Spatially weighted functional clustering of river network data.

Area of Science:

• Environmental Science
• Statistical Modeling
• Geospatial Analysis

Background:

• Spatial covariance is crucial for clustering functional data across geographical areas.
• Traditional methods often use Euclidean distance, which is unsuitable for complex structures like river networks.
• Existing river network covariance models lack extensions for functional data common in environmental studies.

Purpose of the Study:

• To develop a novel method for calculating spatial covariance between functional data points situated along a river network.
• To apply this spatial covariance measure as a weighting factor in functional hierarchical clustering.
• To identify groups of monitoring stations exhibiting similar spatiotemporal characteristics of nitrate pollution on the River Tweed.

Main Methods:

• Developed a method to compute spatial covariance for functional data along river networks.
• Integrated stream distance for covariance estimation within directed networks.
• Employed functional hierarchical clustering, weighting clusters by the novel spatial covariance measure.

Main Results:

• Successfully calculated spatial covariance between functional data from sites on a river network.
• Applied the method to nitrate pollution data from the River Tweed, Scotland.
• Identified distinct groups of monitoring stations with similar spatiotemporal pollution profiles.

Conclusions:

• The developed method effectively incorporates spatial covariance for functional data in river networks.
• This approach enhances the identification of spatiotemporal patterns in environmental monitoring data.
• The findings provide valuable insights into nitrate pollution dynamics along the River Tweed.