Revisiting seasonal dynamics of total nitrogen in reservoirs with a systematic framework for mining data from existing publications. (1st August 2021)

• Record Type: Journal Article
• Authors: Guo, Zhaofeng; Boeing, Wiebke J.; Xu, Yaoyang; Yan, Changzhou; Faghihinia, Maede; Liu, Dong
• Is Part Of: Water research. Volume 201(2021)
• Publication Date: 2021-08-01
• Subjects: Reservoir -- Total nitrogen -- Seasonal dynamics -- Data mining; Water -- Pollution -- Research -- Periodicals; 363.7394
• DOI: 10.1016/j.watres.2021.117380
• Languages: English
• ISSNs: 0043-1354

Abstract

Highlights: A framework is proposed to examine dynamic patterns of water quality parameters. Combination of time-series cluster and GAMLSS model helps in pattern recognition. Seasonal patterns of TN were identified (summer valley, summer peak, spring peak). Improved data accessibility and availability facilitate pattern recognition. Framework can be extended in its applicability for pollution control and management. Abstract: Investigation of seasonal variations of water quality parameters is essential for understanding the mechanisms of structural changes in aquatic ecosystems and their pollution control. Despite the ongoing rise in scientific production on spatiotemporal distribution characteristics of water quality parameters, such as total nitrogen (TN) in reservoirs, attempts to use published data and incorporate them into a large-scale comparison and trends analyses are lacking. Here, we propose a framework of Data extraction, Data grouping and Statistical analysis (DDS) and illustrate application of this DDS framework with the example of TN in reservoirs. Among 1722 publications related to TN in reservoirs, 58 TN time-series data from 19 reservoirs met the analysis requirements and were extracted using the DDS framework. We performed statistical analysis on these time-series data using Dynamic Time Warping (DTW) combined with agglomerative hierarchical clustering as well as Generalized Additive Models for Location, Scale, and Shape (GAMLSS). Three patterns of seasonal TN dynamics were identified. In Pattern V-Sum, TN concentrations change in a "V" shape, dropping to its lowest value in summer; in Pattern P-Sum, TN increases in late summer/early fall before decreasing again; and in Pattern P-Spr, TN peaks in spring. Identified patterns were driven by phytoplankton growth and precipitation (Pattern V-Sum), nitrate wet deposition and agricultural runoff (Pattern P-Sum), and anthropogenic discharges (Pattern P-Spr). Application of the DDS framework has identified a key bottleneck in assessing the dynamics of TN — low data accessibility and availability. Providing an easily accessible data sharing platform and increasing the accessibility and availability of raw data for research will facilitate improvements and expand the applicability of the DDS framework. Identification of additional spatiotemporal patterns of water quality parameters can provide new insights for more comprehensive pollution control and management of aquatic ecosystems. Graphical abstract: Image, graphical abstract