Application of high frequency surrogate modeling of instream chloride to improve understanding of chloride mobilization in cottage country, Ontario

Abstract

Estimating Chloride (Cl) concentration and loading in streams is important to the management of salt application especially in cold regions with sub-zero temperature in winter. Traditional Cl monitoring is based on lab analysis of instream samples collected on a weekly or biweekly basis. However, this method is too coarse in temporal resolution to evaluate acute (less than 96 hours; CCME, 2011) water quality. This project examines the relationship between instream specific conductivity (SC) and Cl concentration, builds SC-Cl surrogate models, and evaluates Cl mobilization for Precambrian shield catchments located in south-central Ontario, Canada. Cl concentration in the catchments in the Muskoka watershed monitored by Dorset Environmental Science Center (DESC) since the 1970s, appear to never exceed the Canadian water quality guidelines. However, with various Cl-introducing salt applications on cottages roads (e.g. road salt, dust suppressant), Cl may accumulate in soils and groundwater in and around road crossings, and can be mobilized into streams and could contribute to concern in the future for downstream aquatic systems. For the DESC-monitored catchments, streamflow water chemistry (weekly to biweekly) samples and local meteorological data has been recorded since 1976. Starting in 2011, 10-min interval observations of temperature, precipitation, SC and discharge are available. In this study, a SC-Cl surrogate modeling approach is used to build relationships between instream SC and Cl concentration for four of the catchments. Models are then used to generate high frequency (10-min) instream Cl concentration and loading, improving the temporal resolution of Cl loading estimation compared with traditional estimation based on linear interpolation of weekly or bi-weekly field samples. For one of the four catchments (DE6), the new high frequency instream Cl estimates are used to investigate Cl mobilization by using computational functions of autocorrelation (ACF), cross-correlation (CCF), continuous wavelet transform (CWT) and wavelet transform coherence (WTC). High frequency data results in large differences in the estimation of Cl loading especially for the month of April when large temperature differences between day and night promote daytime melt on roadways. Changes in discharge can play both a concentration effect and dilution effect on instream Cl concentration. While most previous studies focus on study of instream Cl in urban regions, this study demonstrates that application of SC-Cl models are feasible in rural region with low instream Cl concentration.