NPDES Permits in New England
Bays Eutrophication Model Evaluation Group
Meeting Report, December 15, 1999
Submitted to OMSAP June 21, 2000
The Massachusetts Water Resources Authority (MWRA) initiated in 1991 a long-term research program to develop a set of tested predictive state-of-the-art circulation and water-quality models for the Massachusetts Bay/Cape Cod Bay region for understanding and predicting the impact of their facilities on the Massachusetts marine environment (including Boston Harbor). The immediate objectives of this effort were to: a) develop the coupled numerical models, b) collect field data for model initialization, forcing and calibration, c) evaluate model predictions through comparison with new field data, and d) use the models to predict potential impacts of changes in the Boston sewage treatment and ocean outfall system then being proposed. Scientists at the United States Geological Survey (USGS) and HydroQual, Inc were contracted to develop the circulation and water-quality models and conduct calibration and evaluation simulations for two one-year periods (1990 and 1992) when sufficient field data were available. The MWRA established a Model Evaluation Group (MEG) in 1992 to advise the MWRA, USGS, and HydroQual on this model development effort. The results of these model studies for 1990 and 1992 and a final review by the MEG were presented in a series of reports in 1995 (see Appendix 1).
At that time, the MEG recommended that the circulation and water-quality models be used to extend the 1992 simulation through 1994. The MWRA Mass Bays baseline-monitoring program showed two unusual water-quality events in these years, an intense phytoplankton bloom in fall 1993 and very low levels of dissolved oxygen in bottom waters in late 1994. The MEG felt that model/data comparisons for these two years would provide added insight into the regional environmental dynamics and predictive model skills. The USGS and HydroQual did examine 1993-1994, and the results are presented in the September 1999 HydroQual report "Bays Eutrophication Model (BEM): Modeling Analysis for the Period of 1992-1994". The Outfall Monitoring Science Advisory Panel (OMSAP) formed the Bays Eutrophication Model Evaluation Group (BEMEG) in late 1999 to review this recent report and recommend any changes in either the modeling and/or the monitoring program which would improve the models skill in circulation and water-quality prediction. The BEMEG held an open meeting at the Woods Hole Oceanographic Institution December 15, 1999 to review the model results for 1992-1994 and discuss possible areas for additional work (see Appendix 2). This report summarizes the major conclusions and recommendations of the BEMEG.
In general, the BEMEG was pleased to see that the additional modeling and data comparison effort recommended by the previous MEG had been undertaken. Extending the BEM simulations through 1994 provides additional experience with the model over different environmental conditions and a further test of the model's ability to capture both seasonal changes and episodic events. In particular, the model successfully reproduces the timing and spatial pattern of the winter/spring phytoplankton bloom, the limiting roles played by silica in Cape Cod Bay during winter/spring and inorganic nitrogen in Mass Bay during summer, and the seasonal variation in dissolved oxygen.
The BEMEG noted, however, that the model did not capture the high diatom concentrations observed in fall 1993, nor did it fully capture the low dissolved oxygen observed in late 1994 - the two events that motivated the inclusion of additional model years. Also, the model could not reproduce the large dynamic range in observed phytoplankton concentrations, raising questions as to how well a model can distinguish anthropogenic change from ambient variability. Ongoing data collection suggests that much of the ambient variability may be associated with unresolved variability in open boundary conditions.
The monitoring data presented at the December meeting showed significant increases in ammonium concentrations and chlorophyll in the northern part of Boston Harbor starting from about July 1998 when the effluent from the Deer Island (DI) wastewater treatment plant (WWTP) began to receive secondary treatment. BEMEG member Donald Harleman requested DI effluent data from MWRA for periods both before and after the beginning of secondary treatment. The first data set was for effluent from the new DI primary plant while it was receiving influent from the north system (about two-thirds of the total metropolitan flow) during the period February 1, 1995 through June 30, 1997. The second data set was for effluent from the new secondary plant during the period July 8, 1998 through December 31, 1999 when influent from the north and south systems was combined and received primary and secondary treatment.
