Time Series Analysis for Optimizing Leachate Management in Landfills Under Weather Conditions with Sudden Heavy Rain

Abstract

Leachate management in landfill site is a major issue in both environmental conservation and facility operation. With recent climate change, extremely heavy rains have caused landfill sites to exceed their drainage capacity. This could lead to leakage of leachate and serious damage on the surrounding environment. We proposed models to predict leachate volume, leachate electrical conductivity and leachate temperature, then investigated how to control the waste layer conditions and reduce the load on leachate treatment facility. In the models, we set rainfall and temperature as explanatory variables and used Auto-Regressive with eXogenous (ARX) and Gaussian Process Regression (GPR). Under non-linear or unexpected conditions, GPR predicted leachate volume, leachate electrical conductivity, and leachate temperature with higher accuracy and fewer relearing than ARX. GPR having such characteristics was considered relatively suitable for the management of leachate and landfill condition. These results suggest that continuous collection of training data and iterative refinement of the proposed prediction models are essential for practical leachate management under variable weather conditions.