BAY-DELTA SCIENCE CONFERENCE: A new model to address legacy Gold Rush mercury in the Delta

Polluted runoff from the New Idria Mine, a Superfund site, one of 47,000 abandoned mines in California. Photo by MHL Radio.

It can be a bit hard for some to comprehend, but it is true: Here in the 21st century, California is still being impacted from actions taken in the 19th century: Mercury mined in California’s Coastal Range and used in the Sierra Nevada Gold Rush in the 1800s continues to contaminate water that flows into the Delta today. 

In 2010, the Central Valley Regional Water Board adopted a Delta methylmercury TMDL, including a control program to reduce methylmercury and inorganic mercury in the Delta.  The first phase has been completed, which included developing a model for mercury in the Delta.

At the 2021 Bay-Delta Science Conference, Jamie Anderson, Ph.D., Senior Engineer in the Department of Water Resources Delta Modeling Section, discussed the new model and what has been learned.  She acknowledged the many collaborators who contributed to this work, including Reed Harris, Dave Hutchinson, and many staff in DWR’s Division of Environmental Services and Delta Modeling Section.

Mercury is a legacy contaminant in the Delta originating from the Gold Rush era when hydraulic mining in the Sierras used high-pressure power hoses to wash away the mountainsides to find the gold ore.  As a result, the pollution from these 19th-century mining operations was washed into the sediments of the creeks and rivers, eventually finding its way down to the Delta, where it still exists today.

The map on the slide from the US Geological Survey shows the mercury mines in red.  Mercury is a naturally occurring element in the Coastal Range, which was mined and delivered to the gold mines in the Sierra where it was then used to amalgamate the gold.  

Today, there are over 47,000 abandoned mines in California.  Every time it rains, mercury from the abandoned mercury mine shafts or the mine tailings washes into the streams, eventually making its way into the Delta.

Dr. Anderson noted that the concern about mercury in the Delta is because it’s a neurotoxin, bioaccumulating and magnifying through the food chain, exposing people who eat contaminated fish.

The Delta Mercury Control Program established a total maximum daily load TMDL for mercury; one of its goals is to reduce the open water methylmercury loads in the Delta.  The map shows that the central Delta in green is in compliance with the TMDL, and the northern and southern parts of the Delta, shown in pink, are out of compliance.

As part of the phased adaptive management approach for implementing the TMDL, mechanistic mercury models were developed for the Delta and Yolo Bypass.  However, the focus of this presentation will be only on the Delta mercury model.

The Delta model was completed in phase one of the project, with the final report submitted to the Central Valley Water Quality Control Board in August of 2020.  Phase 2 of the Mercury Control Program will begin later this year when dischargers will implement mercury and methylmercury control programs based on the findings of Phase 1.

Developing the mercury model

The mercury model builds upon the Delta Simulation Model 2 (DSM 2), an existing hydrodynamic and water quality model developed and maintained by the California Department of Water Resources.  The model boundaries are Sacramento, Vernalis, and Martinez.  The DSM 2 model grids are shown in blue, inflows in green, outflows in red, and a tidal boundary at Martinez.  The Yolo bypass model, shown in the upper left of the map, provides the mercury fluxes at the Yolo bypass.

To extend the DSM 2 model for mercury cycling, three mercury constituents were added: methyl, inorganic, and elemental mercury, as well as suspended sediments in a two-layer sediment bed.  The calibration period was October 1999 to July 2006, corresponding to the CalFed Mercury studies.  Regression equations were created for the mercury boundary conditions, and data from several previous studies was used.

The Delta Mercury model represents the mercury cycle, including key physical, chemical, and biological processes.  A two-layer sediment bed was also added to the model to represent the mercury cycle and processes, including methylation and demethylation, burial and erosion, settling and resuspension, photodegradation, reduction and oxidation, and atmospheric deposition.

It is a true fate and transport model,” said Dr. Anderson.

