A century of water quality sampling in rivers of the United States

Long-term records of water quality in rivers are critical to reveal priorities and limitations embedded in nationwide monitoring efforts. Using ~50 million water quality observations in United States rivers across 11 constituents, we show that the annual number of samples increased steadily by four orders of magnitude between 1925 and 2010, followed by a marked decline of ~40% by 2022. Fewer than 1% of ~24 million river segments were ever sampled, with strong proportional emphasis on larger streams. Sampling intensity varies substantially across constituents and states, highlighting the distributed governance and diverging priorities of water quality programs. Sampling is overwhelmingly diurnal and highly seasonal, especially at higher latitudes. Observations per site decrease with power law scaling for all constituents, implying an emergent space-time sampling tradeoff, with more intensive sampling at human-impacted locations. Concentrations of most constituents increase downstream through river networks while their variability decreases. The patterns of monitoring priorities identified here provide context to current water quality in rivers and informs potential improvements to sampling strategies. From 50 million observations across 11 constituents in the United States, sampling increases four orders of magnitude since 1925, then declines 40% since 2010; under 1% of 24 million river segments are ever sampled, biased to large streams with diurnal seasonal patterns, and power law scaling of samples per site.