An Analytical Framework

A modeling study provides guidance for developing nutrient TMDLs and a comprehensive watershed management plan.

The California Regional Water Quality Control Board–Santa Ana Region (RWQCB) has identified Lake Elsinore and Canyon Lake on its Clean Water Act Section 303(d) list as impaired water bodies. The causes of impairment are excessive levels of nutrients in both lakes; high bacteria levels in Canyon Lake; and low dissolved oxygen (DO), excessive sedimentation, and unknown sources of toxicity in Lake Elsinore. Nutrients from the San Jacinto River watershed delivered to these lakes cause significant algae growth, resulting in unpleasant odors, adverse effects on aesthetics, and impaired recreational use. Moreover, excessive algae growth causes depletion of DO in Lake Elsinore and results in occasional massive fish kills.

The RWQCB, in cooperation with various stakeholders in the watershed, has developed nutrient total maximum daily loads (TMDLs) for Canyon Lake and Lake Elsinore. To support this initiative, the Santa Ana Watershed Project Authority (SAWPA) coordinated the Lake Elsinore and Canyon Lake Nutrient Source Assessment (hereafter referred to as Nutrient Source Assessment) (SAWPA 2003) in cooperation with the RWQCB and the Lake Elsinore and San Jacinto Watersheds Authority (LESJWA). Results of the Nutrient Source Assessment, the TMDL study, and other efforts have provided a sound basis and a good opportunity to develop the San Jacinto Nutrient Management Plan (hereafter referred to as Nutrient Management Plan) (SAWPA 2004). The Nutrient Management Plan provides guidance regarding the control of nutrients and implementation of best management practices (BMPs) in the watershed to assist in the restoration of the lakes to meet beneficial uses. This plan uses the results of the Nutrient Source Assessment and identifies contaminant loadings from various sources throughout the watershed and recommends multiple projects to improve conditions and provide further study for guidance in future management decisions.Watershed Background
The San Jacinto River watershed (US Geological Survey–HUC 18070202), which covers approximately 770 square miles, is located almost 60 miles southeast of Los Angeles. It extends from the San Jacinto Mountains in the north and east to Lake Elsinore in the west (Figure 1). Most of the watershed (99.75%) falls within Riverside County with only a small portion (0.25%) extending into Orange County.

The watershed is essentially a desert region that is considered to have a Mediterranean climate. The average annual rainfall in the watershed is approximately 15 inches (RWQCB 1995). Three types of storms dominate the region: general winter storms, general summer storms, and high-intensity thunderstorms. Winter storms typically last for several days and occur in the wet period that extends from November through May. Thunderstorms can occur at any time of the year, but are most common between July and September. These storms are characterized by short periods of high-intensity rainfall. General summer storms, which normally occur from July through September, are rare events. When these storms do occur, they can result in heavy rainfalls over the course of several days [Riverside County Flood Control and Water Conservation District (RCFCWCD) Hydrology Manual]. With increased urbanization in a watershed with a Mediterranean climate such as the San Jacinto, the magnitude of peak discharges and annual runoff volumes increases while the year-to-year runoff volume variability decreases (Beighley, Melack, and Dunne 2003).

The watershed is a dynamic system with various unique conditions that either enhance or restrict flows through the watershed. The San Jacinto River, Salt Creek, Perris Valley Storm Drain, Mystic Lake, Perris Reservoir, Canyon Lake, and Lake Elsinore are the dominant hydrologic features in the watershed (Figure 2). In many cases, lakes, reservoirs, and other detention facilities impound streamflow. These impoundments can have major impacts on the quantity and quality of the water transported throughout the watershed. Storage of water results in not only the attenuation of peak flows but also increased soil infiltration and other associated losses.

To assess the land use of the San Jacinto watershed, the US Geological Survey (USGS) Multi-Resolution Land Characteristics 1993 data were combined with supplemental data collected by the Eastern Municipal Water District, providing a more detailed coverage of the land use. Land use in the watershed is predominantly agricultural and residential in the valleys and open in the headwaters. Overall, 73.8% of the watershed is open, 18.2% is agriculture, and 7.95% is urban (Table 1).

