SST: relevant recent literature #1
Description
According to the proposal, the goal is to develop a model to predict "uncertainty-quantified adjustments"; paired datasets of “observations-like” and “climate-model-like” versions of ERA5 SST and WV will be used. Machine learning (ML) methods will be used to “learn” the adjustments and the uncertainties.
Here are some relevant studies that I found. The first two are the most relevant ones. As a starting point, we can explore their method and code.
- MAESSTRO: Masked Autoencoders for Sea Surface Temperature Reconstruction under Occlusion- 2024: used masked autoencoders (MAE) to fill cloud-caused gaps (spatially) in high-resolution (1 km) SST maps; compared with spatial Kriging interpolation methods. Code available.
- Machine learning methods to predict sea surface temperature and marine heatwave occurrence: a case study of the Mediterranean Sea-2024: applied ML methods (RF, LSTM and CNN) to predict SST time series and extremes in regional seas using ESA CCi SST and ERA5. Code available
- Learning Sea Surface Height Interpolation From Multi-Variate Simulated Satellite Observations-2024: used multivariate satellite + model data to learn interpolation of ocean fields (SSH, SST) in time using CNN. Code available.
- Enhancing the Resolution of Satellite Ocean Data Using Discretized Satellite Gridding Neural Networks-2024: a ML network (NN) to process irregular satellite samples into gridded higher-resolution SST fields. The data and code used in this study are available upon request to the corresponding author.
- Efficient Data-Driven Gap Filling of Satellite Image Time Series Using Deep Neural Networks with Partial Convolutions-2024: Uses 3-D spatio-temporal partial convolutions (i.e., convolutional layers that operate over space and time) to reconstruct image time series with realistic temporal continuity. They apply a U-Net-like model on incomplete image time series of quasi-global carbon monoxide observations from the Sentinel-5 Precursor (Sentinel-5P) satellite. Code available