Systems and methods for modeling water quality

Inventors

Liu, Hongxing • Xu, Min

Assignees

University of Alabama at Birmingham UAB

Abstract

A system, method, device and computer-readable medium for creating an ensemble model of water quality. The ensemble model is generated by determining a set of optimal component models for spectral regions of a body of water, and combining the optimal models. The optimal models can be based on remote sensing data, including satellite imagery. A K-fold partition approach or a global approach can be used to determine the optimal component models, and the optimal component models can be combined through spectral space partition rules to generate an ensemble model of water quality. The ensemble model not only has improved water quality prediction ability, but also has strong spatial and temporal extensibility. The spatial and temporal extensibility of the ensemble model is fundamentally important and desirable for long-term and large-scale remote sensing monitoring and assessment of water quality.

Core Innovation

The invention relates to generating an ensemble model of water quality for a body of water using multispectral images. The method applies spectral space partition rules to identify one or more spectral regions based on optical characteristics of each identified spectral region, and each spectral region is associated with a selected one of a plurality of optimal component empirical models for determining water quality parameters, with different spectral regions associated with different optimal component empirical models.

After associating each spectral region with an appropriate selected optimal component empirical model, the method applies each selected optimal component empirical model to its associated spectral region. The method then applies an ensemble machine learning algorithm to the plurality of optimal component empirical models to generate the ensemble model of water quality for the body of water.

The method also globally calibrates each of the optimal component empirical models using a set of squared residuals determined by comparing a set of empirical measurements with a set of corresponding predictions generated by the optimal component empirical models, performing a summation of the set of squared residuals, and minimizing the summation of the set of squared residuals.

Claims Coverage

Independent claim coverage includes three independent claims: a method, a system, and a non-transitory computer-readable medium. The core inventive features are spectral space partitioning of multispectral images into spectral regions, assigning each spectral region to a selected optimal component empirical model based on optical characteristics, applying component empirical models per spectral region, combining the component empirical models with an ensemble machine learning algorithm, and globally calibrating the component empirical models using squared residual minimization.

Across the independent claims, the core inventive coverage is the end-to-end ensemble modeling framework that partitions spectral space for multispectral images into spectral regions, associates each region with a selected optimal component empirical model based on optical characteristics, applies those component empirical models to their regions, combines them using an ensemble machine learning algorithm to generate an ensemble model of water quality, and globally calibrates each optimal component empirical model using squared residual minimization.

Stated Advantages

Documented Applications