Relationships between stream nitrate concentration and spatially distributed snowmelt in high-elevation catchments of the western U.S.

Abstract

This study compares stream nitrate (NO3-) concentrations to spatially distributed snowmelt in two alpine catchments, the Green Lakes Valley, Colorado (GLV4) and Tokopah Basin, California (TOK). A snow water equivalent reconstruction model and Landsat 5 and 7 snow cover data were used to estimate daily snowmelt at 30 m spatial resolution in order to derive indices of new snowmelt areas (NSAs). Estimates of NSA were then used to explain the NO3- flushing behavior for each basin over a 12 year period (1996-2007). To identify the optimal method for defining NSAs and elucidate mechanisms underlying catchment NO3- flushing, we conducted a series of regression analyses using multiple thresholds of snowmelt based on temporal and volumetric metrics. NSA indices defined by volume of snowmelt (e.g., snowmelt ≤ 30 cm) rather than snowmelt duration (e.g., snowmelt ≤ 9 days) were the best predictors of stream NO3- concentrations. The NSA indices were better correlated with stream NO3- concentration in TOK (average R2= 0.68) versus GLV4 (average R2= 0.44). Positive relationships between NSA and stream NO3- concentration were observed in TOK with peak stream NO3- concentration occurring on the rising limb of snowmelt. Positive and negative relationships between NSA and stream NO3- concentration were found in GLV4 with peak stream NO3- concentration occurring as NSA expands. Consistent with previous works, the contrasting NO3- flushing behavior suggests that streamflow in TOK was primarily influenced by overland flow and shallow subsurface flow, whereas GLV4 appeared to be more strongly influenced by deeper subsurface flow paths.