design strata and how they interrelate.
The methodology for determining the number of sites per pesticide use area only ensures that with a given probability that the observed maximum annual average CWS pesticide concentration will exceed the true 95th percentile. This does not ensure that the precision of the point estimates of interest will be suitable for inferential purposes of this study.
Sample Numbers and Sampling
The assumption implicit in the Agency background document is that there is a one-time study where a number is measured and the job is done. One SAP member noted this is technically indefensible. The level of uncertainty in temporal patterns, for example, preclude the choice of a single number (be it 10 or 14 samples per year) as the appropriate number of samples to collect. It was suggested that the problem be approached with an explicit realization that multiple "rounds" of sampling will be necessary and that an adaptive design may be necessary.
It was suggested that technological advances that permit sample compositing might be looked at, as a way to integrate temporal sampling. Various problems were noted (analyte degradation, etc.) but it was agreed this might be reviewed for portions of the study.
Observations On Occurrence
Some of the highest concentrations seen in monitoring programs do not appear to be associated with storm runoff. Rather, they appear to represent spills or dumps of pesticides directly into the stream or river. They are not frequent, fortunately, but one such spill could have a substantial effect on the annual mean concentration. There isn't any realistic sampling strategy that has a good chance of capturing these events, but they must be acknowledged.
DETAILED RESPONSE TO THE CHARGE
The specific issues to be addressed by the Panel are keyed to the Agency's background document "FIFRA Scientific Advisory Panel Briefing Document for a Consultation on Monitoring Strategies for Pesticides in Surface Derived Drinking Water", dated May 8, 2000, and are presented as follows:
1.) EFED is recommending the following factors to identify target pesticides for inclusion in the study: extent of use area, environmental fate properties, and risk.
Is this a reasonable strategy for selecting chemicals for the survey design?
Would some other approach better serve the long-term objective of developing and
testing predictive models?