As the Chesapeake Bay Program carries out the 1997 Reevaluation of the progress being made to meet our year 2000 goal to reduce nutrient loads to the Bay by 40 percent, there is a lot of talk about "the Chesapeake Bay Model."

There have already been press reports of what the "model" is telling us and what this or any other model can and cannot do. Essentially, while the Chesapeake Bay modeling effort is probably the most advanced to be applied to a body of water anywhere on earth, it is only one tool among many available to us to measure our progress and to test what additional steps, if any, need to be taken.

Let's start with what it isn't. The Bay model is not a physical construct laid out on the ground somewhere, although there was just a scale model of the Chesapeake built by the U.S. Army Corps of Engineers a number of years ago.

You will not find the Bay model in any form you can observe or handle, because it is essentially a set of computer runs - not normal computer runs, but ones that use very complex programs and large capacities that until very recently were only available from the largest supercomputers. Running the Bay model from beginning to end is in the realm of computing rocket trajectories and extended weather forecasts.

The other thing I've found important to understand is that there is no one "Bay Model." It is really a series of linked models, each of which provides results to the next level. The centerpiece of the set is the Watershed Model, which measures all the sources of nutrient pollution in the watershed and provides the total loads into the Bay.

One important input to the watershed model is the Regional Acid Deposition Model, or RADM, which was developed as part of the EPA's Acid Rain Program, but which can be used to estimate nitrogen loads deposited onto land. Nitrogen is a key nutrient, especially in the more saline parts of the Bay. Because the air is a source of nitrogen loadings to the Bay, both directly and off the lands of the watershed, it is important to measure accurately all sources, some of which can be hundreds of miles to the west. RADM gives us those estimates for the Watershed Model, and recently confirmed that 21 percent of nitrogen enters our system from the air.

The Watershed Model then links to the Chesapeake Bay 3-Dimensional Model by providing the loadings. The 3-D Model takes those numbers and applies Bay hydrography to show the results of the loadings in terms of water quality and living resources in the Bay itself. So, depending on how you count them, there are upward of five separate models contained in what is commonly called "the Bay Model."

It is also important to understand that all the models are checked and rechecked against real data and conditions gathered by the Bay Program and the states through extensive monitoring of the Bay and its watershed.

So, where are we on the modeling effort for 1997's Reevaluation? The last time the model was used extensively was 1992, when specific nutrient reduction goals and caps were set for each major tributary of the Bay. Since then, efforts have gone into improving all aspects of modeling for the Bay, including updated versions with better data.

Major improvements have been made to the air and point source (permitted facilities) components. The nonpoint source elements have proven more of a problem; it is difficult to measure the loadings from agricultural and urban runoff for a number of reasons. First, a lot of conservation practices are being carried out by farmers and others as a normal part of improving the productivity of the land, and the extent of this is difficult to measure. Second, the effectiveness of many management practices is difficult to measure in terms of nutrient loadings. Finally, even if effectiveness can be measured onsite, many variables affect actual loadings to streams, including lag time in the ground water.

The Watershed Model is operational and putting out estimates of progress. Attention is now turning to the components related to the Bay itself, through the 3-D Model, so that the progress to date and different options for the future can be tested for results. The updated Bay components will, for the first time, be able to measure the effects of different loadings to the Bay in terms of the living resources - grasses, fish and benthos. Earlier versions measured only nutrients and oxygen levels.

These results will start becoming available for analysis late this summer, in time to help the Chesapeake Executive Council determine what further actions will assure meeting the 2000 goals. They will also assist Virginia citizens in setting the goals for the Commonwealth's tributaries south of the Potomac.

All of this shows why models are good tools for helping us reach our environmental goals. They can measure the success we have had to date. When doing this, they evaluate actions against a normal rainfall year, so our efforts are not masked by natural weather cycles. They can predict our chances of ultimate success in the future by projecting current levels of effort and accounting for increases and decreases we know are likely to occur. Most important, they can test the various options to do more which are available to us, and see if the results show up in significant reductions in nutrient loadings. Finally, we are able to take those loadings changes and estimate how they will affect the health of the Chesapeake and, for the first time, the living resources of the Bay.

At the same time, we need to be careful to understand the limits of models. First and foremost, they are only as good as the data going in. The "input decks" can be the most sophisticated in the world, but if they are full of errors and missing key data sets, the model results will be gibberish, or worse yet, will be wrong but not so obviously so; in that case, we can rely on results which are flawed and not realize it.

Another thing to keep in mind is that because models provide results for an average year, they do not tell us what the Bay actually feels this year. So while the model can be telling us that we should be seeing positive signs of recovery based on lower estimated nutrient loads, storms and abnormally high precipitation levels can mean that the real picture is very different. Finally, because they are based only upon the data that is available, model results require a great deal of time to evaluate and compare to monitoring and other modeling results.

Reaching consensus on what the model results mean and still meeting our timetable to complete the 1997 Reevaluation for the fall Executive Council meeting will be one of the Program's biggest challenges. There are always those who want to make decisions based on what we know. And there are always those who want to delay decisions until we are sure of what we know.

So remember in the coming months, as the numbers start to flow from the model runs, whether the news is good or bad, it is only one tool being applied to a very complicated picture - an extremely valuable tool, but not the last word. Trends analysis of 11 years of monitoring data need to be completed to provide timely comparisons. Further improvements in those input decks of data are also likely to be needed to give us more accurate results. So far, the model is telling us that we will reach the phosphorus goal but will fall short on nitrogen reduction. If this is so, the real value of the Chesapeake Model lies ahead - when we can use it this summer to test the additional actions we need to close the nitrogen gap.

Winston Churchill once said that democracy was the worst possible form of government - except for everything else that had been tried. He might as well have been speaking about computer models; for all they are criticized, no one has yet produced a better way to understand that very complicated system we call the Chesapeake Bay.

Bill Matuszeski is director of the EPA's Chesapeake Bay Program Office.