Scientific Investigations Report 2006–5209
U.S. GEOLOGICAL SURVEY
Scientific Investigations Report 2006–5209
Monitoring data collected during a 3-year study of sucker behavior in response to water quality provided an opportunity to describe water quality in the study area (approximately the northern one-third of Upper Klamath Lake) at a spatial and temporal resolution that was unavailable prior to 2002. Several questions were addressed using these data: What is the seasonal and interannual variability in water quality? Are there any aspects of seasonal variability that are predictable? If poor water quality is defined as conditions that would be potentially harmful to endangered suckers, where do these conditions tend to occur and how are they distributed throughout the day? What is the spatial extent of poor water quality conditions? Are there predictable differences in water quality between deep and shallow sites?
Each of the 3 years of the study was unique in terms of the seasonal patterns in water quality and the occurrence and severity of poor water-quality conditions, indicating that little predictability can be expected in these from year to year. Potentially harmful pH conditions, defined as greater than 9.7, occurred in late June and early July of all 3 years, coincident with a rapid expansion of the first bloom of the season. When averaged by Julian week, the spatial extent of these conditions was the greatest in the first or second week in July in all 3 years, reaching 0, 8, and 47 km2 (square kilometers) to less than 9.7 around mid-July in all 3 years, and only in 2003 did pH increase again to values greater than 9.7, coincident with a second bloom from the second week in August through the first week in September. Temperatures peaked in late July in all 3 years, but the spatial extent of potentially harmful temperatures, defined as greater than 28 degrees Celsius, was negligible in all 3 years. Dissolved oxygen was the most variable of the three continuously monitored water-quality constituents, as it is determined in large part by bloom dynamics, which also are highly variable from year to year. In each year, there were several periods that could be termed “low dissolved oxygen events” (LDOE) on the basis of the occurrence of concentrations less than 4 mg/L (milligrams per liter). Concentrations less than 4 mg/L in the lower water column for part of the day were common during the 3 study years, but did not result in a fish die-off. The most severe LDOE, which did culminate in a fish die-off, occurred in 2003. This LDOE was characterized by concentrations that were 3 mg/L or less throughout the water column and that persisted throughout the day, for 8 days, at a site located centrally in the study area and the area of the lake that was preferred adult sucker habitat. The maximum spatial extent of weekly averaged concentrations less than 4 mg/L occurred in the week starting on July 27, 2003, and was 39 km2. The next highest spatial extent of these potentially harmful dissolved oxygen conditions was 17 km2 and occurred in the week starting on August 15 in 2004.
There were differences between deep (defined roughly as deeper than 2.5 meters) and shallow sites in the daily timing of the most extreme values of temperature, pH, and dissolved oxygen, although the timing at both deep and shallow sites was consistent from year to year. At shallow sites, temperature and pH tended to peak in the late afternoon, primarily between the hours of 5 p.m. and 7 p.m. The daily minimum in dissolved oxygen tended to occur in the early morning hours, before the sun was high in the sky and when dark respiration had been operating the longest, between the hours of about 6 a.m. and 8 a.m. At deep sites, the timing of extreme values of temperature and pH was in the late afternoon, as at the shallow sites, but the timing of the extreme values of dissolved oxygen was less well defined than at the shallow sites. This is because the deeper sites often experience some degree of thermal stability during the day. Dissolved oxygen can continue to decrease in the lower water column beyond the early morning hours due to ongoing respiratory and other oxygen demanding processes in the water column and sediments, until at some point the water column mixes when the wind picks up or the water surface cools sufficiently. As a result, there was no well-defined time window in which the minimum daily dissolved oxygen concentration tended to occur at the deeper sites. Because low dissolved oxygen conditions tend to be more extreme at the deeper sites, and concentrations potentially harmful to fish are most likely to occur at the deeper sites, the assumption, sometimes made, that the worst conditions of the day for fish (in terms of dissolved oxygen) occur in the early morning hours probably is not justified in this lake.
Because hypoxia was determined to be an important trigger of recent, documented fish die-offs, a section of this report was devoted to discussing the factors contributing to an LDOE. Exceptional stratification in dissolved oxygen and low wind stress were not consistent features of the observed LDOEs, but the most severe LDOEs were coincident with a bloom decline.
