Water-quality conditions near the confluence of the Snake and Boise Rivers, Canyon County, Idaho

Arsenic, antimony, mercury, and water temperature in streams near Stibnite mining area, central Idaho, 2011–17

“…Multiple linear regression models were developed using continuously monitored specific conductance, streamflow, hydrologic indices, and time/season variables (surrogates, collectively) to estimate continuous concentrations of dissolved arsenic, dissolved antimony, and total mercury at the five monitoring sites. Surrogate models were developed using stepwise linear regression analysis as described in Wood and Etheridge (2011). Stepwise linear regression involves testing a number of explanatory variables to determine which are the best predictors of measured concentrations.…”

“…Unlike Wood and Etheridge (2011), instantaneous values of water-quality parameters were paired with values obtained from discrete sample analytical results, rather than daily values. Predictor variables were assessed for their significance (using a p-value of less than 0.05) in estimating the constituent of interest, and the variance inflaction factor (VIF) was used with a maximum threshold of 4 to detect multicollinearity (problematic correlation between variables) as additional predictor variables were assessed in the regression model.…”

Statistical analysis of the water-quality monitoring program, Upper Klamath Lake, Oregon, and optimization of the program for 2013 and beyond

“…The use of continuous water-quality monitor data (single variables or combinations of variables), such as nutrient or suspended-solids concentrations, as a surrogate for discrete biomass measurements is well established (Rasmussen and others, 2009;Wood and Etheridge, 2011), and chlorophyll a has been shown to adequately represent phytoplankton biomass in Upper Klamath Lake (Kann, 1998). Therefore, continuous monitoring data, primarily phycocyanin fluorescence, DO concentration, and pH, might act as a surrogate for chlorophyll-a concentration.…”

Occurrence and transport of selected constituents in streams near the Stibnite mining area, Central Idaho, 2012–14

“…Functions of time and discharge also were used as predictor variables in surrogate regression models if determined to be significant. Surrogate models were developed using stepwise linear regression analysis as described in Wood and Etheridge (2011) using the U.S. Geological Survey R statistical programming package "GSqwsr" (DeCicco and Corsi, 2014). The functional form of the surrogate models is:…”