Wetlands Page Banner
KWERI Our Mission Our Values Our Goals Our Vision Organization Wetlands Projects Strategy Publications Progress Chiloquin.Us

Historical Function of this Lake System


We will describe first our hypothesis for how the UKL/AL system functioned historically. It is important to have a well-reasoned view of how the system worked prior to the arrival of the Nouveau Americans in order to describe well the restoration outcome we might expect to achieve. Such a perspective will also help us prioritize actions that would most efficiently and effectively achieve that end. A couple of the key principles of ecological restoration are to: 1) restore systems to functions as close as possible to those employed naturally and historically, because those functions were evolved in place to deal with the local conditions, and 2) to minimize the engineered element of the restoration process and the final solution to allow natural processes to dominate the effort of restoration and to eventually to pass management to those natural processes (Reference).

The Waters

Historical accounts of the lakes describe it as having been undrinkable and darkly colored. Many perceived this as quite undesirable, but it may not have been indicative of it being toxic nor was it indicative of hypereutrophication. Rather, it was a function of much greater connectivity between the lake and its fringe wetlands, and a much higher tannin content.

Consider how the water entered the lakes during these times. All the creeks and rivers entering the lake system flowed through extensive wetlands, both wet meadows and emergent marshes. Martin Kerns of the Upper Wood RIver Valley described how Sun Creek came down from the uplands and dissipated into the wet meadow that is now his pastureland (personal communication). Much of the Wood River Valley behaved like a huge sponge soaking up water during the spring melt, pulling nutrients from the water, growing vegetation, forming peat in places, and slowly releasing water through the summer and fall. The lake fringe wetlands were essentially deltas for all the streams entering the lake system. The streams split into smaller channels, some still visible on the landscape, meandering their way across this flat land to the lake or disappearing into the marsh. Water emerged from the marsh margins into the lake with reduced nutrient loads, but still rich, and with substantial levels of tannins, humic substances or dissolved organic carbon (DOC), all very closely related terms.

The waters of the lake were still rich in nutrients; the lake was probably eutrophic even then. It supported large populations of several species of suckers, very large resident redband trout and migrations of salmon and steelhead moving through the system to spawn. Of course, the pyramid of life needed to support these populations of large fishes required many small fishes as well as invertebrates, etc.

The Winds and the Waters

The winds of pre-Nouveau American times were probably very similar to today’s winds. The Oregon Water Science Center has gathered a substantial amount of data on the winds affecting UKL and has made this available on their Wind Rose Grapher (http://or.water.usgs.gov/cgi-bin/grapher/graph_windrose_setup.pl) page. This was the source of the wind roses in Figures 3 & 4. By following the NWIS-Web link on each of the wind rose graph pages, you can obtain the larger set of “Current/Historical Observations” data (an example for Mid-North-Lake). The predominant winds blow from the west and/or northwest, depending upon the location on the lake; however, the winds reverse occasionally and blow in from the south and southeast (Figures 3 & 4).

Figure 3: USGS Wind Roses for North UKL Stations
(Click Figure to enlarge in a new tab)
Figure 4: USGS Wind Roses for South UKL Stations
(Click Figure to enlarge in a new tab)

These winds set up large circulating currents or gyres of water flowing with the winds in the shallows and returning against the winds in the deep trench along the west side of the central UKL (Figures 5 & 6; adapted from Wood, et al. 2009, Figures 8 & 9). This has been well documented by the US Geological Survey (Wood, et al. 2008; Wood, et al. 2006). Unfortunately, both these papers relied on data prior to the WRD levee breaching, so there is likely some perturbation of these circulation patterns, but the degree of that perturbation is not well understood at this time. Since most of the levees are still intact, the basics are probably still the same; however, since the levee breaches on the WRD were designed to facilitate flow through the delta, there may be some significant changes.