There are no traditional point-source discharges in the watershed, with the exception of discharges from urban stormwater outfalls and the overflow of processed wastewater from dairy and animal feeding operations during acceptable conditions (rainfall event defined in Title 27, Chapter 7, Subchapter 2, Article 1, Section 22562(a), California Code of Regulations and 40 CFR Part 412). Major nonpoint-source contributors in the watershed include agricultural lands, dairies, feedlots, grazing, land development, and urban runoff.

Model Selection and Development
In support of the Nutrient Source Assessment, a modeling system was developed of the San Jacinto River watershed and Canyon Lake. The modeling system can be divided into two components representative of the processes essential for accurately modeling nutrient loading and internal mass balances of the lakes. The first component consisted of a watershed model that predicted stormwater runoff and transport of nutrients as a result of rainfall events (and direct, non-storm loadings to water bodies). The second component included a lake model to predict the response and mass balance of nutrients within the water column of Canyon Lake and resulting overflow and loading of nutrients to Lake Elsinore.

Modeling the San Jacinto River watershed and Canyon Lake presented a challenge using currently available modeling tools. The system involves various unique features including hydraulic issues in the San Jacinto River (e.g, agricultural diversion channels), impacts of agricultural BMPs, hydrology sinks (Mystic Lake), and an arid climate and significant groundwater pumping that results in essentially no baseflow at various locations throughout the San Jacinto River and its tributaries during normal conditions. In addition to TMDL development, the model was utilized to support development of a watershed management plan through testing of alternative scenarios resulting from various management and environmental factors. Such scenarios resulted from the augmentation of input data collected in ensuing monitoring efforts and analysis of various management strategies or BMPs. Therefore, model flexibility was a key attribute for model selection.

The Loading Simulation Program in C++ (LSPC) (USEPA 2003a) was used to simulate watershed processes, including hydrology and pollutant accumulation and washoff. LSPC is a component of the TMDL Modeling Toolbox (USEPA 2003b, 2004), which has been developed through a joint effort between the Environmental Protection Agency (USEPA) and Tetra Tech Inc. It integrates a geographical information system, comprehensive data storage and management capabilities, a dynamic watershed model (a re-coded version of the USEPA’s Hydrological Simulation Program–FORTRAN [HSPF]) (Bicknell et al. 1996), and a data analysis/post-processing system into a convenient PC-based windows interface that dictates no software requirements. Similar watershed models have been applied successfully for the headwaters of the San Jacinto River (Guay 2002) and in other southern California watersheds (Ackerman et al. 2001; Watson et al. 2003). The flexibility of LSPC resulted in a powerful tool for linking model output to a separate receiving water model of Canyon Lake, modifying and tailoring model code to address specific issues in the San Jacinto River watershed (e.g., losses in Mystic Lake), and analyzing model output for assessment of spatial and temporal variability of nutrient loads from alternative land uses.

For simulation of Canyon Lake and prediction of nutrient loads to Lake Elsinore as a function of overflows of Canyon Lake dam, a simplified application of the Environmental Fluid Dynamics Code (EFDC) was used. EFDC is a comprehensive three-dimensional model capable of simulating hydrodynamics, salinity, temperature, suspended sediment, water quality, and the fate of toxic materials (Hamrick 1992, 1996). Since the primary purpose of the Canyon Lake model was to provide an estimate of nutrient loads to Lake Elsinore, it was not necessary to develop a fully configured three-dimensional hydrodynamic and eutrophication simulation model in the present stage of the modeling study. Instead, a simplified, two-dimensional hydrodynamic and nutrient transport model was developed. This model includes a hydrodynamic sub-model and a coupled nutrient fate and transport sub-model. The hydrodynamic sub-model is capable of simulating Canyon Lake water circulation in a depth-integrated fashion based on water budget and momentum balance. The nutrient fate and transport sub-model was developed based on a simplification of kinetics for total phosphorus and total nitrogen. The major sources of the two constituents are the nonpoint-source load from the watershed model and the benthic flux from the sediment. Important sinks for the two constituents are characterized in the model by a first order decay process, as well as the overflow and seepage through the dam. The EFDC computer code was modified to enable use of total phosphorus and total nitrogen as surrogates for ortho-phosphate and ammonia during operation.

