Effects of Middle Creek Marsh Restoration on

Clear Lake Water Quality

E. E. Van Nieuwenhuyse

Jones & Stokes Associates

Abstract - Water quality in Clear Lake is seriously impaired by high phytoplankton production. High phytoplankton production is primarily due to excessive phosphorus loading from diffuse sources throughout the Clear Lake catchment. The bulk (>90%) of this phosphorus is transported to the lake attached to silt and clay-sized sediment. The Middle Creek sub-catchment (i.e., the area drained by Scotts, Middle and Clover creeks) presently accounts for some 70% of the total phosphorus load to Clear Lake's Upper Arm. A previous study suggested that the restored marsh might retain up to 70% of this Middle Creek phosphorus load. By contrast, silt and clay retention coefficients calculated from HEC-6 modeling runs performed by the U.S. Army Corps of Engineers for this study suggested a phosphorus retention coefficient (RP) of 15%. A discussion of the assumptions used to calculate these two widely differing estimates of RP suggested that the actual retention coefficient of the restored marsh would more likely average about 40%. Calculations based on a phosphorus balance model for Upper Arm indicated that a 40% RP for the marsh would reduce the average total phosphorus concentration (TP) of the lake's Upper Arm from its historical (1969-1991) average of 170 mg m-3 to 120 mg m-3 . A global scale regression model for shallow (mean depth <5 m) temperate-latitude lakes (n=221) indicated that this roughly 30% reduction in TP would cause average chlorophyll concentration (a measure of phytoplankton production) in Upper Arm to decline from its historical (1969-1991) average of 92 mg m-3 (calculated from estimates of mean biovolume and average chlorophyll content) to a new average value of 70 mg m-3. An appreciable reduction in the frequency and severity of algal blooms would accompany this reduction in average chlorophyll concentration. These results suggest that marsh restoration would lead to noticeable improvements in Clear Lake water quality, but that the goal of halving phosphorus loading to the lake will probably require additional conservation efforts in the Clear Lake watershed.


Phosphorus loading from diffuse sources throughout the Clear Lake watershed during the wet season (November-April) is thought to be largely responsible for the excessively high levels of phytoplankton biomass that too often characterize the lake during the dry season (Richerson et al. 1994). The "algal blooms" that accompany this excessively high level of algal production seriously impair water quality by forming unsightly surface scums, fouling beaches, creating disagreeable odors, and contributing to low dissolved oxygen concentrations near the lake bed. Algal blooms may also promote processes that lead to bioaccumulation of mercury in the lake's invertebrate, fish and waterfowl communities (Suchanek et al. 1997).

A Clean Lakes Diagnostic/Feasibility Study for Clear Lake (Richerson et al. 1994) set a phosphorus reduction goal for Clear Lake that would require halving the existing loading rate of phosphorus from the watershed. Restoration of some 490 ha of marshland near the mouth of the 33,400-ha Middle Creek catchment has been proposed as one way to help accomplish this target (Richerson et al. 1994). This strategy assumes that reduced phosphorus loading will lead to reduced phytoplankton abundance, fewer and less intense algal blooms and thus better water quality.

The goal of this study was to quantitatively evaluate these assumptions by determining: (1) how much of the total phosphorus load currently entering the lake originates from the Middle Creek catchment; (2) how effectively the marsh is likely to function as a net sink for Middle Creek phosphorus loads; (3) how in-lake phosphorus concentration is likely to respond to a reduction in phosphorus loading from Middle Creek; and (4) how Clear Lake phytoplankton production is likely to respond to reduced phosphorus concentration. Only the Upper Arm was modeled because it accounts for 72% of lake surface area and 63% of its total volume and because much of the nuisance algae conditions in the lower arms are caused by algae produced in the Upper Arm and subsequently transported to the lower arms by wind or water currents. This approach simplified analytical methods and was considered adequate for the purposes of this study.

