Development of statistical regression and artificial neural network models for estimating nitrogen, phosphorus, COD, and suspended solid concentrations in eutrophic rivers using UV–Vis spectroscopy

Environ Monit Assess (2023) 195:1114 https://doi.org/10.1007/s10661-023-11738-0 RESEARCH Development of statistical regression and artificial neural network models for estimating nitrogen, phosphorus, COD, and suspended solid concentrations in eutrophic rivers using UV–Vis spectroscopy Yanping Lyu · Wenpeng Zhao · Tsuyoshi Kinouchi · Tadahiro Nagano · Shigeo Tanaka Received: 8 June 2023 / Accepted: 17 August 2023 © The Author(s) 2023 Abstract River water quality monitoring is crucial for understanding water dynamics and formulating policies to conserve the water environment. In situ ultraviolet–visible (UV–Vis) spectrometry holds great potential for real-time monitoring of multiple water quality parameters. However, establishing a reliable methodology to link absorption spectra to specific water quality parameters remains challenging, particularly for eutrophic rivers under various flow and water quality conditions. To address this, a framework integrating desktop and in situ UV–Vis spectrometers was developed to establish reliable conversion models. The absorption spectra obtained from a desktop W. Zhao College of Hydraulic Science and Engineering, Yangzhou University, Yangzhou 225009, China spectrometer were utilized to create models for estimating nitrate-nitrogen ( NO3-N), total nitrogen (TN), chemical oxygen demand (COD), total phosphorus (TP), and suspended solids (SS). We validated these models using the absorption spectra obtained from an in situ spectrometer. Partial least squares regression (PLSR) employing selected wavelengths and principal component regression (PCR) employing all wavelengths demonstrated high accuracy in estimating NO3-N and COD, respectively. The artificial neural network (ANN) was proved suitable for predicting TN in stream water with low N H4-N concentration using all wavelengths. Due to the dominance of photo-responsive phosphorus species adsorbed onto suspended solids, PLSR and PCR methods utilizing all wavelengths effectively estimated TP and SS, respectively. The determination coefficients (R2) of all the calibrated models exceeded 0.6, and most of the normalized root mean square errors (NRMSEs) were within 0.4. Our approach shows excellent efficiency and potential in establishing reliable models monitoring nitrogen, phosphorus, COD, and SS simultaneously. This approach eliminates the need for time-consuming and uncertain in situ absorption spectrum measurements during model setup, which may be affected by fluctuating natural and anthropogenic environmental conditions. T. Nagano · S. Tanaka Civil Engineering and Eco-Technology Consultants Co., Ltd, 2‑23‑2 Higashi‑Ikebukuro, Toshima‑Ku, Tokyo 170‑0013, Japan Keywords Water quality monitoring · In situ UV– vis spectroscopy · Statistical regression models · Artificial neural network · Wavelength selection Supplementary Information The online version contains supplementary material available at https://doi. org/10.1007/s10661-023-11738-0. Y. Lyu · W. Zhao (*) · T. Kinouchi (*) Department of Transdisciplinary Science and Engineering, Tokyo Institute of Technology, 4259 Nagatsuta‑Cho, Midori‑Ku, Yokohama, Kanagawa 226‑8503, Japan e-mail: T. Kinouchi e-mail: Vol.: (0123456789) 13 1114 Page 2 of 16 Introduction With the development of agriculture, industrialization, and urbanization, the deterioration of water quality in the natural environment is becoming a severe regional and global issue due to the increased loading of nutrients, organic carbon, and other toxic substances (de Waal et al., 2022; Ho et al., 2019; Rashid & Romshoo, 2013). Climate change will influence the water quality dynamics and ecosystem by increasing carbon dioxide concentration, air temperatures, more intense precipitations, etc. (Alexander et al., 2013). Water quality issues have been at the forefront of Sustainable Development Goals (SDGs), in which Goal 6 specifically aims to “ensure availability and sustainable management of water and sanitation for all” (UNESCO, 2015). Thus, water quality monitoring is increasingly important for sustainably managing water resources and maintaining ecosystem stability. Water quality monitoring is a prerequisite for understanding the factors that drive the deterioration and for establishing water quality conservation and improvement goals. Traditional methods, such as manual water sampling and laboratory analysis of individual constituents in the samples, provide accurate data but are labor-intensive and inefficient, especially for frequent monitoring with short intervals. To address this, automatic instruments based on chemical reactions have been developed for real-time monitoring the water quality of streams, lakes, and wastewater (Bodini et al., 2018; Fang et al., 2019, 2022). However, these instruments generally require high maintenance costs and consume many chemicals. In the field of analytical chemistry and geochemistry, ultraviolet–visible (UV–Vis) sensors have emerged as efficient tools for analyzing soluble inorganic salts and organic compounds in water by recording the spectral absorbance (Birdwell & Engel, 2010; Willard et al., 1988). For operational purposes in the field rather than in a laboratory, the in situ UV–Vis spectroscopic technology has been developed rapidly and applied for continuous monitoring of the water environment under different hydraulic conditions (Pesántez et al., 2021; Zhang et al., 2022). For wastewater monitoring purposes, in situ UV–Vis spectrometer has been successfully applied to measure various parameters, including suspended solids (SS) and chemical oxygen demand (COD) (Brito et al., 2014; Langergraber et al., 2003). Several studies have explored the use of Vol:. (1234567890) 13 Environ Monit Assess (2023) 195:1114 in situ UV–Vis spectrometers for monitoring nitrogen, carbon, phosphorus, and other components in different environments, such as springs and tidal marshes, by applying statistical models (Huebsch et al., 2015; Etheridge et al., 2014). However, there is limited research on their application in water bodies with diverse biogeochemical compositions, especially eutrophic rivers containing pollutants from farmlands, forests, and urban areas. Monitoring pollutant concentrations to obtain the loading from eutrophic rivers is increasingly important due to the nutrient enrichment and accelerating eutrophication in the receiving lakes (Ho et al., 2019; Izmailova & Rumyantsev, 2016; Smith et al., 1999). The greatest challenge in applying in situ UV–Vis technology for monitoring stream water quality is establishing robust relationships between specific water quality parameters and UV–Vis absorption spectra under the conditions with complicated biogeochemical constituents in the stream. Interference from compounds with distinctive absorbance can lead to overestimated water quality parameters (Uusheimo et al., 2017). Moreover, high concentrations of suspended particles, particularly during flood events, significantly alt (...truncated)