Integrated data fusion and mining techniques for monitoring total organic carbon concentrations in a lake

Abstract

Monitoring water quality on a near-real-time basis to address water resource management and public health concerns in coupled natural systems and the built environment is by no means an easy task. Total organic carbon (TOC) in surface waters is a known precursor of disinfection by-products in drinking water treatment such as total trihalomethanes (TTHMs), which are a suspected carcinogen and have been related to birth defects if water treatment plants cannot remove them. In this paper, an early warning system using integrated data fusion and mining (IDFM) techniques was proposed to estimate spatiotemporal distributions of TOC on a daily basis for monitoring water quality in a lake that serves as the source of a drinking water treatment plant. Landsat satellite images have high spatial resolution, but such application suffers from a long overpass interval of 16 days. On the other hand, coarse-resolution sensors with frequent revisit times, such as MODIS, are incapable of providing detailed water quality information because of low spatial resolution. This issue can be resolved by using data or sensor fusion techniques, such as IDFM, in which the high-spatial-resolution Landsat and the high-temporal-resolution MODIS images are fused and analysed by a suite of regression models to optimally produce synthetic images with both high spatial and temporal resolution. Analysis of the results using four statistical indices confirmed that the genetic programming model can accurately estimate the spatial and temporal variations of TOC concentrations in a small lake. The model entails a slight bias towards overestimating TOC, and it requires cloud-free input data for the lake. The IDFM efforts lead to the reconstruction of the spatiotemporal TOC distributions in a lake in support of healthy drinking water treatment.