The modeling system was calibrated and validated with instream-flow and water-quality data collected at various instream stations throughout the watershed from 1991 to 2001 and 2003, as well as stage data and water-quality data collected from four Canyon Lake stations from 1997 to 2000 and 2003. However, few water-quality data were available for a significant wet-weather event that resulted in the fill and overflow of Mystic Lake and subsequent transport of nutrients from the upper portions of the San Jacinto River watershed to Canyon Lake and Lake Elsinore. Therefore, the predictive capability of the model during larger storm events was not thoroughly tested and requires future study as data are collected.

Model Calibration and Validation
After initially configuring the San Jacinto River watershed model, model calibration and validation were performed. Calibration refers to the adjustment or fine-tuning of modeling parameters to reproduce observations. The calibration was performed for different LSPC modules at multiple locations throughout the watershed. This approach ensured that heterogeneities were accurately represented. The model validation was performed to test the calibrated parameters at different locations or for different time periods, without further adjustment. Upon completion of the calibration and validation at selected locations, a calibrated dataset containing parameter values for each modeled land use and pollutant was developed.

Calibration and validation were completed by comparing time-series model results to monitoring data. Output from the watershed model was in the form of hourly/daily average flow and hourly/daily average concentrations for the modeled nutrients for each of the subwatersheds. Flow-monitoring data are available at USGS flow gauging stations located throughout the watershed, while water-quality monitoring data are available at these locations and additional locations where flow was not monitored.

Hydrology was the first model component calibrated and involved a comparison of observed data from in-stream USGS flow gauging stations to modeled in-stream flow and an adjustment of key hydrologic parameters to result in most closely representing the system and reproducing observed flow patterns and magnitudes. USGS gauge stations are shown in Figure 3. The period of record for each gauge varies, with stations 11070270, 11069500, and 11070500 having the most years, and stations 11070365, 11070210, and 11070465 (all three located just upstream of Canyon Lake) limited to less than one year of data in 2001 (Table 2).

The calibration years were selected based on an examination of annual precipitation variability and the availability of observation data. The periods selected were determined to represent a range of hydrologic conditions, including low-, mean-, and high-flow conditions. Calibration for these conditions was necessary to ensure that the model accurately predicted a range of conditions for a longer period of time. Details regarding location, period of historical record, and selected periods for calibration and validation are listed for each gauge in Table 2.

Key considerations in the hydrology calibration included the overall water balance, the high-flow/low-flow distribution, storm flows, and seasonal variation. Two criteria for goodness of fit were used for calibration: graphical comparison and the relative error method. Graphical comparisons are extremely useful for judging the results of model calibration; time-variable plots of observed versus modeled flow provide insight into the model’s representation of storm hydrographs, baseflow recession, time distributions, and other pertinent factors often overlooked by statistical comparisons. The relative error method was used to further support the goodness of fit evaluation through a quantitative comparison. A small relative error indicates a better goodness of fit for calibration. Examples of both methods are shown in Figure 4 and Table 3 for the San Jacinto River at USGS gauge 11069500.

After hydrology was sufficiently calibrated, water-quality calibration was performed. Modeled versus observed in-stream concentrations were directly compared during model calibration. The water-quality calibration consisted of executing the watershed model, comparing water-quality time series output to available water-quality observation data, and adjusting pollutant-loading and in-stream water-quality parameters within a reasonable range. The objective was to best simulate concentrations occurring during low flow, mean flow, and storm peaks at water-quality monitoring stations representative of different regions of the basin (and different land uses, in particular).

Water-quality monitoring data have been collected from various stations throughout the San Jacinto River watershed by LESJWA, the RWQCB, and the RCFCWCD (Figure 5). Based on the availability of data from these stations and characteristics of their respective watersheds, the stations listed in Table 4 were selected for water-quality calibration and validation.

Water-quality modeling parameters adjusted during the calibration process included pollutant buildup, washoff, and subsurface concentrations (primarily interflow for the San Jacinto watershed). Water-quality calibration adequacy was primarily assessed through review of time-series plots. An example time-series plot, representing observed and model-predicted total nitrogen (TN) concentrations at station 318, is shown in Figure 6. Looking at time-series plots of modeled versus observed data provided more insight into the nature of the system and was more useful in water-quality calibration than a statistical comparison. Flow (or rainfall) and water quality were compared simultaneously, providing insight into conditions during the monitoring period (dry period versus storm event). Due to the relative lack of water-quality monitoring data, statistical comparisons were not made. If additional data are collected in the future, it may be beneficial to perform error analyses such as correlation (R-squared), Root Mean Square Error, and Mean Absolute Error.