Methods and Assumptions

Phosphorus Load Reduction. Estimating the phosphorus load reduction potential of the restored marsh required first an estimate how much of the total phosphorus load to the Upper Arm is contributed by the Middle Creek watershed under existing conditions. This was done using historical hydrologic data for Upper Arm streams summarized in the Diagnostic Report (Richerson et al. 1994) and flow-weighted mean total phosphorus concentration calculated from samples collected by Lake County between 1993 and 1996 in the Middle Creek drainage (i.e., Middle, Scotts and Clover creeks) and other streams draining into Upper Arm. The historical mean annual discharge for each group of streams was multiplied by the flow-weighted mean phosphorus concentration to estimate contribution to total mean annual phosphorus loading to Upper Arm.

The next step was to quantify an independent estimate of phosphorus retention coefficient (Rp) for the restored marsh. This estimate was calculated using output from a HEC-6 sediment transport model (Jones & Stokes Associates, Inc. 1997). The sediment transport model provided estimates of clay and silt transport at a number of transects above within and downstream of the proposed marsh area under existing conditions and under fully restored (100-year floodplain) conditions. Sediment transport under three events were compared: one annual event after one year, ten annual events after 10 years and the 100-yr event after 10 years. For each event run, the amount of each sediment size class transported past the most downstream transect (Lucern cutoff bridge) under the fully restored condition was subtracted from the amount transported past this transect under existing conditions. The amount of silt or clay retained was multiplied by literature value estimates of average phosphorus content of each size class; specifically, 1.8 mgP g-1 for clay and 0.75 mgP g-1 for silt (Syers et al. 1969) to estimate phosphorus retention. This coefficient was multiplied by Middle Creek's contribution to total phosphorus loading to Upper Arm to estimate loading reduction resulting from marsh restoration.

Prediction of In-lake Total Phosphorus Concentration. The phosphorus mass-balance model of Vollenweider (1969) was used to predict average phosphorus concentration in Upper Arm from predicted external phosphorus loading rates. This model is:

TP = Lp [z( +)]-1 (Equation 1)

where TP=mean total phosphorus concentration in the water column of Upper Arm (mg m-3), Lp=mean areal phosphorus loading to Upper Arm (mg m-2 yr-1), z=mean depth of Upper Arm (m),

=phosphorus sedimentation coefficient (yr-1) and =mean hydraulic flushing rate (yr-1). Mean depth, lake surface area and flushing rate were estimated from data presented in Table 3.1 of the Diagnostic Report (Richerson et al. 1994). The sedimentation coefficient was estimated from long term mean TP, annual phosphorus loading rate, mean depth and flushing rate by rearranging Equation 1 and solving for . This method indicated that sedimentation coefficient in Upper Arm averaged about 0.75 yr-1. This value was assumed constant for purposes of predicting TP under different estimates of retention coefficient for the restored marsh.

Prediction of Phytoplankton Abundance. Average seasonal (May-October) abundance of phytoplankton in Upper Arm was predicted from TP using the global scale model of

Van Nieuwenhuyse (1993):

Log Chl = -2.53 + 3.29 Log TP - 0.60 (Log TP)2 (Equation 2)

where Chl = mean surface chlorophyll concentration. Chlorophyll concentration under existing conditions (1969-1991) was calculated from mean summer biovolume and average chlorophyll content data for blue green algae and non-blue green algae. Chlorophyll content values for each functional group were based on mean values tabulated in Reynolds (1984). Equation 2 was used because it was developed specifically for shallow temperate-latitude lakes.

Results and Discussion

Phosphorus Load Reduction. Hydrologic data for streams draining into Upper Arm indicated that project area streams (Clover, Middle and Scotts creeks) contributed on average about 57% of total inflow into Upper Arm (Table 1). Water chemistry data collected between 1992 and 1996 by Lake County indicated that flow-weighted mean total phosphorus concentration was substantially higher in project area streams (644 mg m-3) than in other streams draining into Upper Arm (342 mg m-3). These data indicated that the total P load to Upper Arm averages about 169 metric tons per year. The contribution of Middle Creek to this total phosphorus load was 71%; considerably higher than its contribution to hydrologic loading. This difference indicates that erosion is more intense among project area streams than among the other streams draining into the Upper Arm of Clear Lake.