The results of Harleman's analysis of these two effluent data sets are summarized as follows:
- 1. The average concentrations of ammonia (NH3) and organic nitrogen
during the 2.5 years of primary-only effluent at DI prior to mid-1998
a) NH3 in primary effluent = 13.2 mg/L, with 11% removal of ammonium by primary treatment, and
b) organic nitrogen in primary effluent = 8.2 mg/L, with 15% removal of organic nitrogen by primary treatment.
- The average concentrations during the 1.5 years of primary plus
secondary treatment effluent at DI since mid-1998 are:
a) NH3 in secondary effluent = 18.7 mg/L, with 7% increase of ammonium by primary plus secondary treatment, and
b) organic nitrogen in secondary effluent = 4.7 mg/L, with 62% removal of organic nitrogen by primary plus secondary treatment.
- The effect of secondary effluent on chlorophyll is related to the relative change in the effluent ammonium concentration following the change from primary to biological treatment. NH3 in secondary effluent increased from 13.2 to 18.7 mg/L, an increase of 42%.
- The change in the average raw sewage to DI following the combining
of the north and south systems in mid-1998 was as follows:
a) NH3 increased from 14.8 to 17.5 mg/L, an increase of 18%, and
b) organic nitrogen increased from 9.7 to 12.5 mg/L, an increase of 22%.
These results indicate that about half of the 42% increase in effluent ammonium following secondary biological treatment is from the increase in ammonium in the raw sewage due to the combination of the north and south systems. The other half is due to the ammonia generated by the biological treatment process itself. Use of these results in future model studies are discussed below.
Based on the information presented at the December meeting and in accordance with the above observations, we offer the following recommendations.
- Massachusetts Bay exhibits a significant fall diatom bloom, which is thought to provide the last large flux of carbon to the bottom before complete winter turnover and may strongly influence benthic secondary production and other bottom processes. While the impact of the fall bloom on bottom water dissolved oxygen in the fall when bottom temperature and respiration tend to be high and in the following spring are not known, we recommend that the present water quality model be modified to include a third algae component to allow direct simulation of the fall bloom.
- With the significant improvement in computational speed over the last decade, we recommend that the water-quality model use the same spatial grid as the circulation model. In particular, within the existing water-quality model domain, the horizontal and vertical grid spacing should be the same as used in the circulation model. This should eliminate questions about errors introduced by the spatial collapsing scheme presently used in the water-quality model. The spatial resolution used in the circulation model should be revisited, to see if this should also be improved for future studies.
- Although "projection" runs (i.e., model runs comparing conditions with existing versus future outfall, and with primary versus secondary treatment) were apparently not part of the current Statement of Work, we recommend these comparisons be continued and tracked as additional changes are made to the model. New "projection" runs should be made for existing versus future outfall locations, and with primary versus secondary treatment using the organic nitrogen and ammonium effluent data presented in the Summary section above. Also, the variation among projection runs should be gauged against the year-to-year variability (both measured and simulated) to assess the relative impacts of anthropogenic change versus ambient variability.
- Both field data and model results indicate that the major source of nutrient input to the Mass Bays system occurs through advection of western Gulf of Maine waters around Cape Ann. The volume and property fluxes of this upstream source are poorly sampled by the present monitoring program, so that changes within the Mass Bays due to real changes in the upstream boundary conditions will be missed in model simulations conducted with poorly known "climatological" boundary conditions. We recommend three specific actions: A) Conduct model studies to determine the sensitivity of the model simulations for 1992-1994 to realistic changes in the upstream boundary conditions. B) Develop a plan to begin collecting in-situ time-series measurements of currents and water properties along the upstream section of the model open boundary. The lack of direct measurements to characterize the upstream inflow and its variability severely limits any model predictive skill. Questions concerning the types, locations, and frequency of sampling need to be addressed, e.g., should measurements be made along the open boundary of the circulation model or the open boundary of the water-quality model or both. C) Investigate recent efforts to develop a Gulf of Maine ocean observing system. The University of Maine is funded to develop such a system (GoMOOS) in the next two years, and the Regional Association for Research on the Gulf of Maine (RARGOM) is considering efforts to augment or complement the Maine GoMOOS plan. MWRA could play an important role in helping to design an effective Gulf of Maine observing system that could provide near real-time measurements of currents and water properties on the open boundaries of the MWRA circulation and water-quality models. The combination of data collected by the MWRA monitoring program and a Gulf of Maine observing system offers the best hope of providing timely accurate open boundary conditions.
