The paper by Turner and co-authors tackles the very timely question on how well we can reconstruct carbon inventories given the sparse observations. Using an Ensemble Optimal Interpolation approach and synthetic observations, the authors show that a large fraction of ocean carbon and its variability can be reconstructed using temperature and salinity measurements in the top 100 meter, however, reconstruction skill decreases when the top 2000 meters are considered. The authors propose a new way to use sparse observations to better understand historical carbon cycle changes, i.e., an important quantity in light of future changes driven by man-made emissions.