A Comparison of Time‐Frequency Signal Processing Methods for Identifying Non‐Perennial Streamflow Events From Streambed Surface Temperature Time Series

Abstract

Nonperennial stream networks constitute over half of total stream lengths worldwide (Datry et al., 2014). However, knowledge of the spatial and temporal variability of non-perennial flows is often limited (Costigan et al., 2016) and stream gauging networks are typically biased toward perennial streams (De Girolamo et al., 2015;Eng et al., 2016). Even where gauges on nonperennial streams exist, often insufficient data are available to identify spatial variations in flows within a catchment. Since the duration and volume of surface flows control both the availability and quality of water for flora and fauna along stream reaches (Datry et al., 2016), as well as the potential for groundwater recharge (Shanafield et al., 2012;Morin et al., 2009;Tooth, 2000), identifying and quantifying nonperennial flow patterns is critical.The need for low-cost technologies that can be widely deployed has spurred several efforts to develop methods of streamflow detection and quantification beyond the traditional gauging station. The availability of inexpensive, easily programmable sensors has led to the development of several types of streamflow

Exploring the feasibility of using the soil temperature to identify preferential and lateral subsurface flows

“…(2001) and Partington et al. (2021), who found a damping effect in temperatures in the soil beneath the riverbed because of rainfall or streamflow events in nonperennial streams.…”

“…(2001) and Partington et al. (2021). These authors found interruptions in the daily soil temperature cycle during water flow events in nonperennial streams comparable to saturated infiltration into an agricultural soil.…”

“…Smaller discontinuities may be corrected by gap‐filling procedures (Groh et al., 2020); however, larger periods without data can only be reconstructed in a predictive way by simulations using a validated model (e.g., suction cup data; Weihermüller et al., 2011). In contrast to tensiometer data, soil temperature is more easily recorded and less prone to sensor failures (Partington et al., 2021). Thus, quasicontinuous time series of soil temperature data could be used to identify deviations from uniform flow patterns by making assumptions for the relationship between soil temperature and water flow.…”

Stream Network Dynamics of Non‐Perennial Rivers: Insights From Integrated Surface‐Subsurface Hydrological Modeling of Two Virtual Catchments

“…Recent improvements in sensor technology gave the opportunity to use high‐resolution photography, remote sensing (Dralle et al., 2023; Shawky et al., 2019; Spence & Mengistu, 2016), and other innovative devices to support the monitoring of the stream length variations. Notable examples include temperature sensors (Arismendi et al., 2017; Blasch et al., 2004; Constantz et al., 2001; Keery et al., 2007; Partington et al., 2021), stage camera systems (Herzog et al., 2022; Kaplan et al., 2019; Noto et al., 2022; Perks et al., 2016; Tauro et al., 2014; Tauro, Olivieri, et al., 2016; Tauro, Petroselli, et al., 2016), and electrical resistance sensors (Chapin et al., 2014; Floriancic et al., 2018; Goulsbra et al., 2014; Jensen et al., 2019; Kaplan et al., 2019; Paillex et al., 2020; Zanetti et al., 2022). All of them proved to be useful for collecting data at high temporal resolution and have the benefit of being cost‐effective and automatic, allowing scientists to observe the sequences of activation/deactivation of the different nodes of the network (Durighetto et al., 2023) and any discontinuity in the wet stream length at the event scale, rather than at the monthly or seasonal scale.…”

Research on modulation recognition method of electromagnetic signal based on wavelet transform convolutional neural network

“…At present, the identification methods based on the signal characteristics of electromagnetic signals mainly use manual methods to extract signal characteristics and realize discrimination [5]. Common electromagnetic signal feature extraction methods include time domain analysis [6], frequency domain analysis, instantaneous autocorrelation, spectral correlation and time-frequency domain analysis. These methods generally transform the sampled signal and extract its features [7], and realize signal recognition through the feature vectors with high separability.…”