Figure 5: Gyres from Prevailing W-NW Winds
(Click Figure to enlarge in a new tab)
Figure 6: Current Gyres created by S-SE Winds
(Click Figure to enlarge in a new tab)
The Seiches or Wind Tides

The large circulation patterns in the lake, however, only tell half the story. Another phenomenon of wind passing over water is a seiche or wind tide. These occur when a wind of sufficient velocity passes over water for a sufficient time over a sufficiently long stretch of open water or fetch. Under these conditions, the water is physically moved in the direction of the water much like the moon and sun combine to create a tide. UKL is a sufficiently large lake that such conditions develop almost daily. In the morning the winds tend to be calm, building through the day until late afternoon or evening when they drop off. The result of this is a movement of water along the axis of the wind direction such that the surface elevation is different at different locations around the lake.

There are three primary surface elevation gauges deployed by the United States Geological Survey (USGS) that monitor lake levels consistently starting in the spring months through to the fall. These gauges are located:

View Larger Map

near the Harbor Isle development in Klamath Falls at the south end of the lake
(Lat 42° 15’00”, long 121° 48’55; #11507000)

View Larger Map

off Rattlesnake Point just north of Hank’s Marsh
(Lat 42° 20’55”, long 121° 49’35”; #11505900)

View Larger Map

at Rocky Point on Pelican Bay
(Lat 42° 28’40”, long 122° 05’12”; #11505800)

The surface elevation and wind data sets overlap considerably and have been used in several reports. For example, Wood, et al. (their Figure 6, 2008), in their report describing the wind gyres, also documented the fluctuations of lake surface level due to daily (diel) wind cycles and produced models to simulate the rises and falls of the water levels (ibid, Figure 14). The models were very good at predicting the cycle timing, and reasonably good at predicting the amplitude, but the amplitude tended to be somewhat overestimated.

We have taken these wind and lake surface elevation data sets comprising over 24,000 combined observations from between May 2009 and October 2012 and run them through a different sort of analysis. This analysis will be discussed in detail in “Appendix A: Seiches on Upper Klamath Lake,” but we introduce the result here. Various, reiterative multiple regression analyses were done seeking to understand the correlation between winds and differences in lake surface levels at Rattlesnake Point (RP) and near Klamath Falls (KF). For wind measurements, we used those from the USGS Mid-Lake gauge

View Larger Map

(Lat 42° 23’ 12”, long 121° 51’ 59.1”; #422312121515900)

We found that the best correlation was obtained by comparing the average wind speed over 60 minutes with the lake level measurements over the last 30 minutes of that period. During this 60 minutes, seven wind speed and direction measurements were taken, and over the last 30 minutes, two lake level measurements were recorded. The wind speed and direction measurements were treated as vectors with direction relative to a line between the two lake level gauges. Using simple trigonometry, the vectors of the wind parallel and perpendicular to the lake level station axis were resolved. The positive direction was designated from RP to KF because that is the prevailing wind direction, and the negative direction in the opposite direction from KF to RP. Similarly, when the KF lake surface elevation gauge read higher than the RP gauge, the difference was positive and when RP was higher than KF, the difference was negative.

Using those conventions, the data was plotted with the wind speeds on the horizontal axis and the lake surface level differences on the vertical axis (Figure 7). These data could be moderately well fit to polynomials from 3rd to 6th-order, but the best correlation was the 5th-order polynomial. With that said, the differences among the various order polynomial fits were very small (note the magenta line in Figure 7 is the 3rd-order polynomial) and all accounted for about 45% of the variance. With such a large dataset, the coefficients derived for the polynomials were very significant, all with p-values < 0.0001. It is also evident from the data that stronger winds produce a much greater effect, proportionally. For example a wind of 11 mph produces only an inch or two of seiche over the distance from RP to KF, while a wind of 22-27 mph will stack up water at one end of this axis almost 6 inches deeper than at the other, and note that this is just a portion of the “fetch” or length of the lake over which the wind is blowing. RP is 11 km from KF, while the distance from the WRD to KF, the north-south axis of UKL, is about 27 km and the distance from Rocky Point to RP, the east-west axis of the lake, is about 25 km. The distance from the mouth of Sevenmile Canal to KF, essentially the length of the combined UKL-AL lake system, is ca. 40 km. It is probable, therefore, that seiches over these longer fetches in the range of 3-4 in are very common and seiches in the range of 1-2 ft are occasional.

