Google Earth Engine for Sea Surface Temperature Mapping in Marine Environments: A Review

Abstract

Sea surface temperature (SST) is a fundamental oceanographic variable for describing air–sea exchange, ocean circulation, fisheries habitat, coral thermal stress, coastal processes, and climate variability. The rapid expansion of satellite archives has improved SST observation, but conventional desktop workflows remain constrained by data volume, preprocessing requirements, and reproducibility. This review evaluates the role of Google Earth Engine (GEE) as a cloud-computing environment for SST mapping in marine and coastal research. A structured narrative review was conducted using peer-reviewed literature and authoritative dataset documentation published mainly between 2017 and 2026, with emphasis on studies that employed GEE, satellite-derived SST products, validation procedures, or marine applications. The synthesis shows that GEE substantially reduces data-handling barriers and supports scalable time-series analysis through readily accessible products such as MODIS Aqua/Terra Ocean Color L3, NOAA OISST v2.1, NOAA WHOI SST, and JAXA GCOM-C/SGLI SST. However, dataset choice remains application-dependent. Coarse, gap-filled products are effective for climate-scale trends and anomalies, whereas thermal infrared sensors with kilometre- to hectometre-scale observations are more appropriate for coastal gradients but require stricter cloud screening, atmospheric correction, and validation. The principal methodological risks are confusion between land surface temperature and true marine SST, inadequate quality-flag filtering, nearshore mixed pixels, inconsistent temporal compositing, and weak validation. A reproducible workflow is proposed that integrates dataset selection, quality assurance, temporal compositing, anomaly analysis, in-situ or multi-product validation, and uncertainty reporting. GEE is therefore best regarded as an analytical infrastructure rather than an SST algorithm itself; scientific reliability ultimately depends on sensor physics, product quality, validation design, and transparent code-based processing.