TDS Prediction with Wavelet Analysis and Trend-Seasonal Decomposition and Machine Learning Algorithms - Case Study: Karkheh River, Iran

Aug 5, 2026

Upload by Sage

Water quality assessment is critical as river resources are increasingly stressed by population growth, climate change driven by human activity, and changing consumption patterns. This study proposes a signal-analysis framework to predict total dissolved solids (TDS) in the Karkheh River, Iran. Continuous wavelet transform decomposes water-quality time series into trend, seasonality, and residual components, from which temporal-dynamics features are extracted. Nonlinear machine learning models (XGBoost, Random Forest, Decision Tree) compare scenarios for feature construction. Results from 50 years of data from three stations reach over 95% accuracy, with signal-driven features improving RMSE by about 30% and supporting scalable real-time monitoring and earlier warning in semi-arid systems.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages17

Reading Time43 min

AccessFree to download

Abstract

Keywords

Introduction

Importance of water quality monitoring

Role of TDS and electrical conductivity (EC)

Regional context and relevance to Iran

Water Conservation Science and Engineering (2025) 10:63 [https://doi.org/10.1007/s41](https://doi.org/10.1007/s41) 101-025-00390-z TDS Prediction with Wavelet Analysis and Trend‑Seasonal Decomposition and Machine Learning Algorithms, Case Study: Karkheh River, Iran Hossein Amini1 · Farshid Fakheri2 · Jaffer Yousuf Dar3 · Reza Shakeri4 · Hossein Nejati4 · ManYue Lam1 · Banafsheh Zahraie4 · Reza Ahmadian1 Received: 14 January 2025 / Revised: 31 May 2025 / Accepted: 17 June 2025 © The Author(s) 2025 Abstract Water quality assessment is definitely important, as the available water resources are highly stressed by population growth, climate change due to anthropogenic activities, and a significant change in consumption patterns. This study aims an innovative framework to predict the total dissolved solids (TDS) with more accuracy in the rivers, case study: “Karkheh River”, Iran, with the integration of signal analysis with machine learning algorithms. First, continuous wavelet transform (CWT) was applied to decompose the time series of water quality variables (e.g., Ca, HCO3, SO4, and Cl) into their trends, seasonality, and residuals, extracting features that capture temporal dynamics. These features served as input for non-linear machine learning models (XGBoost, Random Forest, Decision Tree) in differenct scenarios to compare which way of adding new feature would improve the model performance in terms of the TDS predictions. Adding new features characterized by only TDS signal analysis improved the TDS predictions and was compared with adding all variables signal characterization and compared with only using raw data to predict TDS level. Using a 50-year dataset from three different hydrometric stations, the models could achieve over 95% accuracy, and XGBoost outperformed others in terms of taking the advantage of the new extracted features from signals. The results indicates that signal-driven features significantly contribute to ccurately TDS prediction by 30% improvement in RMSE, and it can offer a scalable approach for real-time water quality monitoring in semi-arid river systems, which leads to a better early warning system for designing future mitigation strategies. Keywords Water quality · Machine learning · TDS · Signal analysis * Hossein Amini [AminiH@cardiff.ac.uk](AminiH@cardiff.ac.uk) Farshid Fakheri[farshid73@aut.ac.ir](farshid73@aut.ac.ir) Jaffer Yousuf Dar [Yousuf.Dar@igb-berlin.de](Yousuf.Dar@igb-berlin.de)[ ](Yousuf.Dar@igb-berlin.de)Reza Shakeri [Shakeri.R@ut.ac.ir](Shakeri.R@ut.ac.ir) Hossein Nejati [Hossein.nejati@ut.ac.ir](Hossein.nejati@ut.ac.ir)[ ](Hossein.nejati@ut.ac.ir)Man Yue Lam [LamM7@cardiff.ac.uk](LamM7@cardiff.ac.uk) Banafsheh Zahraie [bzahraie@ut.ac.ir](bzahraie@ut.ac.ir) Reza Ahmadian [AhmadianR@cardiff.ac.uk](AhmadianR@cardiff.ac.uk) 1 Hydro-Environmental Research Centre, School of Engineering, Cardiff University, Cardiff, UK 2 Department of Civil and Environmental Engineering, Amirkabir University of Technology, Tehran, Iran 3 Department of Experimental Limnology, Leibniz-Institute of Freshwater Ecology and Inland Fisheries, Berlin, Germany 4 School of Civil Engineering, College of Engineering, University of Tehran, Tehran, Iran Introduction Water is a fundamental and highly limited resource, essential for the sustenance of life and the health of ecosystems. Water quality is a pivotal priority for public health, ecosystem sustainability, and economic development, and it faces unprecedented challenges from population growth, climate change, and anthropogenic activities [23 , 36 , 60] . Given the strong and direct correlation between water quality and public health, governments and international organizations have made it a top priority [16] . Globally, the abovementioned challenges in freshwater resources are mainly from agricultural runoff, industrial discharges, and urban wastewater degrading surface and groundwater quality [13, 41]. According to t

FAQs

What is the main goal of this study?

To predict total dissolved solids (TDS) in rivers more accurately by integrating signal analysis methods with machine learning.

How are time-series signals processed before machine learning?

Continuous wavelet transform (CWT) decomposes variables into trend, seasonality, and residual components, generating features that represent temporal dynamics.

Which machine learning model performed best and why does it matter?

XGBoost outperformed other models by leveraging the extracted signal-driven features, yielding over 95% accuracy and about 30% RMSE improvement.

Aug 5, 2026

Upload by Sage