Cropland is the major source of carbon lost to the atmosphere and contribute directly to emissions of greenhouse gases. There is, however, large potential for cropland to reduce its carbon ux to the atmosphere and sequester soil carbon through soil and crop managements. The managements include no-tillage, perennial and/or deep root crops, irrigation, and organic fertilization etc. But these estimations over cropland remain largest uncertain among all other terrestrial biomes. In most models in CMIP5, the cropland is generally treated similarly as grassland without accounting for realistic crop phenology and physiology processes and crop and soil manage- ments. In this study, we will evaluate how well cropland is represented in CMIP5 simulations and how to improve the representations and reduce the uncertainties over cropland. We will compare the modeled biogeochemical variables against multiple observational data including various remote sensing products and in-situ data.