Calculations based on the HEC-6 sediment transport model indicated that marsh restoration would result in the retention of about 40% of incoming silt load and 4% of incoming clay load for the annual flood. The 100-yr flood event analysis indicated similar silt retention, but considerably less clay retention (Table 2). Because most of the loading to lakes is accomplished by "average" events over the long run (Leopold et al. 1964), the 4% and 40% values were used for purposes of predicting phosphorus loading to Upper Arm.

The retention of silt was greater than for clay-sized particles because clay-sized particles are smaller and lighter and thus settle out less rapidly than silt-sized particles. Clay-sized particles, however, have a higher average phosphorus content in part because of their greater surface area to volume ratio. Taking this differential phosphorus content into account indicated that the phosphorus retention coefficient (Rp) for the marsh would average about 15% (Table 2).

The 15% retention coefficient estimated from the sediment transport analysis is considerably lower than the 70% Rp suggested by storm flow samples collected up and downstream of Tule Lake during the Diagnostic Study (Richerson et al. 1994). The 15% value is probably an underestimate given that the sediment transport model did not include the effects of channel complexity and vegetation in the restored marsh. Both factors would greatly increase resistance to flow and thus hydraulic residence time during high flow events. Submerged vegetation and leaf litter also tend to accumulate clay particles on their surfaces. These mechanisms would allow a greater than predicted amount of silt- and clay-sized particles to be retained in the marsh, thus reducing the amount of particulate inorganic P reaching the lake.

By contrast, the 70% estimate of Rp based on storm flow samples collected during a single event probably overestimates the true value. The accuracy of the 70% value, even as a measure of the short term Rp for Tule Lake marsh, is questionable because accurate estimates of flow downstream from the marsh were not available (Richerson et al. 1994). Moreover, even if retention during a single event was 70%, the long term annual mean retention coefficient would almost certainly be substantially lower. This hypothesis is consistent with results presented in Table 5.15 of the Diagnostic Study report indicating that the average annual P retention coefficient for the entire Clear Lake system is 77% (range, 53-96%). It seems unlikely that a single marsh complex could retain the same percentage of incoming phosphorus as the entire Clear Lake system. Consequently, the 15% and 70% values were used in this study as upper and lower limits and a 40% value was selected as a plausible long term average value for subsequent analyses.

Prediction of Total Phosphorus Concentration. The long term average external load estimate of 169 metric tons per year (Table 1) indicated that the areal P loading rate to Upper Arm (i.e., loading per unit surface area of Upper Arm) averaged 1333 mg m-2 yr-1. This areal loading rate decreased as the phosphorus retention coefficient assumed to characterize the restored marsh increased (Table 3). As areal loading decreased from its present value of 1333

mg m-2 yr-1 to the 671 mg m-2 yr-1 value expected with 70% retention, the average phosphorus concentration predicted by Equation 1 for Upper Arm declined from its historical annual mean value of 170 mg m-3 to 85 mg m-3 (Table 3). The expected concentration assuming a 40% long term average value of Rp was 120 mg m-3.

Prediction of Phytoplankton Abundance. Routine monitoring data from the California Department of Water Resources for the period 1969-1991 indicated that seasonal (May-October) mean biovolume of blue green and non-blue green algae in Upper Lake was highly variable. The percentage of the total biovolume attributed to blue green algae also ranged widely; from 4% to 98% of total algal biovolume. Blue green biomass has a lower chlorophyll content than green algae or other forms of algal biomass. Taking this differential chlorophyll content into account resulted in estimates of seasonal mean chlorophyll concentration that ranged between 13 and 356 mg m-3, averaging some 92 mg m-3 over the entire 22 years of available records (Table 4). These estimates provided a basis for comparison with other relatively shallow temperate latitude lakes elsewhere in the world.

