A Cross-Media Approach to Saving the Chesapeake Bay E L A I N E L . APPLETON
n 1992,Robin Dennis and his colleagues at EPAS National Exposure Research Laboratory in Research Triangle Park, NC. began searching for a "Grand Challenge." The researchers from the Atmospheric Modeling Division were looking for a broad scientific problem that would stretch their skills in supercomputer simulations of air quality problems and be a suitable component of the National High Performance Computing and Communication "Grand Challenge" ~.supercomputing . -research program. They found By linking powerful that challenge in a ground-breakingcrosssupercomputer models, media study of the Chesapeake Bay that scientists are studying simulates the transfer of pollutants from one the role of atmospheric medium-in thiscase, the air over the baydeDosition in the to the 64.000-sauaremile watershed and dChesapeake Bay's water timately the bay itself. The- cross-media quality problems. project builds on the modeling work under way at the Chesapeake Bay Program, a cooperative agreement among the jurisdictions in the hay's estuary: Maryland, Virginia, Pennsylvania, the District of Columbia, the Chesapeake Bay Commission, and EPA. Earlier this year the North Carolina air modelers joined their supercomputer model of the bay's airshed with the Bay Program's watershed model and the U.S. Army Corps of Engineers' threedimensional water quality model. The initial results are already revealing new insights into regulatory strategies needed to reduce eutrophication in the bay. By using supercomputer models to "try out" different control strategies designed to reduce nutrient flows, regional and federal regulators are testing the long-term effective-
550 A.
VOL. 29, NO. 12.1995 iENVIRONMENTAL SCIENCE & TECHNOLOSV
ness of potentially expensive, complex, and politically sensitive regulatory decisions. The health of the Chesapeake Bay, whose nine tributaries bring water from as far away as NewYork and West Virginia, has declined for decades. From 1990 to 1992 alone, the tributaries dumped 600 million Ib of nitrogen and 30 million lb of phosphorus into the bay, according to the US. Geological Survey ( 1 ) . Nutrients, which come from utilities, sewage treatment plants, farm fertilizer runoff and animal waste, and automotive exhaust, create huge algal blooms that deprive underwater grasses of sunlight and use up oxygen when they decompose. causing anoxia, or "dead water'' in up to 15%of the bay's total volume (2). Since 1983,the Chesapeake Bay Program has been studying how to control these excess nutrient loadings. In 1987, the group called for phosphorus and nitrogen loading reductions of 40% below 1985 base loads by the year 2000, a goal estimated to produce a 20% reduction in dead waters. Despite studies indicating that atmospheric deposition contributes 20-35% of the millions of kilograms of nitrogen that wend their way into the bay (31, the Bay Program to date has controlled only waterborne nutrients. "The Bay Program has always determined air sources to be uncontrollable," said Lewis Linker. However, the water-only nitrogen reduction strategies have not worked. A formal evaluation of the role atmospheric deposition will play in the overall bay nutrient reduction strategies is scheduled for 1997.
Nitrogen not easily controlled As a result of implementing controls such as banning the use of phosphate-laden detergents, 48% of the phosphorus reduction goals had been met by 1992, when the program conducted its second major review. But nitrogen loading is not as easily understood or as easily controlled. By 1992 only9% of
0013-936X195/0929~55OA$~Q.00/0 1995 American Chemical Society
the nitrogen reduction goal had been met (21, most significantly through point source reductions, such as biological nutrient removal, at sewage treatment plants affecting the West Shore basin (4). This was particularly important because nitrogen reduction is more effective than phosphorus reduction in reducing anoxic volume days (5).Today the bay’s health remains in jeopardy. According to the US.Geological Survey nitrogen continues to increase in the Susquehanna and Potomac Rivers, which are major bay tributaries, “Historically,there has not been much in terms of water quality controls [of nitrogen] being effective,” said Dennis. Nitrogen nonpoint source controls are difficult to implement, point source nitrogen controls have only recently begun, and there are uncertainties in watershed retention of nitrogen, he said. “Au we know is that nitrogen reduction is nowhere near on target, so there’s a lot of interest in what air quality can do to help.” Because nitrogen cycling through the atmosphere is nonlinear and far ranging, deposition at any one area comes from several sources. Thus, it is impossible to gather precise nitrogen source attribution data through monitoring don& quality modeling is necessary, reported Dennis (3).Once nitrogen is deposited, it moves through the watershed and various portions of the bay in different ways, depending on current movement in the tributaries and uptake by various receptors such as soil, trees, and benthic organisms. Therefore, water quality modeling is needed to transform the atmospheric loads deposited throughout the basin to nitrogen loads delivered to the bay.
