Search for: All records

Abstract Protecting surface water quality can be complicated by high spatiotemporal variability. Pollutant sources and transport pathways may be identified through sufficiently high‐density monitoring sites and high‐frequency sampling, but practical considerations necessitate tradeoffs between spatial and temporal resolution in water quality monitoring network design. We examined how tradeoffs in sampling density and frequency affect measures of spatiotemporal variability in water quality, emphasizing pattern stability over time. We quantified the spatial stability of stream water quality across >250 monitoring sites in the intensively monitored watershed draining to Lake Okeechobee, FL using Spearman's rank correlations between instantaneous observations and site long‐term means for each parameter. We found that water quality spatial patterns for geogenic, biogenic, and anthropogenic parameters were generally stable on decadal timescales for all solutes, and that sampling densely in space yields more information than sampling frequently in time. Variations in spatial stability decreased with increased sampling density but not with greater sampling frequency, attesting to the dominance of spatial variability over temporal variability. For nutrients, the spatial coefficient of variation (CV) was approximately double the temporal CV. Spatial stability of most solutes was similar across flow conditions, but high‐flow monitoring allows for more sites that effectively capture the long‐term spatial patterns of nutrient sources. Water quality monitoring regimes can be optimized for efficiency in capturing water quality patterns and should be adjusted to focus more on spatial variation. We discuss potential improvements for water quality monitoring, particularly in watersheds where scarce resources necessitate tradeoffs between sampling density and frequency. 

ABSTRACT The deterioration of stream water quality threatens ecosystems and human water security worldwide. Effective risk assessment and mitigation requires spatial and temporal data from water quality monitoring networks (WQMNs). However, it remains challenging to quantify how well current WQMNs capture the spatiotemporal variability of stream water quality, making their evaluation and optimisation an important task for water management. Here, we investigate the spatial and temporal variability of concentrations of three constituents, representing different input pathways: anthropogenic (NO₃⁻), geogenic (Ca²⁺) and biogenic (total organic carbon, TOC) at 1215 stations in three major river basins in Germany. We present a typology to classify each constituent on the basis of magnitude, range and dominance of spatial versus temporal variability. We found that mean measures of spatial variability dominated over those for temporal variability for NO₃⁻and Ca²⁺, while for TOC they were approximately equal. The observed spatiotemporal patterns were robustly explained by a combination of local landscape composition and network‐scale landscape heterogeneity, as well as the degree of spatial auto‐correlation of water quality. Our analysis suggests that river network position systematically influences the inference of spatial variability more than temporal variability. By employing a space–time variance framework, this study provides a step towards optimising WQMNs to create water quality data sets that are balanced in time and space, ultimately improving the efficiency of resource allocation and maximising the value of the information obtained. 

Abstract We present a curated water chemistry data set for lotic systems across the contiguous US containing 35,000,000 records from 290,000 locations. These records are spatially joined to high‐resolution national hydrography data sets, providing information on watershed area, network position, and other hydrographic information. Our curation process follows best practices applied to raw query results from the Water Quality Portal, followed by assigning network context (position and watershed attributes) to each site from the high‐resolution National Hydrography Data set. The ChemLotUS data set currently includes 11 analytes selected to represent geogenic, biogenic, and anthropogenic processes: calcium, conductivity, pH, total suspended solids, turbidity, dissolved oxygen, total organic carbon, chlorophyll a, nitrate, soluble reactive phosphorus, and total phosphorus. All records from the raw query were modified during curation, most notably by removing duplicated observations, converting units, and aggregating strongly correlated chemical forms. Following curation, 65% of the original records were preserved, with significant reductions from raw to curated data in the means of nine constituents and, more notably, in the standard deviations of all constituents. 95% of monitored river reaches were linked to three or fewer monitoring sites, with sample patterns revealing a strong measurement bias to high order streams. We demonstrate the functionality of ChemLotUS by identifying spatiotemporal patterns in water quality at the CONUS‐scale, including diurnal variations of dissolved oxygen, pH in headwaters compared to their corresponding river mouths, and total suspended solids as a function of stream order. ChemLotUS enables new opportunities for investigations of continental scale variation in and controls on water quality.