Determination of hyporheic travel time distributions and other parameters from concurrent conservative and reactive tracer tests by local‐in‐global optimization

Abstract

AbstractThe complexity of hyporheic flow paths requires reach‐scale models of solute transport in streams that are flexible in their representation of the hyporheic passage. We use a model that couples advective‐dispersive in‐stream transport to hyporheic exchange with a shape‐free distribution of hyporheic travel times. The model also accounts for two‐site sorption and transformation of reactive solutes. The coefficients of the model are determined by fitting concurrent stream‐tracer tests of conservative (fluorescein) and reactive (resazurin/resorufin) compounds. The flexibility of the shape‐free models give rise to multiple local minima of the objective function in parameter estimation, thus requiring global‐search algorithms, which is hindered by the large number of parameter values to be estimated. We present a local‐in‐global optimization approach, in which we use a Markov‐Chain Monte Carlo method as global‐search method to estimate a set of in‐stream and hyporheic parameters. Nested therein, we infer the shape‐free distribution of hyporheic travel times by a local Gauss‐Newton method. The overall approach is independent of the initial guess and provides the joint posterior distribution of all parameters. We apply the described local‐in‐global optimization method to recorded tracer breakthrough curves of three consecutive stream sections, and infer section‐wise hydraulic parameter distributions to analyze how hyporheic exchange processes differ between the stream sections.