A collaboration is born When Robin Dennis first began looking for a “grand challenge,” environmental model coordinator Lewis Linker of the Chesapeake Bay Program had just begun to direct the linkage oftwo supercomputer models: one that simulates land use and its effect on water quality and one that simulates the bay itself. It was a huge task designed to provide regulators with the scientific information needed to reduce the millions of pounds of nutrients, primarily nitrogen and phosphorus, that flow into the bay yearly. Linker’s objective was to pinpoint the nutrient sources and understand how they act upon receptors, including the water in the bay’s main stem, its tributaries, and the land surrounding the bay. Working with the Anny Corps of Engineers, Linker was combining bay and estuary modeling results, but he realized that in an area the sue and complexity of the Chesapeake Bay it was impossible to get accurate estimates without including airborne sources. Little data existed, however, on the effects of atmo-
spheric deposition of nitrogen to the bay, a problem that the Bay Program suspected was significant. Since 1988, when Environmental Defense Fund research indicated that atmospheric nitrate deposition was a major anthropogenic source of nitrogen to the bay (31, Linker and his colleagues had wanted to add comprehensive simulations of air deposition to their water quality and land use simulations. However, to their knowledge, no one had ever attempted this kind of cross-media modeling. When Dennis heard about the Chesapeake Bay Program, he seized upon the idea of using EPKs Regional Acid Deposition Model (RADM), a supercomputer model designed to simulate the conditions that cause acid rain, to study atmospheric nitrogen deposition to the hay. RADM already included nitrogen as one of the species it could simulate. Dennis offered to link RADM to Linker’s water quality models. It was auspicious timing. The 1992 bay agreement amendments mandated that the signatories “incorporate into the Nutrient Reduction Strategies an air deposition component which builds upon the 1990 Clean Air Act Amendments (CAAA)and explores additional implementation opportunities to further reduce airborne sources of nitrogen entering Chesapeake Bay and its tributaries.” Linker accepted Dennis’s offer. It is not surprising, suggested Dennis, that neither his office nor EPKs Office of Air Quality Planning and Standards, which had been planning NO, VOL.29. NO 12.1995/ENVIRONMENTALSClENCEBTECHNOLOGY.551
A
reduction strategies in its efforts to reduce tropospheric ozone, had not heard of Linker’s work. “By taking a broader cross-media perspective,” he said, “we see that we’re in a potential win-win situation, where something that we do for one pollutant actually benefits another problem area. In our typical bureaucratic mind-set where we’re in little cubicles, we would have ignored that. We wouldn’t have knOWll.”
And so the process of linking the three modelsthe watershed model, the Chesapeake Bay Water Quality Model (CBWQM), and the RADM airshed model-was begun. Air deposition source loads could be fed into the watershed model to determine nitrogen deposition and movement on land, and from there into the estuary model to determine their effects in the water. By the beginning of 1995, Dennis and Linker had joined all three models.
