OCNET: Global Daily Chlorophyll-a Concentration Reconstructions

OCNET global daily Chlorophyll-a products are a reconstructed dataset of ocean surface chlorophyll-a concentration, generated by a convolutional neural network (CNN) model. The model uses environmental variables like sea surface temperature, salinity, photosynthetically active radiation, and sea surface pressure to fill gaps in satellite observations. The dataset provides daily global coverage from 2001 to 2021 at a 0.25-degree spatial resolution.

Use Cases

• Analyzing long-term trends in global phytoplankton biomass based on reconstructed chlorophyll-a time series.
• Validating and improving biogeochemical ocean models using gap-filled environmental data.
• Studying the correlation between phytoplankton distribution and environmental drivers like sea surface temperature and salinity.
• Providing input data for climate studies that require continuous, large-scale ocean surface variables.

Strengths

• Covers a long temporal range from 2001 to 2021, enabling analysis of multi-decadal trends.
• Uses a machine learning model (OCNET CNN) that incorporates multiple environmental variables for improved gap-filling over traditional interpolation methods.
• Provides high spatiotemporal resolution with daily frequency and 0.25-degree global coverage.

Limitations

• Dataset size, row count, and specific column details are not provided across any source, limiting precise technical assessment.
• The reconstruction is focused on open ocean areas; performance in coastal or complex regions may differ.
• Conflicting file format information exists between sources (ASCII vs. netCDF), creating uncertainty for users.

Provenance

Source

Hong Zhongkun, Tsinghua University

Collection Method

Satellite remote sensing data (chlorophyll-a) combined with reanalysis data (SST, SAL, PAR, SSP) processed through a custom convolutional neural network (OCNET) for reconstruction.

Time Range

2001 to 2021

Geography

Global ocean