Hydrological Processes · 2017 · 66 citations · 53 references
Natural ResourcesEarth SciencesHydrologyCatchment HydrologyHydrogeologyWatershed HydrologyLand Cover Effects On WatershedsNutrient RunoffCatchment Nutrient DynamicsCatchment Scale Nutrient DynamicsCoastal Plain EnvironmentsRiver BiogeochemistryRiver Discharge RelationshipsRiver Discharge VariabilityRiverine Nutrient FluxSlope Breaks
Abstract Riverine solute versus discharge (C–Q) relationships provide information about the magnitude and dynamics of material fluxes from landscapes. We analysed long‐term patterns of C–Q relationships for 44 rivers in Florida across a suite of geogenic, nutrient, and organic solutes and investigated land cover, watershed size, and surficial geology as controls on these patterns. Solute concentrations generally exhibited far less variability than did discharge, with coherent solute‐specific behaviours repeated across watersheds. Geogenic solutes generally diluted with increasing discharge, whereas organic solutes generally enriched; patterns for nutrients were highly variable across watersheds, but on average exhibited chemostasis. Despite strong evidence of both geologic and land cover controls on solute flow‐weighted concentrations, these variables were poor predictors of C–Q slopes (β) or relative coefficients of variation (CV C :CV Q ). CV C :CV Q generally increased with watershed size, and wetland area appeared to influence C–Q patterns for base cations and organic solutes. Perhaps most importantly, we observed significant slope breaks in C–Q association in approximately half of our observations, challenging the generality of using single power functions to describe catchment solute export patterns. For all solutes except phosphorus (P), C–Q slopes decreased above statistically identified breaks (slopes for P increased), with breaks consistently at or near median flow (i.e., 50% flow exceedance probability). This common pattern significantly impacts solute load estimates; failing to account for slope breaks overestimates nitrate and total organic carbon loads as much as 125% and underestimates P loads as much as 35%. In addition to challenging generic power‐law characterization of C–Q relationships for these coastal plain rivers, and exploring the load estimate consequences thereof, our study supports emerging insights about watershed hydrochemical behaviours across a wide array of solutes.
53
R: A Language and Environment for Statistical Computing
R Core Team · 2000 · 353K citations · Full text
Natural Sciences, Computer Science, Statistical Computing +10
Power-Law Distributions in Empirical Data
SIAM Review · 2009 · 7.6K citations · Full text
Power-law distributions in empirical data
Aaron Clauset, Cosma Rohilla Shalizi, M. E. J. Newman · Figshare · 2018 · 6.8K citations · Full text
Estimating regression models with unknown break‐points
Vito M. R. Muggeo · Statistics in Medicine · 2003 · 2.3K citations