Synthesis Report: Integrating probability and targeted survey designs in regional stream condition assessments with examples from southern coastal California

Ambient water quality monitoring programs have traditionally relied on the use of targeted monitoring designs (in which sampling locations are assigned non-randomly) to assess the condition of aquatic resources. As probability-based surveys (in which site locations are assigned randomly) have become more widely adopted, there has been a growing desire to combine data from targeted sites with data from probability sites when making overall assessments of resource condition. We evaluated the potential for combining datasets and present an approach to combining data from the two types of surveys. We compiled biological, chemical and habitat data from 63 probability sites and 133 targeted monitoring sites in southern coastal California and compared the distribution of analyte values collected under the two designs.

This analysis clearly demonstrates that both targeted and probabilistic sampling designs are necessary for effective non-point source water quality monitoring programs. Probability designs are the only survey type capable of giving an objective overview of the range of conditions present in the resource of interest. By contrast, targeted designs often are needed to fill in critical data gaps and answer site-specific questions. We conclude that the most effective way to ‘integrate’ probability and targeted surveys is to use probability surveys to produce an unbiased regional framework for interpreting targeted results and to supplement this framework as needed with sites targeted to fill data gaps.