Nutrient Source Assessment
To assist in TMDL development and watershed management decision-making, modeling analysis predicted the sources of nutrients throughout the watershed and the transport to Lake Elsinore and Canyon Lake. These model results provide decision-makers with information regarding the location of sources and conditions that control the transport of nutrients in the watershed. This information is critical to prioritization of those management options selected as model scenarios to be tested in support of development of an optimal Nutrient Management Plan. Results were reported for three different years representing varying hydrologic conditions when (1) Mystic Lake and Canyon Lake overflowed, (2) Canyon Lake overflowed but Mystic Lake did not, and (3) neither Mystic Lake nor Canyon Lake overflowed. Scenarios 1, 2, and 3 were represented using model results from water years (WY) 1998, 1994, and 2000, respectively (water years extend from October 1 through September 30). The selected model years provide sufficient insight into the nutrient load distribution under the range of conditions (extreme wet and dry periods).

The watershed was divided into nine zones for analysis of spatial variability of nutrient sources and transport throughout the watershed (Figure 7). Total phosphorus (TP) and TN loads are reported for various sources, including failed septic systems, 11 land uses, and loads resulting from overflow of Canyon Lake to Lake Elsinore. For zones 1 and 2, loads were reported as contributions to Lake Elsinore and Canyon Lake, respectively. This aided in analysis of impacts to the lakes. However, for zones 3 through 9, loads were reported in terms of the amount transported from the zone (including transport of loads from upstream zones). This allowed for tracking of nutrient loads (by source) as they are transported through the watershed.

For illustration of model results, TP and TN loads for 1998 (extreme wet year) are shown in Figures 8 and 9. Results were reported in a similar fashion for each modeled year for guidance in TMDL development and watershed management planning (SAWPA 2003).

Development of a Nutrient Management Plan
The development of a watershed strategy for nutrient management was a multistep process that required assessment of previous studies, input from stakeholders, and modeling analysis. Use of modeling tools utilized for TMDL development provides assurance that the recommended strategy is consistent with future regulatory goals for the watershed. To guide the decision process for strategy development, an advisory group, a subcommittee of the San Jacinto River Watershed Council consisting of key stakeholders in the watershed, was consulted on a regular basis for input and updates on the progress of the project. Utilization of previous modeling tools and studies, combined with consultation with local experts and stakeholders for guidance, resulted in the development of a strategy based on the best and most complete information available so that solutions to nutrient impairments in the watershed are scientifically sound and justified.

From the Nutrient Source Assessment, guidance was provided regarding those areas and sources that would benefit from load reductions resulting from management practices. For instance, for very wet conditions when Mystic Lake is full and overflowing, dairy/livestock land uses are relatively large contributors of TN in zones 7 and 6. However, for drier years, dairy/livestock is a much smaller source of nutrients in these zones. For all three hydrologic scenarios, agricultural practices are a major source of nutrient loads for most zones, with the exception of headwater areas where loads are dominated by background sources. It is important to note that although forested (background) areas were determined to be the primary source of nutrients for zones 8 and 9 during wet periods, these loads were primarily the result of higher storm flows from these land uses, and less the result of mismanagement of a nutrient source. Moreover, such loads were contained by Mystic Lake under normal flow conditions.

The priority areas that were recommended for land-use-specific nutrient load reductions were zones 1 through 6 (zone 1 is Lake Elsinore; zone 2 is Canyon Lake) where impacts to Lake Elsinore and Canyon Lake are likely under normal hydrologic conditions (Mystic Lake does not overflow). Considerations were made regarding BMPs specific to land uses that can be applied to critical areas in the watershed. For instance, urban/residential BMPs in zones 1, 2, 4, and 5 may result in noticeable reductions in nutrient loads to the lakes. Model scenarios determined that nutrient load reductions from urban areas in headwater zones 4 and 5 would result in noticeable reductions at Canyon Lake. Identification of sites and types of urban BMPs in these zones can be investigated through future study and modeling analyses using additional tools currently under development by the USEPA (Riverson et al. 2004) for integration with the LSPC modeling system. Also, reductions in septic loads from zones 1, 2, and 5 may be selected as priority management scenarios for consideration.

