A-3 
covariance. Further, trials of spatial modeling on fixed station Chesapeake Bay water 
quality data by Dr.s Christman and Curriero were conducted to begin to evaluate 
spatial modeling procedures. These exercises, literature review and discussions 
leading to consensus opinion are the basis of our findings. In August 2006, the 
working group supplied preliminary findings and related text for use in the 2006 
CBP Addendum to Ambient Water Quality Criteria that is now under review. 
FINDINGS 
1. The CFD approach is feasible and efficient in representing water quality 
attainment. 
The CFD approach can effectively represent the spatial and temporal dimensions 
of water quality data to support inferences on whether regions within the Chesa¬ 
peake Bay attain or exceed water quality standards. The CFD approach is 
innovative but could support general application in water quality attainment 
assessments in the Chesapeake Bay and elsewhere. The CFD approach meshes 
well within the Chesapeake Bay Program’s monitoring and assessment 
approaches, which have important conceptual underpinnings (e.g., segments 
defined by designated uses). 
In accepting the CFD as the best available approach for using time-space data, the 
panel contrasted it with the previous method and those sustained by other juris¬ 
dictions. The previous method used by the Chesapeake Bay Program, similar to 
the approaches used in other states, was simply based on EPA assessment guid¬ 
ance in which all samples in a given spatial area were compiled and attainment 
was assumed as long as > 10% of the samples did not exceed the standard. In this 
past approach all samples were assumed to be fully representative of the specified 
space and time and were simply combined as if they were random samples from 
a uniform population. This approach was necessary at the time because the tech¬ 
nology was not available for a more rigorous approach. But it neglected spatial 
and temporal patterns that are known to exist in the standards measures. The CFD 
approach was designed to better characterize those spatial and temporal patterns 
and weight samples according to the amount of space or time that they actually 
represent. 
2. CFD curves are influenced by sampling density and spatial and temporal 
covariance. These effects merit additional research. Conditional simulation 
offers a productive means to further discover underlying statistical proper¬ 
ties and to construct confidence bounds on CFD curves, but further directed 
analyses are needed to test the feasibility of this modeling approach. 
The panel finds that the CFD approach in its current form is feasible, but that 
additional research is needed to further refine and strengthen it as a statistical tool. 
The CFD builds on important statistical theory related to the cumulative distribu¬ 
tion function and as such, its statistical properties can be simulated and deduced. 
Through conditional simulation exercises, we have also shown that it is feasible 
to construct confidence ellipses that support inferences related to threshold curves 
appendix a 
Fhe Cumulative Frequency Diagram Method for Determining Water Quality Attainment 
