Dataset to train, validate and reconstruct POC over the global ocean for 2009-2013 based on PlankTOM12

The distribution of in situ UVP5 measurements over the period 2009-2013 was used to create synthetic data by sampling a global biogeochemical ocean model PlankTOM12 at UVP5 time and location. The synthetic data set is used to train, validate and test Machine Learning methods to reconstruct particulate organic carbon concentration. These data are part of publication at the GMD.本研究利用2009-2013年间的原位UVP5（Underwater Vision Profiler 5）观测数据的分布特征，通过在UVP5的观测时刻与位置处对全球生物地球化学海洋模式PlankTOM12进行采样，生成合成数据集。该合成数据集被用于训练、验证与测试机器学习方法，以反演颗粒有机碳浓度。上述数据集为发表于《地球科学模型开发》（Geoscientific Model Development, GMD）期刊的学术成果的一部分。