Cross-country modeling The modelers are now running different scenarios to evaluate different control strategies.Whatwould happen, for example, in the year 2000 if regulators impose limit-of-technology control strategies to reduce nitrogen emissions from utilities that export nutrients into certain tributaries? To do a model “run,” specialists in Research lli552 A
VOL. 29. NO. 12, 1995 I ENVIRONMENTAL SCIENCE &TECHNOLOGY
angle Park o r h a p o l i s , MD, send these queries, or “model input files,” via high-speed communication lines to a Cray C90 computer at EPA’s National Environmental SupercomputingCenter in Bay City, MI, where the runs are performed. pqically, the airshed model is run first so results can be used as input into the watershed and water quality models. The 3-D airshed model tracks nutrient emissions across the eastern United States by dividing the area into 20,000 square cells, each 80 km on a side. A fine grid of 60,000 20-km-by-20-km cells covers the Chesapeake Bay region. Piled one on top of another, the cells create 15vertical layers reaching about 15km high to the top of the free troposphere. By simulating the airshed in three dimensions, modelers can determine vertical as well as horizontal movement of nutrients. The water quality model represents the bay as 4073 computational cells that average 6 mi long, 2 mi wide, and 5 ft deep, stacked up to 15 layers in the deepest parts of the bay. The model includes a hydrodynamic model that simulates water flow rates into the bay, the mixing of the bay with coastal waters, and the mixing of water within the bay. One “what-if“ scenario run on the airshed model alone requires 300 h of computer time. “RADMis a monster,” said Linker. The watershed and water quality models are smaller, he said, with a typical watershed run taking 3 W O h. It is not surprising that the modelers are caught between their desire to refine the model into ever-smaller grid cells and time steps and the growing amount of computer resourcessuch refinement requires-the developers are continually running up against the limits of computational technology Combining air quality and water quality modeling in cross-media simulations is rife with challenges, including such problems as calibrating equivalent units of emissions in air and water and determining the sources of far-ranging atmospherically deposited nitrogen or the retention rates of nitrogen through forests. And, given the amount of data simulated in these models, data validation via traditional field monitoring is challenging if not seemingly insurmountable. For example, Linker’s watershed model, designed to simulate nutrient loads delivered to the stuary under different management scenarios, includes hydrological data (temperature,precipitation, wind, solar radiation, and other factors) from 1984 to 1994, nonpoint s o m e loads, and nine types of land uses. In river reaches, modelers calculate nonpoint source loads, point source loads, and water supply diversions on a one-hour time step. The complexity of cross-mediamodeling is exacerbated by emissions inventories that change as more information becomes available. Such massive complexity poses real-world problems, said Linker. “As the simulations become more and more extravagant, there’s a greater and greater reliance on simple observed data to make sure we’re getting the simulation right.” In fact, calibratingthe models is critical to “getting the right answers for the right reasons,” he said. The developers calibrate the models by comparing simulation results with observed data. In addition, a panel of scientists from
computaoonal cells that
ion. temperature. wind cloud cover. land use
Regional Acid Deposition Modal In development since 1983. completed in 1989. Divides eastern Unlted States into 80 x 80 km cells with a tine gndolmxm kmcellsoverthe oaywalershea Computes
outside the Chesapeake Bay Program reviews the models quarterly.
Trial run lor future regulations Despite these challenges, the integration that today exists among the models has given Bay Program officials information about nitrogen deposition they did not have before. ’&pically. policy makers ask the modelers to create different scenarios hy “tweaking” emissions sources to project the outcomes of various control strategies. For instance, said Linker, they have used models to ask, “Do we need
to control both nitrogen and phosphorus, or can we control only one to meet our water quality ohjectives? Or are the nutrients equal in different places, so if we reduce a kilogram of nitrogen in the Susquehanna River at the top of the bay. would it be equivalent to reducing a kilogram in the James River at the lower end of the bay?“ It turns out that geography is indeed important, and that reducing nitrogen at the northern and central portions of the bay has a far greater effect on anoxic days measured in the hay’s main stem than does reducing the same amount of nutrient at the south VOL. 29. NO. 12.1995 I ENVIRONMENTAL SCIENCE &TECHNOLOGY. 5 5 3 A
btsbov‘s emissions sources - ~ ~ . ~ _ _ , “airshedand ~~~
~
~
~~~
~
~
~
a cmssmedia model was used this summer 10 establisn the s u e of me wap+&.n Bay’s aushed. the geographic region that contribures 70.80% h e Mmaspheric &ate deposhon tothe natershed. lnnlal reSLllS show 8 t h airshed is mcfe than five limes the size of the bav s walershed. deral ratherthan regional regulations may mputs into the bay. The modeling also mdi anonary sources contribute equally to nitro
Modeling also has shown that, although utility and mobile source emissions are almost equally responsible for nitrogen deposition IO the eastern (uriliries of NO, emissions, and mobile sources 35%), they are geographically separated. The utility emissions come more from the west, and the mobile sowces affecting the bay come from the more densely populated eastern seaboard (2). ~~~
looking to ozone reduction
864 A
.
