Forest carbon sink in China: Linked drivers and long short-term memory network-based prediction

Abstract

The study contributes to existing literature by proposing a decomposition framework combining production-theoretical decomposition analysis with Gini coefficient, and demonstrating superiority of long short-term memory neural network (LSTM) in predicting forest carbon sink (FCS) in the case of small dataset. We linked FCS with socioeconomic factors ignored by most previous studies (e.g., urbanization, income inequality, technical efficiency and technological change). We then investigated China's forestry sector as a case during 2005–2017. The results showed that FCS and its technical efficiency presented an upward trend during the study period. In terms of drivers, GDP and urbanization were the positive drivers driving the increase in FCS. The dominant driver has changed from urbanization to the urban-rural income inequality in explaining the distribution pattern of residential income for determining FCS. Furthermore, LSTM showed the excellent performance for projecting city-level FCS. We finally provided some policy proposals according to the results. • Income inequality, urbanization and forest carbon sink (FCS) nexus were quantified. • A novel framework combining production-theoretical decomposition analysis with Gini coefficient was proposed. • Long short-term memory neural network was used to predict city-level FCSs.