Landscape filtering of hydrologic and biogeochemical responses in managed catchments

Abstract

Here, we examine hydrologic and biogeochemical responses of managed catchments regarding the degree of stochastic, nonlinear filtering of hydro-climatic and anthropogenic drivers. We utilize three types of analyses to examine the time-series records of catchment responses: (1) statistical analysis, based on probability distribution functions (pdfs); (2) temporal Lorenz inequality, based on Gini coefficients, and (3) contingency or memory assessment, based on spectral analysis. We present analytical expressions for pdfs and statistical moments for several catchment responses, explicitly linking stochasticity of drivers and catchment filtering properties. We also examine relative temporal inequality of two catchment signals: ΦG = G2/G1, where Gi is the Gini coefficient of each signal. We use spectral analysis to evaluate the relative memory of two signals: Φα = α2/α1, where αi is the slope of spectral power versus frequency of each signal in log–log plots. We use these metrics to evaluate: (1) filtering of precipitation (P) inputs to discharge (Q) outputs from four locations in different moisture regimes in the US, and identify human impacts on catchment hydrology; and (2) dominance of hydrologic variability for nitrate wet deposition flux (Ω) patterns for nine US urban sites, and in the patterns of exported solute loads in intensively managed catchments. Our analyses of long-term monitoring in selected managed catchments suggests that increasing human impacts on landscapes: (1) cause hydrologic and biogeochemical processes to exhibit increasing functional homogeneity; (2) contribute to shifts in memory between catchment drivers and responses; and (3) decrease the temporal inequality of nutrient export dynamics.