A classification approach to reconstruct local daily drying dynamics at headwater streams

Abstract

Headwater streams (HSs) are generally naturally prone to flow intermittence. These intermittent rivers and ephemeral streams have recently seen a marked increase in interest, especially to assess the impact of drying on aquatic ecosystems. The two objectives of this work are (a) to identify the main drivers of flow intermittence dynamics in HS and (b) to reconstruct local daily drying dynamics. Discrete flow states—“flowing” versus “drying”—are modelled as functions of covariates that include information on climate, hydrology, groundwater, and basin descriptors. Three classifiers to estimate flow states using covariates are tested on four contrasted regions in France: (a) a linear classifier with regularization (LASSO for least absolute shrinkage and selection operator) and two non‐linear non‐parametric classifiers, (b) a one‐hidden‐layer feedforward artificial neural network (ANN) classifier, and (c) a random forest (RF) classifier. The three classifiers are compared with a benchmark classifier (BC) that simply estimates dominant flow state for each month based on observations (without using covariates). The performance assessment over the period 2012–2016 carried out by cross‐validation shows that the three classifiers for flow state based on covariates outperformed the BC. This demonstrates the predictive power of the covariates. ANN is the classifier that globally achieves the best performance to predict the daily drying dynamics whereas both RF and LASSO tend to underestimate the proportion of drying states. The covariates are ranked in terms of relevance for each classifier. The monthly proportion of drying states provided by the discrete observation network has a major importance for the three classifiers ANN, LASSO, and RF. This may reflect the proclivity of a site to flow intermittence. ANN gives higher importance to climatic and hydrological covariates and its non‐linearity allows a greater degree of freedom.

Mediterranean intermittent rivers and ephemeral streams: Challenges in monitoring complexity

“…Arrows depict the links between each compartment and variables within stations is not only costly but may be prohibitive due to active geomorphological processes and restricted accessibility (Beaufort, Carreau, & Sauquet, 2019;Beaufort, Lamouroux, Pella, Datry, & Sauquet, 2018). Our conceptual model shows that climate is the main driver affecting all processes through precipitation, wind action, and temperature fluctuations, which are mediated by catch- Conceptual framework compartmentalizing the natural and human-induced exogenous and indigenous variables that drive Mediterranean intermittent rivers and ephemeral streams.…”

“…The complexity of Mediterranean IRES arises not only from the number and variability of the controlling variables but also from the interactions between them. Our conceptual model shows that climate is the main driver affecting all processes through precipitation, wind action, and temperature fluctuations, which are mediated by catch- stations is not only costly but may be prohibitive due to active geomorphological processes and restricted accessibility (Beaufort, Carreau, & Sauquet, 2019;Beaufort, Lamouroux, Pella, Datry, & Sauquet, 2018). On the ground wet and dry mapping is sometimes used to compensate the absence of gauges (Datry, Fritz, & Leigh, 2016;Turner & Richter, 2011).…”

Aqua temporaria incognita

“…Background elevation data from https://land.copernicus.eu/imagery-in-situ/eu-dem/eu-dem-v1-0-andderived-products. The stream network was obtained from http://www.sandre.eaufrance.fr/ Hydrological models (Gutiérrez-Jurado, Partington, Batelaan, Cook, & Shanafield, 2019;Ward, Schmadel, & Wondzell, 2018;Williamson, Agouridis, Barton, Villines, & Lant, 2015;Yu, Bond, Bunn, Xu, & Kennard, 2018), topographic data (Prancevic & Kirchner, 2019) and statistical approaches (Beaufort, Carreau, & Sauquet, 2019;Durighetto, Vingiani, Bertassello, Camporese, & Botter, 2020;González-Ferreras & Barquín, 2017;Jaeger, Sando, et al, 2019;Konrad & Rumsey, 2019;Russell, Gale, Muñoz, Dorney, & Rubino, 2015;Snelder et al, 2013) have been used to predict where streams are temporary and can be used to determine where additional data on the state of temporary streams may be most useful. However, to train and validate these models, more observations of the state of temporary streams and stream network dynamics are needed.…”

Reconstructing Spatiotemporal Dynamics in Hydrological State Along Intermittent Rivers

“…Previous research demonstrated the potential of parametric approaches for wet/dry mapping [15], flow permanence simulation [16] and estimating probability of intermittence [17]. Furthermore, neural networks and random forest approaches have been demonstrated to be well suited for statistical simulation of intermittence [18]. Other recent studies have also taken a random forest approach to simulating intermittence [19][20][21], and some have compared approaches including neural networks, regression trees and multivariate adaptive regression splines [18,22].…”