Time series modeling of methane gas in underground minesTime series modeling of methane gas in underground mines Peer-Reviewed Journal Publication Δημοσίευση σε Περιοδικό με Κριτές 2024-02-232022enMethane gas is emitted during both underground and surface coal mining. Underground coal mines need to monitor methane gas emissions to ensure adequate ventilation is provided to guarantee that methane concentrations remain low under different production and environmental conditions. Prediction of methane concentrations in underground mines can also contribute towards the successful management of methane gas emissions. The main objective of this research is to develop a forecasting methodology for methane gas emissions based on time series analysis. Methane time series data were retrieved from atmospheric monitoring systems (AMS) of three underground coal mines in the USA. The AMS data were stored and pre-processed using an Atmospheric Monitoring Analysis and Database Management system. Furthermore, different statistical dependence measures such as cross-correlation, autocorrelation, cross-covariance, and variograms were implemented to investigate the potential autocorrelations of methane gas as well as its association with auxiliary variables (barometric pressure and coal production). The autoregressive integrated moving average (ARIMA) time series model which is based on auto-correlations of the methane gas is investigated. It is established that ARIMA used in the one-step-ahead forecasting mode provides accurate estimates that match the direction (increase/decrease) of the methane gas emission data.This study was sponsored by the Alpha Foundation for the Improvement of Mine Safety and Health, Inc. (ALPHA FOUNDATION), contract number AFCTG20-103.http://creativecommons.org/licenses/by/4.0/Mining, metallurgy & exploration.3951961–1982 Diaz Martinez Juan Agioutantis Zacharias Αγιουταντης Ζαχαριας Christopoulos Dionysios Χριστοπουλος Διονυσιος Schafrik Steven Luxbacher Kray Springer Methane gas concentration Atmospheric monitoring systems Time series modeling ARIMA forecasting