The PSR-FA-NAR model for assessing and forecasting environmental impacts: An empirical analysis of Changning–Weiyuan shale gas play in China

Abstract

The largest ever investment in shale gas resources has induced potential environmental threats in China. The assessment and forecasting of environmental impacts associated with shale gas production is highly challenging due to the characteristics of high uncertainty, nonlinearity, and complexity. This paper proposes a new hybrid model by combining the pressure-state-response (PSR) framework with the firefly algorithm (FA) and a nonlinear auto-regressive (NAR) dynamic neural network (the PSR-FA-NAR model) to detect and forecast the state of the environment as well as send warning signals for shale gas production. Then, an empirical sample, the Changning-Weiyuan national-level shale gas pilot zone that produces more than 50% of Chinese shale gas output, is used to test the effectiveness of the proposed model. The results show that Changning play will predictably face severe environmental threats imposed by rapid development, and the model is not only able to capture nonlinearity time-series and present cause-effect relationships but is also able to improve the predictive performance and forecasting accuracy. It proves that the PSR-FA-NAR model can effectively address the problems with high dimensionality, complexity, and nonlinearity and provides a practical methodology to quantify and identify the potential environmental impacts of unconventional oil and gas production.