end of the bay. Linker warns, however, that anoxia is just one measure of health, and that nutrient reduction in southern tributaries may have other positive effects. To understand the effects of federal regulations such as the 1990 Clean Air Act Amendments on the bay, Dennis thought it was important to determine the geographical boundaries of the airshed, dek e d as “that contiguous region of air emissions that contributes a majority [70-80%1 of the emissions to the bay watershed and main stem” (2).Although earlier runs of the combined water and estuary models had indicated that atmospheric nitrate deposition contributed at least 25% of the total nitrogen to the bay, it was still undear exactly where it was coming from. Duringthe summer of 1995, after first determining that the range of influence of nitrogen, especially the termination product nitric acid, is, conservatively, 700-800 km, Dennis estimated that the airshed is at least 5.5 times the size of the bay’s watershed. That means the airshed extends as far as Montreal and Toronto, Canada, Detroit, MI, and Cincinnati, OH. NO, emissions from this airshed conStiNte about 30% of the NO, emissions in the entire eastern United States and Canada and account for approximately 70% of the anthropogenic nitrogen deposited in the bay watershed, said Dennis. He noted that emissions from west of the bay affect it the most because of prevailing winds (3).However, he cautioned that, because of bias in the model, this is a conservative estimate and, in fact, the range of influence of nitrogen could be as far as 1000 km.
VOL. 29. NO. 12,1995 I ENVIRONMENTAL SCIENCE &TECHNOLOGY
EPA policy makers are taking these results seriously. Until now, regulators have dealt with Chesapeake Bay problems solely on a state and regional basis. Now, said John Bachman, associate director for science policy at EPKs Office of Air Quality Planning and Standards, “If Robin Dennis’s preliminary modeling tuns out to be right, and his footprint for the Chesapeake Bay goes all the way to Ohio, then it turns out to be a federal issue, because you need federal regulations to convince Ohio to care about this. We’re looking at that.” Initially, policy makers are looking to ozone reduction plans, which inchde NO, reductlon strategies, to help the bay. Modelers compare “milestone” scenarios-which assume either no new control strategies beyond 1985 or achievement of the Bay Program’s 40% nitrogen loading reduction goal-with scenarios that could occur if federal omne reduction strategies are implemented. The latter model run indicates that full CAAA implementation would reduce nitrogen entering the bay and provide an additional slight reduciion in anoxia days over the 20% reduction projected by the Bay Agreement ( 4 ) . But Bachman is also interested in the results of additional modeling runs that examine the effects of strategies devised by the Ozone Transport Commission (OTC),the 13-memberregional board made up of northeastern states and the District of Columbia. OTC controls are more stringent than current CAAA strategies and would, according to Dennis, double the positive effects on the bay of the full implementation of CAAA controls alone. Preliminarymodeling runs show that an even more effective strategy would be to implement OTC controls on the entire Chesapeake Bay airshed. Another issue is whether there is a need to implement ozone restrictions beyond the c m n t summer restrictions. Model runs show that it is equally important to the bay to regulate ozone in the spring, and perhaps also in the fall and winter. Bachman said, “That’scritical for us to know. I don’t want to control ozone in winter as well as summer, but I have to know that as a policy maker, and these models are helpful in that regard.” Bachman has asked Dennis and Linker to run more scenarios to determine if and where wintertime controls are necessary. “It certainly will be critical within the first few hundred ki-
beaan
s-media emissions trading
Am . i.e a EPA - in- August wil mvesbgatethe parsibiii of cmssmtfra emissions trading as a new approach to &wing nitrogen loadings to Chapeake Bay. W i n g with the Envimmnemal O e h e Fund (EOR. the A~encyhopes to devise a NO.tradino framework dona the lines of e W n a dioxide The Chesapeake M-Pmiectwillexamine tho $a& bi!Ay of using emissions trading between air aad water sources, includingtrading credii bmween pwer plants and mobile sources, to reduce the sancspheric deposition of nitrogen to the bay. EPKs Chris Knopeg who is helping to direct the pmiect acknowledged that cross-media trading cwld be a long way off.“he concern of the Bay Program is mat they‘re not the W4 they‘re not Rabbnal, and sa they can’t sey ‘All of you stsfss need la implement comoh‘ Their concem with mrdtbnedia trading is that it would appear to be loosening [exkthgl conBdS.” Moreover, m do 0 1 7 kid of smigsionstrading requires a h r w g h un-
I
I.
