48 
xCHLA = DATAFLOW measured chlorophyll a, parameter expression 
used to differentiate it from lab derived chlorophyll a based 
on nutrient samples. 
xCHLA*tributary = interaction term of DATAFLOW measured chlorophyll a 
with tributary system 
xlnSalin = DATAFLOW instrument derived salinity measurement, 
parameter used to differentiate this data from routinely meas¬ 
ured salinity with other instrumentation. 
Questions regarding K d -Turbidity relationships: 
1. Does the 1.5 root transformation that worked well to linearize the K d - 
Turbidity relation for VIMS data work well for MD DNR data? Yes. 
2. Does one K d -Turbidity model work for all tributaries? No. 
3a. Is chla (chlorophyll a) an important predictor? Yes, but contribution is less 
than Turbidity. 
3b. Is chla effect same for all tributaries? No. 
3c. Is it better to use chla or logchla? Chla 
4. Is Salinity a useful predictor? Yes 
5. Is there a seasonal effect? Not much 
6. Can Tributaries be grouped so that calibration terms are uniform within 
group? Yes - the 15 tributaries form 6 groups. 
The following details provide the supporting analyses for the answers to the ques¬ 
tions above: 
1. Does the 1.5 root transformation that worked well to linearize the K d - 
Turbidity relation for VIMS data work well for DNR data? Yes. 
To address this question, a series of linear regression analyses were done use root 
transformations ranging from the 1.1 root to the 2.9 root. R-square and root mean 
square error for this series are reported (Table D-l) as measures of goodness of fit. 
Table D-1. R-square and root mean square error from a series of linear regression 
models where K d is the dependent variable and the independent variables include 
Tributary, root(turbidity), Tributary*root(turbidity), chlorophyll, Tribu- 
tary*chlorophyll. The root transform of turbidity ranges from 1.1 to 2.9. Note: 
“chlorophyll” refers to chlorophyll a measurements. 
appendix d 
Derivation of IQ Regressions 
