The logistic regression approach could be also be adapted to assess attainment of the 
7-day and 1-day mean criteria components as well as other duration-specific criteria, 
where and when a body of observational data is available at frequencies relevant to 
the time frame. High frequency ‘buoy’ data sited at sentinel locations, where contin¬ 
uous records extend over days, weeks and months, would offer opportunities to 
develop logistic models of the relationship between exceedance/attainment and the 
temporal means. EPA recommends that this method be actively developed for 
possible employment for attainment assessments of the instantaneous minimum 
dissolved oxygen criteria (see next section for details) while additional high 
frequency data are collected and more complex, detailed methods described below 
are being developed. 
SPECTRAL ANALYSIS METHOD 
The time series approach utilizes a statistical procedure known as spectral analysis 
to synthesize a complete record of dissolved oxygen concentrations at short interval 
time steps over time. The synthetic record is developed using continuous measure¬ 
ment data from nearby locations to develop a model that predicts the short-interval 
variations in concentration. That model is combined with the long-term pattern of 
variability derived from data collected routinely, monthly to twice monthly, at the 
fixed-stations located in the assessment unit. The synthetic dissolved oxygen record 
can then be used in the same way that data collected using a continuous meter would 
be used. This time series approach has only been applied in a limited way to date and 
further development is needed in order for it to fully meet the needs of a publishable 
Chesapeake Bay dissolved oxygen criteria assessment methodology (see pages 183- 
185 in U.S. EPA 2003). EPA recommends that this development work proceed 
simultaneously with the development of the logistic regression and that the spectral 
analysis method replace the logistic regression in the future should it prove a more 
robust method. 
COLLECTION OF CONTINUOUS MEASURES OF DISSOLVED 
OXYGEN CONCENTRATION 
The most rigorous approach for assessing attainment of the high frequency dissolved 
oxygen criteria would be to collect continuous measures of dissolved oxygen 
concentration at representative locations and depths throughout each spatial assess¬ 
ment unit. The temporal and spatial density of such data would need to be sufficient 
to enable all of the dissolved oxygen criteria to be assessed simply by calculating 
means at the appropriate time scales (e.g. 30-day, 7-day, 1-day) or by observing 
violations of the instantaneous minimum criteria values. However, continuous 
collection of high frequency dissolved oxygen concentration in the Bay is expensive 
both in purchasing the equipment and maintaining it. It is also difficult or impossible 
to find sufficiently representative locations where the equipment can be affixed to 
buoys or fixed pilings. Finally, it is expensive and labor-intensive to maintain the 
equipment and sensor calibration once it is deployed due to the effects of weather, 
turbulence, biological fouling and human interferences (e.g. accidents, thefts). 
Nevertheless, the collection of at least some continuous dissolved oxygen data will 
be critical for use in the other two statistical analysis-based assessment methods 
appendix e 
Potential Methods for Assessing Shorter Duration Dissolved Oxygen Criteria 
