Accuracy assessment of global historical cropland datasets based on regional reconstructed historical data——A case study in Northeast China

作     者：Li BeiBei Fang XiuQi Ye Yu Zhang XueZhen 

作者机构：Beijing Normal Univ Sch Geog Beijing 100875 Peoples R China Chinese Acad Sci Inst Geog Sci & Nat Resources Res Beijing 100101 Peoples R China 

出 版 物：《Science China Earth Sciences》 (中国科学（地球科学英文版）)

年 卷 期：2010年第53卷第11期

页      面：1689-1699页

核心收录：

学科分类：07[理学] 08[工学] 

基　　金：supported by National Natural Science Foundation of China (Grant Nos.40571165,40901099) Beijing Normal University Independent Research Fund (Grant No.2009SAP-2) National Key Technology R & D Program (Grant No.2007BAC03A11) 

主　　题：LUCC dataset accuracy assessment Northeast China 

摘      要：Historical cropland datasets are fundamental for quantifying the effects of human land use activities on climatic change and the carbon cycle. Two representative global land-use datasets, the Global Land Use Database (termed SAGE dataset) and the Historical Database of the Global Environment (termed HYDE dataset) have been established and used widely. Despite improvement of data quality and methodologies for extracting historical land use information, certain dataset limitations exist that need to be quantified and communicated to users so that they can make informed decisions on whether and how these land-use products should be used. The Cropland data of Northeast China (CNEC) is based on calibrated historical data and a multi-sourced data conversion model, and reconstructs cropland cover change in Northeast China over the last 300 years. Us- ing the CNEC as a reference, we evaluated the accuracy of cropland cover for SAGE and HYDE in Northeast China at spatial scales ranging from the entire Northeast China to provinces and even individual raster grid cells. Neither SAGE nor HYDE reflects real historical land reclamation. Cropland areas in SAGE are overestimated by 20.98 times in 1700 to 1.6 times in 1990. Although HYDE is better, there are significant disagreements in cropland area and distribution between HYDE and CNEC, especially in the 18th and 19th centuries. The proportion of total grid cells whose relative error was greater than 100% was 63.55% in 1700 and 53.27% in 1780. Global cropland dataset errors over Northeast China originate mainly from both the reverse calculation method for historical cropland data based on modern spatial patterns, and modern land-use outputs from satellite data.