UKL Seiche-Wind Relationship
Figure 7: Seiche-Wind Relationship

Anecdotal evidence of a seiche in this latter range was reported under conditions of a strong south wind at the WRW. The Wood River was visibly flowing upstream with enough current to form eddies on the upstream side of the bridge entering the WRW, and the entire riparian zone was inundated upstream of the bridge. It was estimated that this seiche was in the 1-2 ft range (Rob Roninger, BLM, personal communication, 2013). This is reminiscent of why the Klamath Tribes referred to the Link River as Yulalona purported to mean “back and forth.” The phenomenon was last reported by the Spokane Spokesman-Review on July 23, 1918 when a seiche on July 18th of that year stopped the flow of the Link River leaving dry riverbed, and apparently, fish for those willing to bend over to pick them up. Figure 8 shows men and boys standing on the dry riverbed with the boys holding their catch of fish. It has been indicated by Native Americans (reference) that before the Link River Dam was built, this was at least a common lifetime experience for local residents.

Dry Bed of Link River
Figure 8: Men and boys standing on the wind blown dry bed of the Link River
Courtesy of the Klamath County Museum

The data in Figure 7 are also very interesting that there were much higher seiches produced under some circumstances, while at other times the winds seemed much less effective. This is, of course, consistent with the wind variable accounting for only ca. 45% of the variance. Multiple regression modeling of the other parameters available yielded little further benefit (data not shown). The wind vector perpendicular to the line between the two stations was most significant, but only more significant than the 4th-order term in the simple polynomial (as judged by larger t-ratios, data not shown). This perpendicular wind vector had a positive effect, so it may have simply indicated the general windy conditions of the time. It was also found that the variance of the wind speed and direction during the 60 min averaged period was significant, but again, it was a small effect. This modeling effort is continuing and this page will be updated if interesting results emerge.

Figure 9: Distribution of Mid-Lake Wind Speed Moving Averages
(Click Figure to enlarge in a new tab)

The distribution of the wind speed vectors along the RP to KF axis at the USGS Mid-Lake meteorological station is shown in Figure 9. The distribution is best modeled by three overlapping normal distributions corresponding to high and low winds in either direction and relatively calm conditions. The averages (means μ1, μ2 & μ3, Figure 9highlighted light blue) were 9 mph in the prevailing (positive) direction from the north, 1.3 mph north winds even under calm conditions, and -4.4 mph when winds were southerly. Interestingly, the dispersion measure (standard deviations σ1, σ2, & σ3), that is, the measure of how much the winds varied during southerly, calm and northerly winds, respectively, was largest (5.82 mph) for southerly winds. A practical effect of this was that even though the average southerly wind was weaker, the highest winds were approximately the same (60 min moving average of 20-22 mph) in both directions.

Figure 10: Distribution of Seiche Moving Averages between KF & RP
(Click Figure to enlarge in a new tab)

The distributions of seiches between Rattlesnake Point and Klamath Falls are shown in Figure 10. What was most interesting was that when the JMP statistical software was asked to find the best continuous fit for the seiche distribution, JMP determined the best fit for the seiches was also three overlapping normal distributions, just as was the case for the wind distributions. The averages and breadth of these distributions are shown in the upper right corner of Figure 10.

