Influence of various water quality sampling strategies on load estimates for small streams

Abstract

Abstract. Extensive streamflow and water quality data from eight small streams were systematically subsampled to represent various water-quality sampling strategies. The subsampled data were then used to determine the accuracy and precision of annual load estimates generated by means of a regression approach (typically used for big rivers) and to determine the most effective sampling strategy for small streams. Estimation of annual loads by regression was imprecise regardless of the sampling strategy used; for the most effective strategy, median absolute errors were -30% based on the load estimated with an integration method and all available data, if a regression approach is used with daily average streamflow. The most effective sampling strategy depends on the length of the study. For 1-year studies, fixed-period monthly sampling supplemented by storm chasing was the most effective strategy. For studies of 2 or more years, fixed-period semimonthly sampling resulted in not only the least biased but also the most precise loads. Additional high-flow samples, typically collected to help define the relation between high streamflow and high loads, result in imprecise, overestimated annual loads if these samples are consistently collected early in high-flow events.

Trends in water quality of five dairy farming streams in response to adoption of best practice and benefits of long-term monitoring at the catchment scale

“…On average these were AE25% for TN and TP, and up to 50% for SS and were similar to other estimates of uncertainty (e.g. Robertson and Roerish 1999).…”

“…Stream loads leaving four of the catchments were initially calculated using the products of flow-weighted mean concentrations (the sum of simultaneous products of concentration and stream flow divided by the sum of the individual flows) from the monitoring data, and true mean flow for each period from the continuous flow record (Ferguson 1987;Robertson and Roerish 1999). Because continuous flow measurements in Waikakahi Stream began in 2001, load estimates for 1996-1998 were made using individual flow measurements made at each time of sampling, and using the 2001-2011 mean annual flow (Table 1) to approximate the 1996-1998 period.…”

Managing pollutant inputs from pastoral dairy farming to maintain water quality of a lake in a high-rainfall catchment

“…1) to determine whether monthly sampling from one site adequately described stream water quality and enabled accurate estimates of loads and specific yields. Robertson and Roerish (1999) concluded that 'for study durations of two or more years, routine fortnightly sampling is likely to give the least biased and most precise load estimates. Furthermore, additional high-flow samples result in imprecise, overestimated annual loads if samples are consistently collected early in high-flow events'.…”

“…Storm-flow concentrations of SS, TN and TP were recorded and relationships with flow evaluated in order to calculate storm loads (Ferguson 1987;Robertson and Roerish 1999). Loads were calculated using percentile flows to weight the contributions from routine monitoring and from flood flows.…”

“…Average annual stream loads of TN, TP and SS at the catchment outlet were calculated from the products of flow-weighted mean concentrations and mean flow, using all of the routine (fortnightly and monthly) monitoring data for the period June 2004 to June 2011 (Hirsch et al 1982;Ferguson 1987;Robertson and Roerish 1999). Storm-flow concentrations of SS, TN and TP were recorded and relationships with flow evaluated in order to calculate storm loads (Ferguson 1987;Robertson and Roerish 1999).…”