ing for air quality, forest resources and water quality, for example. The National Science and Technology Council's efforts will hopefully provide strong directives from the U.S. President's office that advance inter-agency integration. Some success has already been met in coordination and standardization among the several federal agencies which collect geo-special data and prepare interpretive images representing land use and cover. Collectively, they have really been able to produce a ｂｅｔｔｅｒ product at less total cost.
d. Between models and monitoring
The state of the art of the management models has progressed to the point that managers have considerable confidence in their results. The temptation is to ask how the Bay is doing simply by inputting data such as wastewater loadings and area of agricultural lands under nutrient management practices. The deterministic model will produce a result which is clear and unambiguous. In contrast, monitoring results are complex, hard to inter-relate, subject to confounding natural variation, and associated with error terms. As a result progress in the management strategies are currently being reassessed by the Chesapeake Bay Program mainly by model predictions; the ability to use monitoring results to verify these predictions has greatly lagged the sophistication and appeal of the models. Concerted efforts are now underway to use monitoring results more effectively in validating predicted progress toward restoration goals.
e. Between monitoring and research
The way in which new discoveries and emerging scientific understanding are used in design and interpretation of monitoring is an ad hoc and inefficient process. At the same time, routine environmental monitoring can produce very valuable and otherwise unaffordable data for use by the research community. The Scientific and Technical Advisory Committee of the Chesapeake Bay Program is examining ways to improve the connection between research and monitoring. Some of the focal areas relate to the processes through which environmental quality affects living resources, application of research in the interpretation of water quality data from the monitoring of small watersheds, and the extrapolation and interpolation of monitoring results to different scales. At the national level, the National Science and Technology Council's integrated framework for monitoring includes the concept of Index Sites around the nation at which intense, long-term monitoring and research will be conducted.
3. APPLYING NEW TECHNOLOGIES
a. Chesapeake Bay Observing System The Chesapeake Bay Observing System (CBOS) is a network of buoys through out the Bay from which are deployed various sensors of estuarine and weather conditions. These sensors make nearly-continuous measurements and telemeter data to a base station at my Center's Horn Point Laboratory from which they become available in near-real time via the Internet. The sensors measure winds, air temperature and humidity, current speed and direction, water temperature and salinity, and dissolved oxygen. Other sensors of chemical (e.g. nutrient concentrations) and biological (e.g. fluorescence, acoustic signals, etc.) can be added as needed or developed. Some of the buoys (presently four) are permanently deployed in the main stem of the Bay and other smaller buoys (rover buoys) are deployed for shorter periods of time where they are needed (e.g. in the tidal tributaries).
CBOS is producing knowledge of short-term variability and the processes causing observed patterns of water quality that is impossible to deduce from the monthly or fortnightly sampling employed in the existing Bay monitoring program. This short-term variability (e.g. the movement of waters with low dissolved oxygen into shallow environments as a result of cross-bay winds) may be as biologically important as the mean conditions. Also, the results are showing that, as a result of this variability, low-frequency shipboard sampling may sometimes produce poor estimates of mean conditions. Efforts are underway to use the nearcontinuous CBOS data to develop error estimates and temporal interpolation for low frequency sampling.