Selected wind and seiche quantiles are shown in the lower left of Figures 9 & 10. A quantile is the fraction of measurements that fall within a range of values. So in Figure 9, the top 20% of northerly winds, between the 100th and 80th quantiles, were between 11.3 and 22.7 mph, and the top 20% of southerly winds, between the 20th and 0th percentile, were between an actual 1.2 mph northerly wind and a 21.7 southerly wind, again an indication of the prevalencce of northerly winds at the Mid-Lake station. The quantiles chosen for presentation for the seiches were only the top and bottom 10% of seiches, because the distributions were rather narrow. Under northerly winds, the top 10% (100th-90th) ranged from a little over an inch to almost 11 inches. In front of southerly winds, the bottom 10% (10th-0th) ranged from approximately one to a bit over 6 inches. The most prominent observation from these data was that it was only the top and bottom 2 percentiles (100th-98th & 2nd-0th) that represented substantial seiches between approximately 2 and 10 inches to the north or south. However, if one considers that the seiche maxima only occur every day or so and that these analyses were done with moving averages to bring out more meaningful data as opposed to fleeting numbers, one can appreciate that 2% of any day is a half hour. As such, one might interpret the data we have at this point as telling us that 2 to 10 inch seiches are quite common during any given day.

Taken together, these two types of observations regarding water currents and seiches come into play when we consider the historical shoreline of UKL surrounded by wide zones of lake fringe wetlands. In Figure 11, it is clear from the wind rose graphs along with the current gyres of Figures 5 & 6 that in pre-Nouveau American times the wind served to sweep water along the shallows, and when the wetlands were not isolated from the lake ( historical wetlands; Figures 11), even small winds would push a few inches of water into the wetlands. There were historically 53,000 acres of lake fringe wetland surrounding Upper Klamath and Agency Lakes. If even a quarter of that was flooded by a 3 inch seiche on any given day due to wind speed and direction, that represents 3,300 acre-ft of water being passed in and out of the wetlands. Positing a stronger wind in a more favorable direction, let’s say 8 inches of water is pushed into half the lake fringe wetland: over 15,000 acre-ft of water would flow in and out of the wetlands.

With that water flowing into the wetlands would go larval suckers, sediment and dissolved nutrients, we hypothesize that significant amounts of all of these would be left behind when the winds died off and the water ebbed back into the lake. Given that higher winds would also suspend a great deal more bottom sediment in this shallow lake system, we believe these higher winds would deposit substantial quantities of sediment, in particular.

When the water returned to the lake, it would carry with it tannins or humic acids (tea-colored organic compounds derived from degrading plant tissues) characteristic of wetland waters. These compounds have been found to inhibit the growth of algae, and are the source of the dark colored waters of Upper Klamath Lake reported historically.

Historical Conclusions

In summary, before separation from its fringe wetlands, the lake would exhale into the wetlands with most wind events water containing sediments rich in phosphate, larval suckers during spring and early summer and nutrients, and then inhale tannins from the wetlands as the wind tide ebbed. It was a living breathing lake. We hypothesize that these currents and seiches were historically a very important mechanism by which the lake cleaned itself and by which drifting larval suckers were pushed into refugia. Separation of the lake system from its wide bands of fringe wetlands was akin to constricting its breathing and suffocating the lake system in P.

In addition, as mentioned at the top of this historical reconstruction, most of the water entering the lake system flowed through wetlands as it entered. This water likewise would have been separated to some extent from its sediments, larvae and nutrients. Isolation of the lakes from these primary mechanisms of purification has led to a profound change in the ecology of the lake system and its hypereutrophication.

Now the question is posed; how much of this ecosystem function can we restore? It clearly won’t happen overnight, but what might be done soon that would have substantial effects in the near future? What might be investigated to substantiate and quantify potential effects, such as those posited here around the seiches of these shallow lakes? Can these lake effects be realized in time to save the sucker populations directly dependent upon them and possibly contribute to the recovery of our Oregon Spotted Frogs and a revitalized Redband Trout fishery?

Table of Contents

  1. Upper Klamath Lake Watershed
  2. Historical Function of the Lake System
  3. Lake System Restoration Opportunities - these sections still in draft
    1. Treatment Wetlands
    2. Recirculating Wetlands and Alum addition
    3. Subsidence Reversal
    4. Reconnection of Large Band Lake Fringe Wetlands to the Lake System
    5. Economic Analysis and Public Education
  4. Walker Rim Groundwater Dependent Ecosystem