The Chesapeake Bay Model

The most recent meeting of the Potomac Watershed Roundtable was in Warrenton, VA at Lord Fairfax Community College and had a series of speakers on the Chesapeake Bay strict pollution diet, the Total Maximum Daily Load (TMDL) mandated by the EPA to the six Chesapeake Bay Watershed states and the District of the Columbia. The TMDL addresses only pollution from excess nitrogen, phosphorus and sediment. No action has been taken or at this time is intended on other pollutants that might be present in the Chesapeake Bay and its tributaries. Two of the speakers, walked the Roundtable audience through some of the issues and uncertainties will impact implementing a plan to achieve the TMDLs on the local level.

Clifton Bell of Malcolm Pirnie, Inc. and Mike Rolband of Wetland Studies and Solutions, Inc. walked the group through some of the issues with the EPA “final” TMDL loading levels and how they were obtained. The TMDLs were created by a series of models of the Chesapeake Bay Watershed that include various land use models, water quality models and watershed models. These computer models are mathematical representations of the real world that estimate environmental events and conditions. The models are at best imperfect, but they are nonetheless the best tool available to view the 64,000 square miles of the watershed. The Chesapeake Bay and its watershed are so large and complex, that scientists and regulators rely on computer models for critical information about the ecosystem’s characteristics and the impact of various environmental actions to reduce pollution. Mr. Bell lifted the curtain to point out some aspects of the models.

Although model simulations are an important part of the Chesapeake Bay regulatory mandate, they are not considered by the EPA to be perfect forecasts. Rather, model simulations are the current best estimates. The Chesapeake Bay Model is really made up of several models that are added together to create the whole: the Watershed Model, the Estuary Model, the Scenario Builder, the Airshed Model, the Land Change Model and the Land Use Models.

The Watershed Model incorporates information about land use, fertilizer applications, wastewater plant discharges, septic systems, air pollution, farm animal populations, weather and other variables to estimate the amount of nutrients and sediment reaching the Chesapeake Bay and which of the major land uses produce these pollutants. This is the most robust and calibrated portion of the model sequence because it is calibrated and validated on the major tributary basin levels where there is decades of measured water quality data available. The Watershed Model divides the 64,000-square-mile Chesapeake Bay watershed into more than 2,000 segments.

The Estuary Model examines the effects that pollution loads generated by the Watershed Model have on water quality. In the Estuary Model, the Chesapeake Bay is further divided into more than 57,000 computational cells and is built on two sub-models: The hydrodynamic sub-model and the water quality sub-model. The water quality sub-model is well calibrated for dissolved oxygen.. However, it is unclear that dividing up the world in ever smaller pieces gives you better resolution when there are not enough hard data points in the cells. Zooming in on a picture without adequate pixels does not increase the resolution.

The Scenario Builder can generate simulations of the past, present or future state of the Chesapeake Bay watershed to explore potential impacts of regulation and management actions and evaluate alternatives. This model creates the assumptions in the regulated community much like Sims game.

The Airshed Model uses information about nitrogen emissions from power plants, vehicles and other sources to estimate the amount of and location where these pollutants are deposited on the Chesapeake Bay and its watershed using information from vehicle registrations and DOT and weather.

The Land Change Model analyzes and forecasts the effects of urban land use and population on sewer and septic systems in the Chesapeake Bay watershed based on: population trends and forecasts, migration, satellite imagery, and waste water treatment plant service area and data. Finally, the Land Use model estimates the types and amounts of pollution that run off a particular land use are based on comprehensive reviews of the latest scientific literature.

Since most of the models were built and calibrated on the data that has been collected over the years they are well calibrated on the major tributaries and basins where most of the sampling and research has been done. The models are designed conservatively; all waste water treatment plants are assumed to discharge their maximum load simultaneously. The best management practices, BMPs, used to manage stormwater and runoff are assumed to be effective only at the low end of their effectiveness range.

In addition, according to Mr. Bell, the model was never designed to be accurate on the local level and this was confirmed in 2008 when the Scientific and Technical Advisory Committee, STAC, Peer Review recommended that the model not be used for local TMDLs. Nonetheless, that is exactly what the EPA did. The models reportedly produce some non-intuitive trends, poorly calibrated results and poor model behavior raising the question of whether the regulatory scheme relies too heavily on the model and will distort desired behavior. They have created a situation where base assumptions are based on the model’s presentation of reality and management practices are not selected for the cost and measured improvement on water quality or ease to maintain, but rather the impact these steps have on model results.