Peak centiles of chlorpyrifos surface-water concentrations in the NAWQA and NASQAN programs. (1st February 2015)

• Record Type: Journal Article
• Authors: Mosquin, Paul L.; Aldworth, Jeremy; Poletika, Nicholas N.
• Is Part Of: Water research. Volume 69(2015)
• Page Start: 261
• Page End: 273
• Publication Date: 2015-02-01
• Subjects: Chlorpyrifos -- Water quality monitoring -- Water sampling -- Censored data -- Survey statistics; Water -- Pollution -- Research -- Periodicals; 363.7394
• DOI: 10.1016/j.watres.2014.11.020
• Languages: English
• ISSNs: 0043-1354

Abstract

Abstract: We provide upper bound estimates for peak centiles of surface water chlorpyrifos concentration readings within spatial, temporal, and land-use domains of the United States Geological Survey (USGS) National Water-Quality Assessment (NAWQA) and National Stream Quality Accounting Network (NASQAN) programs. These datasets have large overall sample sizes but variable sampling frequencies and, for chlorpyrifos, extremely high levels of non-detections. Point and interval estimates are provided for the 90th, 95th, 99th, and the 99.9th centiles, given sufficient sample size. Overall upper bound estimates for the NAWQA program over the period 1992–2011 for the 90th, 95th, 99th, and 99.9th centiles are <0.005, 0.0066, 0.0214, and 0.0852 ug/L, respectively. The estimation method is based on a survey sampling approach, finding centiles of pooled data across aggregates of site-years. Although the population quantity estimated by a pooled data centile is not the easily interpretable average of population site-year centiles, we provide strong support that it bounds this average by a combination of theory, comparison of NAWQA aggregate and direct estimates, and using modeled populations. Graphical abstract: Highlights: We estimate bounds for upper centiles of chlorpyrifos surface water concentrations. Data are taken from the USGS NAWQA and NASQAN programs during years 1992–2011. Analysis domains of interest are spatial, temporal, and land-use. The methodology is non-parametric, allowing for limited site data and extensive non-detections. We provide both theoretical and empirical support for the bounding relationship.