OPEN-LSTM: A Global 1º×1º Monthly Ocean Heat Content Dataset from Remote Sensing Data Based on a Long Short-Term Memory (LSTM) Method (1993-2020)

The OPEN-LSTM dataset estimates Ocean Heat Content (OHC) for upper 2000 m over six different depths (0-100，0-300，0-700，0-1000，0-1500，0-2000 m) at a global scale from 1993 to 2020, based on a Long Short-Term Memory (LSTM) method, via multisource remote sensing observations (SSH, SST, and Winds) combined with Argo-gridded OHC as training data. The resolution of OPEN-LSTM is monthly and one degree, and the time span is from 1993 to 2020. The LSTM neural network method considers long temporal dependence of ocean process to reconstruct a new long time-series OHC dataset (1993-2020) and fill the pre-Argo data gaps from satellite remote sensing observations. The OPEN dataset has been cited by the IPCC AR6 report, and adopted by Big Earth Data in Support of the Sustainable Development Goals (2021) report.