Because the cause for bloom declines is unknown, the fundamental cause of a severe LDOE remains undetermined, but it was determined that severe LDOEs in the northern part of the lake for the most part do not develop there, but rather develop to the south and are transported into the area by the wind-driven circulation. The placement of acoustic Doppler current profilers in the lake in conjunction with the collection of wind data showed decisively the close coupling between the currents in the lake and the surface wind stress, and the fact that under prevailing wind conditions most of the volume of the lake moves in a clockwise loop that is completed in about 10 days. Water enters the northern part of the lake, where the preferred sucker habitat is found, almost entirely through a small cross-section between Eagle Ridge and Bare Island. Water coming through that cross section during the most severe LDOE, in July 2003, was characterized by exceptionally low water-column averaged dissolved oxygen, exceptionally low chlorophyll a, and exceptionally high temperature and ammonia.
The concentration of dissolved oxygen observed at a fixed site on the lake is a complicated function of the wind-driven circulation. Reversals or stalls in the wind increase the residence time in the deep trench along the western shoreline, thus allowing more time for oxygen demanding process to deplete the water passing through of oxygen. These stalls or reversals increase residence time along the northern and eastern broad shallow areas of the lake as well, thus giving oxygen producing photosynthesis more time to increase dissolved oxygen concentration in those areas. Short-term pulses of water low in dissolved oxygen coming through the trench were linked directly to brief wind reversals. Long-term trends in water quality, on a time scale greater than about 10 days, integrate the net effect of all oxygen demanding and oxygen producing processes along the flow path. It was such a trend that culminated in the severe July 2003 LDOE.
An attempt was made to place the 3 study years into a more historical context and, in particular, to the characteristics that distinguish the three most severe recent fish die-off years (1996, 1997, and 2003). A 15-year record of meteorology (air temperature and wind speed collected at the Klamath Falls airport, obtained from the National Climatic Data Center) and water quality (biweekly water-quality profiles collected at three sites in the northern part of the lake by the Klamath Tribes) were used to provide the historical context. Air temperature was used as a surrogate for water temperature, as no long-term continuous records of water temperature exist. Two of the three fish die-off years, 1996 and 2003, ranked at the high end of the range in July temperature, and 1996 and 1997 ranked at the high end of the range in August temperature, on the basis of an analysis of variance test (ANOVA). Thus, high temperatures may contribute to the conditions leading to a fish die-off. The same test ranked all 3 fish die-off years at the low end of the range in July wind speeds. An attempt to correlate wind stress with the degree of dissolved oxygen stratification or with dissolved oxygen concentration using the 15 years of climate data yielded insignificant relations, so the contribution of lower-than-average wind stress to a severe LDOE or a fish die-off probably is not through stratification, but rather something else, possibly the wind-driven circulation.
Attempts were made to rank July–August median water-quality variables in the biweekly dataset collected by the Klamath Tribes with an ANOVA. Most water-quality variables had short term and spatial variability comparable to interannual variability, making it difficult to statistically distinguish July–August distributions between years. The exception was ammonia concentration, which showed year-to-year trends in the July–August distributions suggestive of a climatic signal. The ammonia concentration was highly correlated with two climate variables—mean July–August wind speed, and October to May cumulative discharge in the Williamson River. Because both climate variables were also correlated with May–June ammonia concentration, the correlation with Williamson River discharge is more likely to represent a causal relation, such as high spring runoff bringing in more readily decomposed material, resulting in a higher water column ammonia concentration in late spring and early summer. It is unclear, however, that there is any connection between the climate signal in ammonia concentration and the occurrence of fish die-offs. One of the three fish die-off years, 1997, was characterized by some of the highest ammonia concentrations on record, one year, 1996, was characterized by more moderate, but still high (relative to the entire 15-year record) concentrations, and one year, 2003, was characterized by some of the lowest concentrations in the 15-year record.
Even though fish die-off years were not all years of exceptionally high ammonia concentration, high ammonia concentration (in the fourth quartile of the entire historical distribution of July–August values) at a single index site located in the study area was one of three factors that, in combination, could be used to successfully screen the historical water-quality dataset for those dates which most closely preceded the fish die-offs. The other two factors were dissolved oxygen and chlorophyll a concentration, both in the first quartile. Dates preceding the fish die-offs in all 3 years were passed by the screening, and dates from only three other non die-off years were passed by this screening. Therefore, the idea that one site could act as an “indicator” site was deemed a success, which might be information that could be used in the design of a long-term monitoring program for the lake.
U.S.
Department of the Interior | U.S. Geological
Survey
Persistent URL: https://pubs.water.usgs.gov/sir20065209
Page Contact Information: Publications Team
Page Last Modified: Thursday, 01-Dec-2016 19:13:04 EST