Drawing upon results from the Nutrient Source Assessment, local expertise (stakeholder involvement), and input from LESJWA, specific projects and studies have been recommended and included in the Nutrient Management Plan. Projects have been categorized as either in-lake or watershed projects. In-lake projects include alternatives suggested in separate LESJWA studies (BMPs within Lake Elsinore and Canyon Lake). Watershed projects have been recommended to either reduce or control nutrients either from the source or during transport through the watershed.

Table 5a

Table 5b

For in-lake projects, specific projects were outlined for both Lake Elsinore and Canyon Lake to

• collect additional data for guidance in future planning;
• study specific in-lake processes to provide guidance for future planning and BMP design; and
• implement BMPs to remedy water-quality impairments.

The final list of in-lake projects includes four for Lake Elsinore and four for Canyon Lake. Four of these projects are currently planned by LESJWA and are in the early stages of implementation to improve water quality in the lakes. Although currently planned, discussion of these projects in the Nutrient Management Plan ensured that the ultimate strategy for nutrient reduction and control is comprehensive. These LESJWA projects included

• Lake Elsinore in-lake nutrient treatment;
• aeration of Lake Elsinore;
• aeration/destratification of Canyon Lake; and
• dredging of Canyon Lake.

In addition to currently planned projects, four projects were recommended in the Nutrient Management Plan to provide more information and guidance in future planning and lake management:

• Water-quality monitoring at Lake Elsinore
• Development of a dynamic water-quality model of Lake Elsinore
• Water-quality monitoring at Canyon Lake
• Development of a dynamic water-quality model of Canyon Lake

With proper planning, the above projects can be combined and structured to improve funding opportunities or meet priority schedules. Therefore, the Nutrient Management Plan provides a systematic approach for project definition, goals, and design, to ensure that monitoring studies are consistent with requirements of planned modeling studies and that schedules correspond to periods when information is needed for planning purposes (e.g., revision of TMDLs based on more information).

Eleven projects were identified to address nutrient reduction or control in the watershed. These projects were categorized by specific issues in the watershed, including sources of nutrients (e.g.. urban, agriculture), physical features (e.g., stream hydraulics, Mystic Lake), data collection, and overall nutrient management (e.g., pollutant trading model). Watershed projects were identified to

• collect additional data for guidance in future planning;
• study nutrient loading characteristics to provide guidance for future planning; and
• implement BMPs to reduce nutrient loads from the watershed.

The following is a list of recommended watershed projects included in the plan to address specific issues or sources identified in the source assessment.

Urban

• Recommend urban BMPs (includes model scenario for analysis of impacts to lakes; model scenarios were tested to determine the impact of reduced urban loads from zones 4 and 5)
• Recommend sewer/septic improvements (Canyon Lake and Lake Elsinore zones)
• Interception and treatment of nuisance dry-weather urban runoff (zones 4, 5, and Lake Elsinore and Canyon Lake zones)

Hydraulic Modifications / Mystic Lake

• Recommend data collection for Mystic Lake to support development of future proposed projects (zone 7). Potential projects include the following:
• Treatment of Mystic Lake water and pump to Lake Elsinore
• Hydraulic modifications to low-flow bypass channel
• Hydraulic modification and control of Mystic Lake outflow and storage capacity
• Recommend riparian habitat restoration (San Jacinto River and Salt Creek) and maintenance of San Jacinto River Channel (zones 3, 5, and 6)
• Recommend study to assess San Jacinto River Levee project impacts and potential flooding of dairies (zone 7)

Agriculture

• Recommend study to determine crop-specific agronomic rates for guidance in management of fertilizer and manure application in agricultural areas. Study will include determination of spatial distribution of varying crops in the watershed to assess the associated spatial variance of fertilization and manure application and sources of nutrients (zones 3, 5, 6, and 7).
• Recommend study to determine feasibility and site of a regional organic waste digester to process manure from local dairies
• Recommend development/maintenance of riparian buffers for agricultural land (study combined with riparian habitat restoration project) (zones 3, 5, and 6)