Calibrating the model

As with all modeling projects, data was a challenge, and field data availability was highly variable.  The suspended sediment and mercury data were often collected in different places, at different times, and different resolutions.  So the approach was to calibrate the models separately with the best available data for the time periods.

For suspended sediments, the best data available was from October of 2010 to September of 2013, when the data for suspended sediment was collected every 15 minutes by the US Geological Survey.

Since Mercury is transported on suspended sediments, it was really important to get that suspended sediment dialed in for our Mercury simulations,” said Dr. Anderson.  “So once we had that calibrated, then we calibrated the mercury model for October of 1999 to July of 2006, which is when we had the most Mercury data collected during the Cal fed Mercury studies.”

The graphs below show the suspended sediment calibration; the simulated values are shown in blue, the observed values in orange, and the locations for the calibration plots are highlighted in yellow on the map.

 “As you can see, we’ve got a pretty good fit with the suspended sediment calibration; the data was the highest resolution data we had available for this study,” said Dr. Anderson.  “Since Mercury is transported on the suspended sediments, having a good fit for the suspended sediments was very important for this project.”

The plots of the methylmercury calibration are shown on the slide below.  The simulated results are shown in blue, the observations are indicated by the gold diamonds, and the locations of the plots are highlighted on the map in yellow.

As you can see here, the methylmercury data that was available was more sporadic in both space and time, but we got a pretty good fit to that data as well,” said Dr. Anderson.

Methylmercury fluxes in the Delta were also estimated.  The observed values are shown in black on the slide below, and the modeled values are shown in orange.  Dr. Anderson noted that the modeled and observed fluxes matched quite well, which increases confidence in the model.

The tributaries are the major source of methylmercury, and the sediment-water fluxes in the Delta are very small,” said Dr. Anderson.  “It’s definitely a pass-through system.  A lot of what flows in flows out, but a lot of it settles out, so the Delta is a net sink for suspended sediments, inorganic mercury, and methylmercury.”

Model results

The graph below shows the outflows minus the inflows for methylmercury loads from October 1999 to July 2006.

You can see that under all conditions, the Delta is a net sink for methylmercury during this time period,” said Dr. Anderson.  “It was also a net sink for inorganic mercury and suspended sediments, and the loads were highly variable related in part to hydrologic variability.”

The tributaries are the major source of methylmercury, inorganic mercury, and suspended sediments, as shown in the graph below.  The Sacramento River, shown in blue, is the major source of methylmercury.  During wet years when it’s flooded, the Yolo Bypass is also a major source of methylmercury, shown in yellow.  Dr. Anderson noted that the relative contributions from the inflows from the different tributaries vary from year to year with hydrologic variability.

The chart below shows the monthly methylmercury fluxes for the Delta.  The inflows are the positive values on this graph; exports are the negative values.

There’s a lot of seasonal and inter-annual variability,” explained Dr. Anderson.  “A lot of that is tied to the hydrologic variability in this system.  The Sacramento River, shown in blue, and the Yolo Bypass, shown in yellow, are the largest methylmercury sources.  The Yolo Bypass floods intermittently during flood flows, so when it does flood, it is a major source of methylmercury.  The major outflows from the system are at Chipps Island, where the water flows out into San Francisco Bay.  So the majority of the mercury flows through the Delta and back out.”

The chart below shows the modeled mercury flexes for the entire simulation period from October 1999 to June 2006.  The Sacramento River is the major source for both inorganic and methylmercury at 71% and 52%, respectively.  The Yolo bypass is also a major source, at 18% for inorganic mercury and 34% for methylmercury.  (To learn more about why Mercury is methylated in the Yolo bypass, refer to chapters three and four of the studies.)

Most of the mercury leaves at Chipps Island and flows into San Francisco Bay.  Between 5-8% of the inorganic mercury and methylmercury are exported by the Central Valley Project and the State Water Project, respectively.

Dr. Anderson next presented snapshots from the model results (below); the top panel shows suspended sediment concentration, the middle panel shows inorganic mercury, and the bottom panel is methylmercury.  The results are shown for a high Sacramento River flow, the highest Yolo Bypass flow during the simulation period, the median Sacramento River flow, and a low Sacramento River flow.   

