Long-term land cover dynamics (1986-2016) of Northeast China derived from a multi-temporal landsat archive

Abstract

Northeast China is a major grain production area, an ecological important forest area, and the largest old industrial base which is now suffering from economic growth slowdown and brain drain. Accurate and long-term dynamic land cover maps are highly demanded for many regional applications. In this study, we developed a set of continuous annual land cover mapping product at 30 m resolution using multi-temporal Landsat images. The maps in year 2000 and 2015 were tested using another independent validation dataset and the overall accuracies were 80.69% and 88.38%, respectively. The accuracies of the maps were improved by the integration of multi-temporal Landsat images and post-classification strategies. We found a general trend that the total area of land that experienced a change in land cover each year increased over time. The area change of each land cover type is also detected. The area of forests was 3.92 × 105 km2 in 1986, fluctuated under fire disturbance, but declined in a quite high rate over the period of 1989 to 2006, and finally stayed relatively stable in area around 3.58 × 105 km2. The expansion of croplands was the leading land cover change from 1986 to 2000, and then the total area of croplands slightly declined under the Grain to Green Project of China, while shrublands, grasslands and wetlands began to increase. The area of impervious surfaces increased by more than 502% during the last three decades, and about 73% of the new built-up area was converted from croplands. We also demonstrated the our maps could capture the important land cover conversion processes, such as urbanization, forest logging activities, and agricultural expansion.