A-21 
down with each monitoring cruise, the designated use of each grid cell must also be 
defined based on the available data for each cruise. 
The pycnocline is defined by the water density gradient over depth. Temperature and 
salinity are used to calculate density, which in turn is used to calculate pycnocline 
boundaries. Density is calculated using the method described in: Algorithms for 
Computation of Fundamental Properties of Seawater (Endorsed by 
UNESCO/SCOR/ ICES/IAPSO Joint Panel on Oceanographic Tables and Standards 
and SCOR Working Group 51. Fofonoff, N P; Millard, R C Jr. UNESCO technical 
papers in marine science. Paris , no. 44, pp. 53. 1983). For each column of temper¬ 
ature and salinity data, the existence of the upper and lower pycnocline boundary is 
determined by looking for the shallowest robust vertical change in density of 0.1 
kg/m3/m for the upper boundary and deepest change of 0.2 kg/m3/m for the lower 
boundary. To be considered robust, the density gradient must not reverse direction at 
the next measurement and must be accompanied by a change in salinity, not just 
temperature. 
The depths to the upper pycnocline boundary, where detected, and the fraction of the 
water column below the lower boundary are interpolated in two dimensions. If no 
lower boundary was detected the fraction was considered to be zero. The depth to the 
upper pycnocline boundary tends to be stable across horizontal space and so spatial 
definition of that boundary using interpolation generally worked well. However, 
interpolation of the lower boundary is more complicated because the results can 
conflict with the upper boundary definition or with the actual bathymetry of the Bay. 
As a result, interpolation of the lower boundary was performed based on “fraction of 
water column depth". In that way, the constraints of the upper pycnocline boundary 
definition and the actual depth were imposed and errors related to boundary conflicts 
were eliminated. 
Assessments were performed based on criteria specific averaging periods. The 
instantaneous assessment for deep channel dissolved oxygen was evaluated using the 
individual cruise interpolations. All monthly assessments were based on monthly 
averages of interpolated data sets. To calculate the monthly averages, each interpo¬ 
lated cruise within a month was averaged on a point-by-point basis. Generally, there 
were 2 cruises per month in the wanner months and 1 cruise per month in the cooler 
months. Spatial violation rates are calculated for each temporally aggregated inter¬ 
polation in an assessment period. For example, for a three-year summer open-water 
dissolved oxygen assessment, the twelve monthly average interpolations repre¬ 
senting the four summer months over three years were used. 
3. PROTOCOL FOR INTERPOLATING WATER QUALITY 
The CFD approach uses the proportion of space in attainment in any given month 
estimated using an approach based on a statistical model. The current method uses 
data collected in a specific month at a set of sampling locations within the segment 
of interest to estimate the parameters of the model. The estimated model is then used 
to interpolate likely values at unsampled locations, specifically at a set of prediction 
locations arranged in a grid over the segment. The predictions thus obtained are used 
appendix a 
The Cumulative Frequency Diagram Method for Determining Water Quality Attainment 
