Comparisons of cropland area from multiple datasets over the past 300 years in the traditional cultivated region of China

Land use/cover change is an important parameter in the climate and ecological simulations. Although they had been widely used in the community, SAGE dataset and HYDE dataset, the two representative global historical land use datasets, were little assessed about their accuracies in regional scale. Here, we carried out some assessments for the traditional cultivated region of China (TCRC) over last 300 years; by comparing SAGE2010 and HYDE (v3.1) with Chinese Historical Cropland Dataset (CHCD). The comparisons were performed at three spatial scales: entire study area, provincial area and 60 km by 60 km grid cell. The results show that (1) the cropland area from SAGE2010 was much more than that from CHCD; moreover, the growth at a rate of 0.51% from 1700 to 1950 and -0.34% after 1950 were also inconsistent with that from CHCD. (2) HYDEdataset (v3.1) was closer to CHCD dataset than SAGE dataset on entire study area. However, the large biases could be detected at provincial scale and 60 km by 60 km grid cell scale. The percent of grid cells having biases greater than 70% (<-70% or >70%) and 90% (<-90% or >90%) accounted for 56%-63% and 40%-45% of the total grid cells respectively while those having biases range from 10% to 10% and from 30% to 30% account for only 5%-6% and 17% of the total grid cells respectively. (3) Using local historical archives to reconstruct historical dataset with high accuracy would be a valuable way to improve the accuracy of climate and ecological simulation.