A mixture model approach to infer land-use influence on point referenced water quality

TL;DR

This paper introduces a spatial mixture model with a latent variable to infer land-use influence on water quality, effectively addressing spatial support and compositional data challenges in environmental analysis.

Contribution

The paper presents a novel mixture model approach that incorporates a latent variable to identify land-use influence on water quality, overcoming spatial and compositional data issues.

Findings

Successfully applied to South Australian water quality data

Improves inference of land-use impact on water quality

Addresses spatial support and compositional data challenges

Abstract

The assessment of water quality across space and time is of considerable interest for both agricultural and public health reasons. The standard method to assess the water quality of a catchment, or a group of catchments, usually involves collecting point measurements of water quality and other additional information such as the date and time of measurements, rainfall amounts, the land-use and soil-type of the catchment and the elevation. Some of this auxiliary information will be point data, measured at the exact location, whereas other such as land-use will be areal data often in a compositional format. Two problems arise if analysts try to incorporate this information into a statistical model in order to predict (for example) the influence of land-use on water quality. First is the spatial change of support problem that arises when using areal data to predict outcomes at point locations. Secondly, the physical process driving water quality is not compositional, rather it is the observation process that provides compositional data. In this paper we present an approach that accounts for these two issues by using a latent variable to identify the land-use that most likely influences water quality. This latent variable is used in a spatial mixture model to help estimate the influence of land-use on water quality. We demonstrate the potential of this approach with data from a water quality research study in the Mount Lofty range, in South Australia.