Abstract:Sea surface temperature (SST) is a crucial indicator of heat exchange between the ocean and atmosphere.As one of the most important ocean environmental parameters describing the thermal state of the ocean surface,it is widely used in research and applications such as upper ocean processes,air-sea heat exchange,numerical simulation and forecasting of the ocean-atmosphere system.In-situ observation is one of the most direct and accurate ways to obtain SST,primarily through conventional observation systems such as offshore buoys,coastal stations,and ships.In-situ SST observations serve as the foundation for all other sea temperature data products.Whether it is gridded SST data products,satellite-retrieved SST products,multi-source merged products,or reanalysis data,these SST data products all rely on ship and buoy observation data as the basic support.
Therefore,focusing on ships and buoys as the two primary observation methods for SST,this work collected SST observation data from multiple sources such as ICOADS,GTS,CFSR_OBS,GDAS_OBS,and offshore China.Through decoding,extraction,duplicate checking,standardization,and other steps,a relatively complete and long-term sequence of global SST observation dataset was integrated.Besides the traditional quality control techniques for gross errors,the quality control scheme developed for this dataset also includes a technical approach leveraging model analysis fields for observation data quality control.Specifically,using ERA5 reanalysis data,a unified quality control process was applied to the data.The quality control scheme formulated can remove large variations or anomalies in SST.The final products,the “Global Ship-Based SST Observation Dataset from 1900 to 2023” and the “Global Buoy-Based SST Observation Dataset from 1976 to 2023,” can provide fundamental data support for subsequent SST data evaluation,multi-source SST data merging,and global climate change analysis.This paper further evaluates the data characteristic differences,and the main conclusions are as follows:
After the 1990s,the number of buoy observation records far exceeded those of ship observations.After 2011,buoy observation records reached more than ten times the number of ship observations.The large number of buoy observation records is mainly due to continuous observations over time.However,buoy data have much lower spatial and temporal coverage than ship observations,with poor representation of global or large-scale SST,making it impractical to study global SST trends solely using buoy data.
The quality control results indicate that buoy observation data have a higher accuracy rate and better quality than ship observation data.Error comparison analysis between ships,buoys,and ERA5 reanalysis data also shows that buoy observation data have smaller errors.Therefore,buoy data can be used as a high-precision data source to correct or evaluate other data sources.Ship data has a positive systematic bias relative to buoy data,with an average daily bias of approximately 0.2 ℃.
After 1961,the number of global ship-based SST observations increased significantly.Between the 1970s and 1990s,the number of ship records showed phased high values.After the 1990s,the number of ship observations decreased and stabilized.During 1990—2023,the global SST trend shown by ship data was consistent with that of ERA5 data,showing a slow upward trend.Among them,the global SST observed by ships from 1998 to 2012 showed a “basically unchanged plateau period”, while before and after this period,SST showed a clear upward trend.Although ship data have a relatively smaller number of observation samples compared to buoys,their spatial and temporal coverage is relatively larger,providing a certain reference for global SST trend changes.Combining the accuracy of buoy observation data with the coverage of ship observation data in the future can better apply them to global change research.