Observed values of TP and estimated Chl for the Upper Arm of Clear Lake were compared to the global TP-Chl relationship for shallow lakes (Equation 2). The Clear Lake values were generally scattered along either side of the ascending limb of the model curve. This comparison indicated that Upper Arm phytoplankton abundance is probably still phosphorus limited (i.e., TP levels in Upper Arm are within a range in which Chl would be likely to respond to changes in TP). The mean and 95% confidence interval (illustrated by the crossed bars) for chlorophyll concentration under existing conditions (Figure 2) agreed well with the global model prediction. Thus, this model could be used to predict what chlorophyll concentration would be under the three retention coefficients estimated for the restored marsh.

According to the global model, the modest reduction in TP for Upper Lake expected with a 15% retention coefficient for Middle Creek marsh would result in a similarly modest reduction in phytoplankton abundance (Figure 2). By contrast, with a 70% retention coefficient for the restored marsh, the global model indicated that chlorophyll would decline from its historical average of 92 mg m-3 to about 51 mg m-3. If the marsh retains 40% of incoming P loading, the decrease in chlorophyll concentration would be less dramatic (from 92 to 70 mg m-3), but still highly significant (Figure 2). A number of studies have shown a strong correlation between average chlorophyll concentration and the frequency and severity of algal blooms in lakes (Rast and Lee 1978; Walker 1986). Thus, this analysis of conditions in Clear Lake indicated that even under the most plausible value of 40% retention, a substantial reduction in the frequency and severity of algal blooms in Upper Arm could be expected. This reduction would alleviate water quality problems associated with blooms in Upper Arm and reduce the amount of biomass blown in or otherwise transported from Upper Arm to Lower and Oaks arms.

Recommendations for Future Studies

The model used to predict phosphorus concentration in Upper Arm is quite sensitive to variation in the proportion of external loading contributed by project area streams and by the assumptions made about the phosphorus retention coefficient of the restored marsh.

Neither estimate is very well quantified using existing data.

To improve P loading estimates to the lake, it is recommended that all major inlet streams to Upper Arm be gaged and equipped with remote water sampling devices programmed to collect samples during at least three stages along the flood hydrograph. Experience with other lakes indicates that it is only with this kind of monitoring that reasonably accurate estimates of P loading can be achieved; especially in systems with highly seasonal and flashy hydrologic regimes, such as Clear Lake's inlet streams.

Two recommendations are offered for improving estimates of the marsh's retention coefficient. The HEC-6 model should be refined to include the effects of channel modifications, baffling, vegetation and other site-specific properties of the project area. These refinements should provide a more realistic estimate of retention coefficient. Output from this improved HEC-6 model should be compared with empirical estimates of retention coefficient for the Tule Lake marsh system. This system should be monitored up and downstream during floods and during the draw down season. For this purpose, it would be necessary to reactivate the DWR gage downstream of Tule Lake or to install a new temporary gage. Either gage would have to be rated against ultrasonic velocity meter (UVM) estimates of outflow or against predicted outflow values based on measured inflow, lake level and other factors using a numerical hydrodynamic model. This focused monitoring effort could be incorporated into the more routine monitoring effort required to accurately quantify phosphorus loading to Upper Arm.

Another source of error in this preliminary analysis is the estimate of chlorophyll concentration derived from biovolume estimates. Biovolume estimates are very time consuming and expensive to obtain, whereas chlorophyll concentration is relatively cheap and easy to measure. It is recommended that chlorophyll analysis be added to the routine monitoring program for Clear Lake or that a short term study be conducted to quantify a chlorophyll-biovolume relationship specific to Clear Lake rather than relying, as in this analysis, on point estimates extracted from the literature.

