Predicting DOC Concentration in the Peel River with a Mechanistic Numerical Model

Abstract

Arctic warming is causing increased export of sediments and organic matter via active layer deepening and thermokarst slumps. A mechanistic numerical model was developed using the ReacTran R package to predict riverine dissolved organic carbon (DOC) and total suspended sediment (TSS) concentrations measured during a 2019 field expedition in the Peel River watershed, YT, Canada. In addition to advective transport, two geochemical DOC removal processes were implemented (DOC mineralization and adsorption to mineral surfaces). The power of upstream slump affected area to predict riverine DOC and TSS concentrations was also investigated via a random forest classifier used to identify slump features in the landscape. However, other landscape properties (NDVI, NDMI) proved to be better predictors of riverine DOC and TSS, possibly due to inaccuracies in the classification. Steady state model results indicate that 70–90 % of total DOC input to the river was exported from the downstream boundary unaffected by removal processes, and the 10–30 % of input DOC that was removed was done so predominantly via adsorption to mineral surfaces. Adsorption was driven by high TSS tributaries entering the model domain in its downstream reaches, with the high TSS values possibly due to increased slumping activity in the watersheds of these tributaries. Requisite sensitivity analyses were not performed and offer opportunities for continuation of this work, as does expanding the model to include dynamic inputs and splitting the bulk DOM pool into contributing components.