These are snapshots in time,” she said.  “They’re not intended to be what will always happen during high medium or low flows, but they are snapshots to give you an idea of the kinds of results you can get from this model.”

These are snapshots in time,” Dr. Anderson said.  “They’re not intended to be what will always happen during high medium or low flows, but they are snapshots to give you an idea of the kinds of results you can get from this model.”

The mercury and sediment concentrations are tributary driven,” she continued.  “During high flow periods, the suspended sediments and the inorganic mercury often have lower concentrations in the central Delta than they do at the edges near the tributary.  This is similar to that map shown earlier of the parts of the Delta that are in and out of compliance for methylmercury.  This is also similar to observed patterns of mercury in fish tissue in the Delta – that they’re lower in the central Delta and higher at the edges of the Delta.  Again, our field data are limited, but the model does give us a chance to look at those spatial patterns.”

Dr. Anderson then showed and animation from the model:

A sensitivity analysis was performed with the model analyzing 22 different variables related to the tributary inflows, the suspended sediment loads, and the mercury loads.  They decreased a single variable by 10% for each sensitivity analysis simulation and ran 22 separate simulations.  They then looked at the effects of those variables on the methylmercury exports at four locations: Chipps Island, the State Water Project, the Central Valley Project, and Staten Island.  A parameter was considered sensitive if it caused a 2% or more impact on methylmercury exports.

The chart on the right of the slide shows the sensitive parameters at each of the four locations in order of sensitivity.  “At Chipps Island, it was the flow and methylmercury concentrations in the Sacramento River and the flow and methylmercury concentrations in the Yolo bypass,” she said.  “For Staten Island, the sensitive variables were the flow and methylmercury concentrations of the east side streams, which are the Consumnes and Mokelumne River.”

For the State Water Project, it was most sensitive to methylmercury concentrations in the Sacramento River and then methylmercury concentrations in the San Joaquin River.  The Central Valley Project was the opposite; the methylmercury concentration in the San Joaquin River was the most sensitive parameter and then the methylmercury concentration in the Sacramento River.”


The model results found that the Delta is a net sink for methylmercury.  Tributaries are a major source, including the Yolo bypass when it is flooded.  However, the production of methylmercury within the Delta itself is small.  And there’s large spatial and temporal variability in methylmercury concentrations and fluxes.

We have this great new tool, the DSM 2 Hg Delta Mercury model,” said Dr. Anderson.  “It is a mechanistic fate and transport model that did a good job replicating observed fluxes, so this is a tool we can now use to gain insight into key processes.  And it can be used for scenario testing, especially looking at relative differences between scenarios.”

The model will be available to support phase two of the Delta Mercury Control Program.  The open water Mercury final report is available on the Delta Stewardship Council website, where it is posted as individual chapters; Chapter Five is the Delta Mercury Model.

The report is posted in its entirety on the DWR webpage for environmental services, applied research, and select the mercury tab.  The technical appendices are available upon request.


Example of consumption warning in the Delta due to mercury contamination.

QUESTION: Why wasn’t the American River included as a source to sample for mercury?

Dr. Anderson: “The delta simulation model has its upstream boundary at Sacramento at the I Street Bridge, and the American River is further upstream than that, so the American River is included implicitly in whatever data is available in Sacramento.”

QUESTION: When you did the sensitivity analysis, did you include inorganic mercury or just methylmercury?

Dr. Anderson:  “In the sensitivity analysis, the simulations ran all of the variables, but for presentation purposes to keep it focused, we only focused on the methylmercury.  But they were all run as part of the sensitivity analysis.  And to the question about the time period for the sensitivity analysis, we ran the full simulation period that we had available, which was October of 1999 through June of 2006.  So we wanted the largest amount of hydrologic variability included in that sensitivity analysis as well.

Print Friendly, PDF & Email