- The closing of the Nut Island WWTP in 1998 and switch to secondary treatment at Deer Island made a significant change in the distribution and quality of the nutrient input to Boston Harbor. In particular, high levels of ammonium were observed near the northern harbor. We recommend that the BEM (with improved resolution; see recommendation 2 above) be used to simulate the 1998-1999 period to see how well the model captures the increased chlorophyll observed in and offshore of Boston Harbor. Again, the pre- and post-secondary organic nitrogen and ammonium effluent concentrations presented in the Summary section above should be used in this simulation.
- Related to 4 and 5 above, model mass balance studies should be conducted to determine the temporal and spatial distribution of the source of various pools of nitrogen. That is, how much of the ammonium observed at x, y, z, t is from treatment plant versus the open boundary.
- There is generally good agreement between the sediment model and in-situ measured fluxes of oxygen, ammonium, nitrate, phosphate, silicate, and denitrification, with the exception of some summer data for denitrification and phosphate. The report recommendation that benthic diatoms should be investigated is a good one, but prior to addition of a benthic diatom state variable, the model and field data should be analyzed for the predicted/observed light levels at the sediment-water interface. The predicted areal pattern of light could be used to assist in deciding whether the inclusion of this potential source of oxygen to the model is warranted. If it is, the experimental benthic flux work should include an illuminated treatment.
Recommendations for Report Addenda
- Complete documentation of the circulation and water-quality model parameters used in the 1992-1994 simulation need to be provided, especially noting which values were changed from the values resulting from the 1990/1992 calibration study and/or changed during the 1992-1994 period to better fit with observation. Also note any changes in parameter values suggested by more recent observations.
- Complete documentation of the boundary conditions used in the 1992-1994 simulation should be provided. This includes descriptions of the various parameters, their spatial and temporal variability, and key assumptions used to derive the values specified for the 1992-1994 study. For example, do the surface wind stress and insolation change hourly while the temperature and salinity specified along the Gulf of Maine boundary change only monthly or yearly? Since step changes in boundary conditions generally introduce disturbances that propagate through the model domain, comment on the effects of step changes versus time ramp changes in boundary conditions on the main model results and whether the time dependence of the boundary conditions should be changed in future model simulations.
- The water-quality model takes as input the grid-collapsed temperature and salinity fields as computed by the hydrodynamic model. A comparison of simultaneous temperature, salinity and density from both models should be made for the 1992-1994 period and presented with a discussion about the skill of the water-quality model to capture the vertical stratification seen in the hydrodynamic model. Note that if the hydrodynamic and water-quality models are run on the same grid, there is no need for grid collapse and this recommendation is not necessary.
- One of the major difficulties in comparing model results with in-situ measurement data is the mismatch in spatial and temporal scales between model and nature. We recommend that the experimental uncertainties inherent in the field data due to instrumental and methodological errors and the spatial and temporal scales of natural variability be estimated and shown in all model/data comparison figures.
The water quality model simulations for bottom water dissolved oxygen (DO) during the 1994 stratified season were in the right direction in the sense that they tended to be lower than in previous years, but did not reach the observed minima for the 1994 season. We do not really know if the model's underestimate of the minima is due to a failure to simulate metabolism (primary production may be low in the model compared to observations) or to a lack of sufficient detail in some physical processes (coarse-scale advective or fine-scale vertical processes). Bottom-water DO is influenced by such factors, including local water column and respiration rates, rates of sub-pycnocline water column or benthic primary production, the fine-scale structure of stratification and its relationship to an irregular bottom topography, and horizontal advection of water into local areas. The progression of DO decline in the nearfield sampling area during summer stratification is very sensitive to the combination of these biological and physical factors. Other modeling has shown that different combinations of factors can approximate a given annual bottom-water DO minima, moreover, the temporal progression of decline varies with the mix of factors and specific rate processes. Significantly, the shape of the observed seasonal decline in nearfield DO was not reproduced by the model. Typically (1992-1994), bottom-water DO decline has been slower (or has no decline) early in summer and then has increased towards fall turnover; in contrast, the model tends towards a fairly constant decline for the whole period.
