The King County Conundrum: Spatial-Temporal Stream Water Quality Trends with Increasing Urban Development

Abstract

King County, Washington State, has monitored stream and river water quality monthly at 75 sites for over 40 years for an average of > 23,000 samples collected for each constituent. This sampling was conducted during a time when the population has almost doubled, and conventional wisdom predicts a decline in water quality due to increased urban development. The database includes nitrate, ammonium, and total nitrogen (N), phosphate and total phosphorus (P), alkalinity, pH, conductivity, dissolved oxygen (DO), turbidity, total dissolved solids (TSS), and fecal coliform bacteria (FCB). I quantified long-term trends, contemporary statistical relationships between catchment land-cover and water quality, and seasonal variability in constituent concentrations. The long-term trend analysis showed nitrate-N, ammonium-N, total N, phosphate-P, FCB, TSS, and DO concentrations significantly declined (p < 0.01) in King County streams during the sampling period. Conversely, alkalinity, conductivity, and stream temperature significantly increased. Turbidity, pH, and total P showed no clear temporal trends. Stream water quality data from 2016-2022 was fit to catchment land-cover regression models. Conductivity (r2 = 0.56), phosphate-P (r2 = 0.55), alkalinity (r2 = 0.53), FCB (r2 = 0.48), total P (r2 = 0.44), temperature (r2 = 0.39), total-N (r2 = 0.37), nitrate-N (r2 = 0.26), DO (r2 = 0.24), and pH (r2 = 0.21) multiple-regression models had significant fits (p < 0.01) with land-cover characteristics. Turbidity, total suspended solids, and ammonium-N had insignificant model fits. Developed land was the most important land-cover category for every constituent. Deciduous forest cover was an important secondary variable for total-N, nitrate-N, conductivity, and alkalinity. Agricultural land-cover (largely pasture) was an important secondary variable for the phosphate-P, total-P, and FCB models. The temperature and pH models had no important second variables. Wetlands were an important secondary variable for DO. In general, developed, deciduous forest, and agricultural landcover resulted in higher constituent concentrations, whereas wetlands and open water usually resulted in lower concentrations. The results of this study are paradoxical because the expected stream water quality trends with urban development were observed, whereas dissolved and total N, dissolved P, suspended sediment and fecal bacteria concentrations declined during a multi-decadal period when the population of King County nearly doubled.