Multitemporal analysis of cropland transition in a climate-sensitive area: a case study of the arid and semiarid region of northwest China

Land-use change is becoming an important anthropogenic force in the global climate system through alteration of the Earth's biogeophysical and biogeochemical processes. Cropland, which provides people with food, is the most variable vegetation land-use type and is affected by natural and anthropic forces and in turn affects the environment and climate system. This paper investigates the temporal-spatial pattern of cropland transition in the arid and semiarid region of northwest China, using remote sensing data for the late 1980s, 1995, 2000, 2005, 2008, and 2010. The aim was to clarify the change intensity and conversion pattern of cropland with a view to identifying the effects of a series of governmental policies and their influence on the climate system. Mathematical methodologies including the use of a transition matrix model, dynamic degree model, area-weighted centroid model, and area percentage were employed to analyze the temporal change in cropland. Meanwhile, a gridded zonal model with 10-km(2) resolution was used to detect the spatial pattern of cropland transition. During the period from the late 1980s to 2010, cropland increased dramatically by 23,182.17 km(2), an increase of 13.61 % relative to the area under cultivation in the late 1980s. Cropland transition accumulated in the western oasis-desert ecotone of the study area while it declined in the eastern farming-pastoral ecotone, leading to the westward movement of the cropland centroid. A net decrease in natural vegetation and unused land along with a net increase in built-up land due to cropland conversion was observed in the monitoring period. The three major driving forces of the cropland transition were population growth, economic development, and land-use management governed by the Grain for Green Program. The climate response to different conversion patterns was simply analyzed. However, quantitative assessment of the effect should be undertaken by employing ecosystem and climate models.