Prediction of land-use change and its driving forces in an ecological restoration watershed of the Loess hilly region

Land-use change is one of the important topics of regional ecological restoration research, but it exhibits high interdependencies in social-ecological systems which make it difficult to predict spatiotemporal variability and identify the main drivers. We employ long-term dynamic predictions utilizing an integrated cellular automata–Markov (CA–Markov) model and geographic information system technology, and reveal trends of land-use change and the driving forces in an ecological restoration watershed of the Loess hilly region. Here, we show that dry land showed a rising transfer trend to shrub land, sparse forestland, middle-coverage grassland, low-coverage grassland, and rural settlements from 1995 to 2010. Dry land, forestland, sparse forestland, middle-coverage grassland, and rural settlements will maintain a slight decreasing trend from 2010 to 2020, but other forestland, low-coverage grassland, and reservoir or pond may have a gradual increasing trend. Farmland and forestland in 2050 decrease by 8.17% and increase by 46.3%, respectively, compared with that in 1980. Grassland shows an overall downward trend. Water area increases first and then decreases. Rural settlements increase rapidly with a growth rate of 88.8% from 1980 to 2050. The returning farmland policy and the decline of agricultural population may be the main driving forces and positively affect the transfer of dry land to forestland. Our results may provide underlying insights needed to guide the planning and management of regional land resources.