Other

• Recommend continuation of monitoring of streamflow and water quality throughout the watershed to support future project development and modeling efforts
• Recommend control of trash in the San Jacinto River
• Recommend study to determine pollutant trading options in the watershed

There is no prioritization of projects in the Nutrient Management Plan, as all are deemed valuable and address issues that are diverse and not necessarily comparable using a single prioritization scheme. Rather, the Nutrient Management Plan provides a holistic look at projects deemed important to improving water quality and reducing nutrients in the San Jacinto River watershed. Each of these projects and associated benefits is described in detail in the Nutrient Management Plan to guide planners and watershed managers in future decisions for watershed management and pursuit of funding opportunities.

Advertisement

The Nutrient Management Plan evaluates projects on a common system for comparison of relative benefits and specific issues addressed. Each project was examined relative to key benefits that are considered important factors in development of a comprehensive plan. These benefits include pollutant load control, habitat protection, aesthetic value, lake water quality, lake water quantity, TMDL development (TMDLs developed for Lake Elsinore and Canyon Lake are subject to revision as new data are collected), and TMDL implementation and/or BMPs. Projects outlined in the Nutrient Management Plan are listed in Table 5 with designated benefits marked with “X.” Additional reference tables are provided in the Nutrient Management Plan to categorically describe how each project addresses issues that impact multiple benefits. Based on specific issues of concern, planners and decision-makers can reference these tables for a quick survey of projects, with additional information for each project provided in detailed summaries that include the project background, goals, institutional barriers, estimated costs, and schedule. This information can assist greatly in project selection and planning by LESJWA, SAWPA, the San Jacinto Watershed Council, the RWQCB, and various stakeholders in the watershed.

Conclusions
Due to water-quality impairments observed in Canyon Lake and Lake Elsinore as a result of excess nutrient loads in the watershed, TMDLs are currently under development for TP and TN. A modeling system of the watershed and Canyon Lake was developed to provide the required source assessment and determine load allocations for specific land uses throughout the watershed. Preliminary results of this TMDL study have determined that significant reductions in nutrient loads from the watershed are required to restore the lakes’ water quality to unimpaired conditions. (The final TMDL report is not completed.)

Using the models developed for TMDL calculation, alternative BMPs were tested through model scenarios and a Nutrient Source Assessment was performed to determine locations in the watershed where reductions in nutrient loads would be most beneficial to restoring water quality in the lakes. Use of these models provided assurance that analysis of BMPs and planning options are consistent with implementation plans designed by the RWQCB to meet TMDL goals. Cooperation of stakeholders in identifying management options ensured that alternatives are reasonable and may be pursued in the future as stakeholders are required to meet load reductions defined by the RWQCB.

The San Jacinto Nutrient Management Plan provides a guidance document or roadmap for planning and designing nutrient reduction and control strategies in the watershed. Alternative projects are identified to address numerous sources in the watershed, including agricultural, urban, and dairy runoff; failed septic systems; in-lake sources (e.g., sediment release); and flooding of agricultural areas. Also, special studies are recommended to address data gaps and provide information essential for proper management of the watershed. The Nutrient Management Plan recommends BMPs that could potentially control or reduce nutrient loads to the lakes, or reduce in-lake nutrient concentrations. The combination of these projects and studies provides a comprehensive plan for improving conditions in the watershed and restoring the water quality of Lake Elsinore and Canyon Lake. To assist LESJWA, SAWPA, the San Jacinto Watershed Council, and the RWQCB in planning of future projects in the watershed, the Nutrient Management Plan provides an evaluation of each recommended project based on benefits identified by a stakeholder advisory group to be essential for project selection and prioritization of funding.

Author's Bio: Stephen Carter is a water resources engineer with Tetra Tech Inc. in San Diego, CA.

Author's Bio: Andrew Parker is a water resources engineer with Tetra Tech Inc. in Fairfax, VA.

Author's Bio: Rick Whetsel is a watershed planner with the Santa Ana Watershed Protection Authority in Riverside, CA.

Author's Bio: Mark Norton is a watershed planner with the Santa Ana Watershed Protection Authority in Riverside, CA.