Finally, even though Middle Creek may account for some 70% of phosphorus loading to Upper Arm under existing conditions, restoration of the marsh may retain only 40% of this loading. Consequently, the total loading to Upper Arm would be reduced by 28%; considerably less than the 50% reduction goal defined in the Diagnostic Report. Achieving this goal will require a combination of improved watershed management and perhaps additional marsh restoration elsewhere in the Middle Creek drainage or in other inlet stream systems.


Thanks to Professor Pete Richerson and Mr. Jesse Becker, UC-Davis, for providing a digital copy of the DWR routine monitoring data set for Clear Lake and to Mr. Tom Smythe, Lake County Flood Control and Water Conservation District for providing hydrologic and climate data. This study was funded by the Sacramento District Office of the U.S. Army Corps of Engineers as part of its Middle Creek Ecosystem Restoration Reconnaissance Study, Mr. Rick Dreher, Project Manager.

References Cited

Jones & Stokes Associates, Inc. 1997. Middle Creek ecosystem restoration reconnaissance study. Final. May. (JSA 96-239). Sacramento, CA. Prepared for U.S. Army Corps of Engineers, Sacramento, CA.

Leopold, L. B., M. G. Wolman, and J. P. Miller. 1964. Fluvial processes in geomorphology. W. H. Freeman and Co., San Francisco. 522 p.

Rast, W. and G. F. Lee. 1978. Summary analysis of the North American (U.S. portion), OECD Eutrophication Project: Nutrient loading lake response relationships and trophic state indices. U.S. Environmental Protection Agency. EPA-600/3-78-008.

Reynolds, C. S. 1984. The ecology of freshwater phytoplankton. Cambridge University Press. Cambridge. 384 p.

Richerson, P., T. Suchanek, and S. Why. 1994. The cause and control of algal blooms in Clear Lake: Clean Lake Diagnostic/Feasibility Study for Clear Lake, California. Prepared for Lake County Flood Control and Water Conservation District, California State Water Resources Control Board, and the U.S. Environmental Protection Agency. Lake County, CA.

Suchanek, T. H., P. J. Richerson, L. J. Mullen, L. L. Brister, J. C. Becker, A. Maxson, and D. G. Slotton 1997. The role of the Sulphur Bank Mercury Mine site(and associated hydrogeological processes) in the dynamics of mercury transport and bioaccumulation within the Clear Lake aquatic ecosystem. A report prepared for the USEPA, Region IX Superfund Program. 479 p.

Syers, J. K., R. Shah, and T. W. Walker. 1969. Fractionation of phosphorus in tow alluvial soils and particle-size separates. Soil Science 108(4): 283-289.

Van Nieuwenhuyse, E. E. 1993. The phosphorus-chlorophyll relation in streams and some comparisons with lakes. Ph.D. dissertation, University of Missouri, Columbia, MO.

254 p.

Vollenweider, R. A. 1969. Possibilities and limits of elementary models concerning the budget of substances in lakes (in German). Arch. Hydrobiol. 66: 1-36.

Walker, W. W., Jr. 1986. "Empirical methods for predicting eutrophication in impoundments; Report 4, Phase III: Applications manual," Technical Report E-81-9, prepared by W. W. Walker, Jr., Environmental Engineer, Concord, MA, for the U. S. Army Corps of Engineers Waterways Experimental Station, Vicksburg, MS.

Table 1. Estimated phosphorus loads to Upper Arm of Clear Lake from Middle Creek catchment
and from other catchments draining into Upper Arm.
Contributing Area Catchment Area Mean Annual Discharge1 % of Total Inflow to Upper Arm Total P Concentration2 Annual P Load % of Total Load
(km2) (cfs) (%) (mg m-3) (MT yr-1) (%)
Middle Creek 334 210 57 644 121 71
Others 900 159 43 342 49 29
Total 1234 369 100 --- 169 100
1 Tom Smythe, personal communication
2 Flow-weighted mean

