir inflows vs. modeled reservoir inflows, for the HistoricRef scenario, 1980-2009.

  Annual Nov-May Jun-Oct
Blue River 0.71 -0.48 0.55
Cottage Grove 0.91 0.7 0.43
Cougar 0.83 0.04 0.71
Detroit 0.88 0.12 0.81
Dorena 0.88 0.46 0.8
Fall Creek 0.8 0.08 0.61
Fern Ridge 0.91 0.81 -3.26
Green Peter 0.88 0.29 0.81
Hills Creek 0.9 0.43 0.92

 

Simulated vs. Future Reservoir Inflows

  • We compared total systemwide reservoir inflows between the simulated historical scenario (HistoricRef) and two futures scenarios with different intensities of climate warming. These two scenarios were the Reference Case, which included moderate climate warming (4° C or 7.5° F) warming over the 21st century, and the High Climate Change scenario (with 6° C or 10.5° F) of warming over the 21st century. Refer to the climate page for more information about these scenarios. The analysis suggests that relative to the past, future reservoir inflows to the system will be the same or higher except for the early summer months (May-July) and late fall, primarily for late century.

Total reservoir inflow – Reference scenario

Figure 3. Total daily averaged reservoir inflows for early, middle, and late century periods of the Reference scenario. For comparison, values from the simulated historical scenario, 1980-2009, are also shown.

Total reservoir inflow – HighClim scenario

Figure 4. Total daily averaged reservoir inflows for early, middle, and late century periods of the HighClim scenario. For comparison, values from the simulated historical scenario, 1980-2009, are also shown.

Flood Regulation

  • Model results indicate that regulation of floods at Salem is very high, represented by zero days above flood stage for the simulated historical scenario. Flood regulation at Salem remains high in nearly all years for the Reference and HighClim future scenarios (Fig. 5).
  • The No Reservoirs scenario illustrates the degree of flood regulation provided by the reservoirs, the benefits of which are particularly evident in the mid- and late-century time periods when flood stage is regularly exceeded.
  • Model calibration emphasized summer flows since the project focused on water scarcity. As a result, model fit for reservoir inflows was lower in winter than in summer, introducing some uncertainty in the results regarding impacts of climate change on flooding.

Time reliability of flood regulation on the Willamette River at Salem

Figure 5. Time reliability of flood regulation on the Willamette River at Salem, Ore., for three WW2100 scenarios. The Reference scenario includes a moderate degree of climate warming (weather inputs were derived from the MIROC global climate model). The No Reservoirs scenario includes these same weather inputs, but models behavior of the river system as though no federal reservoirs were present. The High Climate Change scenario models the system with reservoirs, as in the Reference scenario, but includes more intense climate warming (weather inputs derived from the HadGEM global climate model). For more information about WW2100 modeling scenarios, refer to the scenarios page.

Spring and Summer Flow Targets

  • Model results indicate that, with a warmer climate, the number of days when summer targets are met increases in the future relative to the simulated historical scenario. This counterintuitive effect is greater for the Reference scenario than for the HighClim scenario. This reflects the increase in runoff projected for the April-May time period (Figs. 3, 4). However, future reservoir inflows are lower than the simulated historical during the July-August months when the reservoirs appear to provide higher reliability in summer targets.
  • Results for the No Reservoirs scenario also illustrate the role of reservoirs in augmenting flows from July to October. In this scenario (Fig. 6), the number of days when late summer flow targets are met is nearly zero.
  • The No Reservoirs scenario also demonstrates the limited effect of the reservoirs on providing flows for the spring and early summer months. Even without reservoirs, the scenario shows a high reliability of meeting flow targets from April to June.

Time reliability of spring and summer flow targets on the Willamette River at Salem

Figure 6. Time reliability of spring and summer flow targets on the Willamette River at Salem for the Reference, HighClim, and No Reservoirs scenarios. Results are plotted as the ratio of days above the flow target relative to the number of days in a given month.

Conclusions

  • Model fit: basinwide the model tends to underestimate inflows during the winter months but reliably produces inflows during the July to November time period.

  • Flood regulation: climate change appears to have limited to no effect on flood regulation. However, the model underpredicts the historical winter inflows to reservoirs, indicating that this finding has some associated uncertainty.

  • Spring flow targets: climate change appears to have limited to no effect on the ability of the system to meet spring flow targets.

  • Summer flow targets: climate change appears to have a positive effect on the ability of the system to meet summer flow targets in the late century, perhaps because of projected increases in spring runoff.

Related Links & Publications

Contributors to WW2100 Reservoir Modeling

  • Desiree Tullos, OSU Biological & Ecological Engineering (lead)

  • Cara Walter, OSU Biological & Ecological Engineering

  • Kathleen Moore, PhD Student, OSU Geography (Graduated: 2015; now a post-doctoral researcher, OSU Applied Economics)

  • Matt Cox, OSU Biological & Ecological Engineering (now at Interfluve, Hood River, Oregon)

References

Jaeger et. al. (2016). Scarcity amid abundance: Water, climate change, and the policy role of regional system models. Manuscript in preparation.

National Marine Fisheries Service. (2008). Continued Operation of 13 Dams & Maintenance of 43 Miles of Revetments in the Willamette Basin, OR. (PCTS Tracking No. NWR 2000-2117). Portland, OR.

Hashimoto, T., Stedinger, J. R., & Loucks, D. P. (1982). Reliability, resiliency, and vulnerability criteria for water resource system performance evaluation. Water Resources Research, 18(1), 14-20. http://dx.doi.org/10.1029/WR018i001p00014

Ray, P. A., Vogel, R. M., & Watkins, D. W. (2010). Robust optimization using a variety of performance indices. In Proceedings of the World Environmental and Water Resources Congress, ASCE, Reston, VA.

 

Web page authors: D. Tullos, C. Walter, K. Moore
Last updated: August 2016