Variations on a method for evaluating decadal-scale changes in the groundwater quality of two GAMA coastal study units 2004–14, California GAMA Priority Basin Project

Decadal changes in groundwater quality in two study units on the north-central California coast were evaluated by the Priority Basin Project (PBP) of the Groundwater Ambient Monitoring and Assessment (GAMA) Program. Groundwater samples collected from wells during 2004–05 were compared on a pair-wise basis to samples collected from the same wells during 2014. The data set consisted of paired-samples from 50 public supply wells in 2 GAMA-PBP study units (25 wells each in the North San Francisco Bay and the Monterey Bay and Salinas Valley Basins), with analytical results for 160 water-quality constituents. Statistical analysis was done on grouped results for the 59 constituents that were detected in at least 10 percent of the samples during either sampling period to evaluate decadal-scale change by a step-trend analysis.

The data for both of the sampling periods were processed three different ways, resulting in three variations of the paired results to be submitted for statistical analyses. The first evaluation method variation processed data only to facilitate comparison of data when one or both of the results was a non-detection. The second and third variations applied the additional requirement that differences between initial and decadal-sample results exceed a defined threshold to prevent small differences from supporting the conclusion of a step trend. One method for setting the difference threshold between initial-sampling and resampling results is based on criteria used by the GAMA-PBP to determine whether or not replicate results are acceptable. The other difference-threshold-setting method uses a calculated confidence interval around each result based on demonstrated analytical variability for the constituents during each sampling period. Finally, constituents for which decadal-scale changes were statistically significant were identified using the Wilcoxon-Pratt signed-rank test on each of the three evaluation method variations.

Step trends were identified by at least 1 of the 3 method variations for 33 constituents. After considering other factors, however, such as water-quality context and the results of quality-control samples, it was concluded that decadal changes were meaningful for 14 constituents in at least 1 of the 2 study units. Constituents for which step trends indicated meaningful increases were dissolved oxygen, total nitrogen, nitrate, orthophosphate, calcium, chloride, sulfate, iron, and lithium. Constituents for which step trends indicated meaningful decreases were temperature, arsenic, lead, the isotopic ratio of carbon-13, and tritium.