What is it about?
This study introduces a novel and statistically robust water classification system based on a multidimensional approach. Unlike traditional methods, which often rely on ternary diagrams and face limitations due to data closure issues, our scheme utilizes hybrid log-ratios (hlr) of molar concentrations of eight key ionic species. The system is trained on a dataset of over 46,000 simulated water samples. It uses Linear Discriminant Analysis (LDA) and canonical analysis to classify water into 16 primary types, with the capability of distinguishing up to 256 hybrid water classes.
Featured Image
Photo by David Becker on Unsplash
Why is it important?
Water classification is fundamental for hydrology, environmental science, and resource management. Existing classification schemes, such as the widely used Hill-Piper diagram, suffer from inherent mathematical limitations that distort chemical compositions. Our new method overcomes these problems using log-ratio transformations and advanced statistical techniques, resulting in a more accurate and reliable classification. This advancement enhances water quality monitoring, facilitates the identification of contamination sources, and improves decision-making in water management and environmental assessments.
Perspectives
The integration of AI and machine learning in water classification presents new opportunities for improving environmental data analysis. By leveraging LDA and canonical analysis, our approach introduces a data-driven framework that improves classification accuracy while minimizing errors associated with conventional methods.
Lorena Díaz-González
Universidad Autonoma del Estado de Morelos
Read the Original
This page is a summary of: A statistically coherent robust multidimensional classification scheme for water, The Science of The Total Environment, January 2021, Elsevier,
DOI: 10.1016/j.scitotenv.2020.141704.
You can read the full text:
Contributors
The following have contributed to this page