Table 2. Estimate of phosphorus retention coefficient of restored marsh based on output from sediment transport
model (HEC-6).1
Status Model Scenario Simulated Flow Simulated Clay Load2 Simulated Silt Load2 % Clay Retained3 % Silt Retained3 Phosphorus Retained4
(cfs) (MT day-1) (MT day-1) (%) (%) (%)
Without marsh One annual event 4660 927 815 --- --- ---
With marsh 4660 888 488 4.2 40.1 14.8
Without marsh Ten annual events 4660 927 817 --- --- ---
With marsh 4660 889 492 4.1 39.8 14.6
Without marsh One 100-yr event 19180 11653 16300 --- --- ---
With marsh 19180 11480 9848 1.5 39.6 12.7
1 Source: Don Twiss, U.S. Army Corps of Engineers, Sacramento, CA
2 Load at downstream end of Rodman Slough (mouth of Middle Creek drainage)
3 Difference between load under existing conditions and under restored marsh conditions divided by load under existing conditions
4 Assumes phosphorus content of 1.8 mg per gram of clay and 0.75 mg per gram of silt (Syers et al. 1969)

Table 3. Predicted mean total phosphorus concentration in Upper Arm of Clear Lake
for 15%, 40%, and 70% phosphorus retention coefficient values for the restored Middle Creek marsh
Phosphorus Retention Coefficient for Restored Marsh Predicted Total P Loading Rate to Upper Arm Predicted Total P Concentration in Upper Arm1 Percentage Reduction from Historical Mean (170 mg m-3)
(%) (mg m-2 yr-1) (mg m-3) (%)
0 1333 170 0
15 1191 150 12
40 955 120 30
70 671 85 50
1 from Equation 1

Table 4. Seasonal (May-October) mean biovolume and estimated chlorophyll concentration
associated with blue green and non-blue green algae in Upper Arm, 1969-19911.
Year Total Biovolume %Blue green Blue green biovolume Non-blue green biovolume Blue green chlorophyll Non-blue green chlorophyll Total chlorophyll
(mm3 L-1) (%) (mm3 L-1) (mm3 L-1) (mg m-3) (mg m-3) (mg m-3)
1969 5.26 73.3 3.86 1.41 21.5 15.4 37
1970 9.97 79.2 7.90 2.07 44.1 22.8 67
1971 8.57 49 4.20 4.37 23.4 47.9 71
1972 15.11 98.4 14.87 0.24 83.0 2.7 86
1973 8.16 67.1 5.48 2.69 30.6 29.5 60
1974 6.64 97.1 6.44 0.19 36.0 2.1 38
1975 7.13 92 6.56 0.57 36.6 6.3 43
1976 15.75 96.4 15.18 0.57 84.7 6.2 91
1977 12.20 98.3 11.99 0.21 66.9 2.3 69
1978 5.44 81.7 4.44 1.00 24.8 10.9 36
1979 7.19 97.5 7.01 0.18 39.1 2.0 41
1981 17.07 46.4 7.92 9.15 44.2 100.3 145
1982 7.70 90.7 6.99 0.72 39.0 7.9 47
1983 4.33 36.9 1.60 2.73 8.9 30.0 39
1984 40.51 83.3 33.75 6.77 188.3 74.2 263
1985 21.45 97.9 21.00 0.45 117.2 4.9 122
1986 3.92 56.8 2.23 1.69 12.4 18.6 31
1987 21.47 20.7 4.44 17.02 24.8 186.7 212
1988 1.19 3.9 0.05 1.15 0.3 12.6 13
1989 8.89 15.2 1.35 7.54 7.5 82.7 90
1990 38.53 32.2 12.41 26.13 69.2 286.6 356
1991 8.56 33.5 2.87 5.69 16.0 62.4 78
1 Biovolume data are from the California Department of Water Resources routine monitoring program
(data provided by Pete Richerson, UC-Davis). Chlorophyll content is assumed to average 5.58 ug mm-3
for blue green algae and 10.97 ug mm-3 for non-blue greens (Table 7 in Reynolds 1984).
a No data