lometers of the bay,” he said. “The question is where to draw that line.” Another policy question that may be addressed with the aid of cross-media modeling is the cost effectiveness of different regulatory approaches. Chris Knopes, a policy analyst with EPXs Office of Policy Planning and Evaluation, would like to know what the cost differentialis between implementing atmospheric controls and water controls. “The air people are saying they can reduce NO, emissions for $1000 a ton from the stack,” reported Knopes, referring to selectivecatalytic reduction. In contrast, it can cost “a couple of hundred to a couple of hundred thousand dollars per ton” to reduce nitrogen loading from water sources. “The question is what does that [airborne reduction cost] mean in terms of the cost per ton of nitrogen not delivered to the hay? We don’t know yet.” Knopes is hoping that a combination of traditional cost data analysis and crossmedia air and water quality modeling will help answer this question. As the Bay Program gears up for a reevaluation of its nutrient reduction goals in 1997, Linker and Dennis are further refining the model scenarios to add to understanding of source-receptor relationships and nitrogen retention. They also hope to be able to simulate underwater grasses and benthic organisms to provide tributary-specific goals for nutrients based on habitat improvements. This adds an “ecosystem”measure of health to the “anoxic days” measure the program has so far focused upon, said Linker. “Weknow that the 40% nutrient reduction improves dead waters, hut what does it do for the underwater grasses of the bay and the benthic organisms, both of which support the bay’s fisheries?“
derstanding of source-recepmr relatlonshaps, something &at grid-based models such as the ba* airshed model. in contrast to plwne models, aren‘t typically designed to do. Such an incentive program will a h require more Ummugh understanding of the temporal naf loading an how m e n moves through the watershed, including forests, soil, and proundwatec. untwtunateb, the science on this aspect of nitrogen h very weak But tho prcges dthe Bay Program nutriat reduetion goals is up for reevaluation in 1987, and Know and EOF economisr Brian Mollon have h@ hopes that the trading plan, M c h would plaoe a cap on the mass of emissions and rate of deposihbn a h a d by dl s w c e g vi! becoma ERe atmospheric dsposbn p o r t h of the Chesapeake Bay F’mgram‘s Nutrient R e d m n S r a W . -ELAINE L APPLETON
n.dmen
I
The complexity of the questions that the crossmedia modelers address and the urgency of finding regulatory solutions will continue to increase as population and development pressures add more nutrients into the bay Linker and Dennis are now refining the linkages between the models and plan to complete a finalintegrated model by early 1997. “The nutrient reduction goal of the year 2000 really becomes a cap not to be exceeded,” said Linker. “But growth will continue and sewage treatment plants will discharge more loads. We have to get smarter and more cost-effective about the work. That’s driving a lot of the inquiry into air deposition.”
References (1) Zynjuk. L. D. Chesapeake Bay: Measuring Pollution Re-
duction;U.S.Cealogical S w q , Washington,DC, 1995;Fact Sheet FS-055-95. (2) Dennis. R. L. “Usingthe Regional Acid Deposition Model
to Determine the Nitrogen DepositionAimhed of the Chesapeake Bay Watershed.”In AtmosphericDeposition to the Great Lnks and Coastal Wafers;Baker, J., Ed.; Society of Environmental Toxicology and Chemistry:Pensacola. FL, in press. (3) Fisher, D. et al. Polluted Coartal Waters: The Role ofAcid Rain; Environmental Defense Fund New York, 1988. (4)N 1994 National Envimnmental Supercomputing a n t e (NESC) ~ Annual Report US. Environmental Protection Agencp NESC Bay Ciry, MI, Feb. 1, 1995: EPA12081 K-951001. (5) Cerco, C. E; Cole, T. M. “Three-DimensionalEumphication Model of Chesapeake Bay”; U.S.Army Corps of Engineers:Vickshurg. MS. May 1994; Technical Report EL94-4. (6) Blankenship. K.BayJoournall992,JulylAugust,pp. 1. P5. Elaine L. Appleton is afreelance writer based in Newburyport, MA. She is former senior editor at Datamation magazine.
VOL. 29, NO. 12, 19951ENVlRONMENTAL SCIENCE &TECHNOLOGY m566 A