Overall, the patterns suggest that the correct tendency for seasonal DO minima is produced by the model, even if not fully. But one cannot have confidence that results occur for the right reason, i.e., that the dynamics and scales acting as the mechanism for DO decline in nature are indeed drivers in the model. Without that assurance, we cannot have full confidence in the model's ability as an event-scale prediction tool, even though we are confident in it as a broader, scenario-projection tool.
If an event-level predictor is desired of the model, both sensitivity analyses of the model and comparison of observational uncertainties (each recommended above) would help resolve apparent model-observational differences. A simple place to start is to revisit the calibrated model coefficients for select physical and metabolic coefficients (including temperature functions) against the MWRA database. Comparison would note whether critical processes expressed in nature appear to be more dynamic than would be captured by a fixed (i.e., time-invariant) model coefficient.
Appendix 1. Model reports and MEG review released in 1995-6.
Beardsley R, Adams EE, Harleman D, Giblin AE, Kelly JR, O'Reilly JE, Paul JF. 1995. Report of the MWRA hydrodynamic and water quality model evaluation group. Boston: Massachusetts Water Resources Authority. Report ENQUAD ms-37. 58 p.
Blumberg, AF, Ji, Zhen-Gang, Ziegler, CK. 1996. Modeling Near-Field Plume Behavior using a Far-Field Circulation Model. Journal of Hydraulic Engineering. Vol. 122, no. 11. pp. 610-616.
Hydroqual, Normandeau. 1995. A water quality model for Massachusetts and Cape Cod Bays: Calibration of the Bays Eutrophication Model (BEM). Boston: Massachusetts Water Resources Authority. Report ENQUAD 95-08. 402 p.
Signell, R.P., Jenter, H. L. and A.F. Blumberg. 1996. Circulation and effluent dilution modeling in Massachusetts Bay: model implementation, verification and results. Open-File Report 96-015. U.S. Geological Survey. 121 p.
Appendix 2. Model Evaluation Group Membership, December 1999 Meeting Agenda, and Attendees
Outfall Monitoring Science Advisory Panel
Bays Eutrophication Model Evaluation Group Meeting
December 15, 1999, 9:00 AM to 3:30 PM
Woods Hole Oceanographic Institution, Quissett Campus, Carriage House
Dr. Bob Beardsley (chair), Woods Hole Oceanographic Institution
Dr. Eric Adams, Massachusetts Institute of Technology
Dr. Jeff Cornwell, University of Maryland
Dr. Don Harleman, Massachusetts Institute of Technology
Dr. Jack Kelly, US Environmental Protection Agency
Mr. Jay O'Reilly, National Marine Fisheries Service
Dr. John Paul, US Environmental Protection Agency
Observers: Dr. Brad Butman, USGS; Ms. Cathy Coniaris, OMSAP staff; Dr. David Dow, NMFS; Mr. Jim Fitzpatrick, HydroQual; Dr. Rocky Geyer, WHOI; Dr. Anne Giblin, MBL; Dr. Carlton Hunt, Battelle; Dr. Russ Isaac, MADEP; Mr. Rich Isleib, HydroQual; Dr. Wendy Leo, MWRA; Dr. Wayne Leslie, Harvard; Dr. Matt Liebman, EPA; Dr. Jason Link, NMFS; Dr. Mike Mickelson, MWRA; Dr. Andrea Rex, MWRA; Dr. Jack Schwartz, MADMF; Dr. Rich Signell, USGS; and Mr. Steve Tucker, Cape Cod Commission.
9:00 - 9:15
Welcome, Purpose of Meeting, and Introductions
Bob Beardsley, WHOI, MEG Chair
9:15 - 11:00
BEM Model Overview and Results for 1992-1994
Jim Fitzpatrick and Richard Isleib, HydroQual
11:00 - 11:30
Overview of 1992-1999 Baseline Monitoring Features
Carlton Hunt, Battelle
11:30 - 12:00
Questions and Discussion
12:00 - 1:00
1:00 - 3:30
MEG Discussion and Comments for the Final MEG Report
*The MEG met afterwards to draft their final report to the Outfall Monitoring Science Advisory Panel.