LAUSR

Abstract. A spatially explicit cropland distribution time-series dataset is the basis for the accurate assessment of biogeochemical processes in terrestrial ecosystems and their feedback to the climate system; however, this type of dataset is lacking in China. Existing cropland maps have a coarse resolution, are intermittently covered, or the data are inconsistent. We reconstructed a continuously covered cropland distribution dataset in China spanning from 1900 to 2016 by assimilating multiple data sources. In total, national cropland acreage expanded from 77.72 Mha in 1900 to a peak of 151.00 Mha in 1979, but it consistently decreased thereafter to 134.92 Mha in 2016. The cropland was primarily distributed in three historically cultivated plains in China: the Sichuan Plain, the Northern China Plain, and the Northeast China Plain. Cropland abandonment was approximately 43.12 Mha: it was mainly concentrated in the Northern China Plain and the Sichuan Plain and occurred during the 1990–2010 period. Cropland expansion was over 74.37 Mha: it was primarily found in the southeast, northern central, and northeast regions of China and occurred before 1950. In comparison, the national total and spatial distribution of cropland in the Food and Agriculture Organization (FAO) of the United Nations and the History Database of the Global Environment (HYDE) were distorted during the period from 1960 to 1980 due to the biased signal from the Chinese Agricultural Yearbook. We advocate that newly reconstructed cropland data, in which the bias has been corrected, should be used as the updated data for regional and global assessments, such as greenhouse gas emission accounting studies and food production simulations. The cropland dataset is available via an open-data repository (https://doi.org/10.6084/m9.figshare.13356680; Yu et al., 2020).