By the end of this decade, several space agencies and commercial operators will develop and launch the next generation of space-based Earth observation satellites. This paper focuses on imaging spectroradiometers, which measure solar radiation reflected by the Earth’s surface across the ultra-violet, visible and short-wave infrared spectral ranges using hyperspectral sensors. To enable interoperable use of datasets from this growing constellation of sensors, it is essential to understand and manage differences between the measurements of the same target under similar conditions. These differences arise from multiple factors: real variations due to instrument design (e.g., differing spectral responses or viewing geometries), differences in Level 1 processing approaches based on pre-flight and post-launch calibration strategies, differences in higher-level processing into biophysical quantities, and any residual biases between sensors caused by measurement or processing errors — some of which may not be fully captured in reported uncertainties. A consistent approach to the analysis and documentation of uncertainties supports intercomparison between datasets. By assessing whether measurements agree within the expected uncertainty associated with the comparison, or identifying and quantifying systematic deviations when they do not, it becomes possible to validate interoperability and diagnose discrepancies. This paper outlines key concepts for harmonising uncertainty analysis and reporting techniques across missions, including the use of structured uncertainty propagation frameworks such as uncertainty trees. The goal is to initiate a roadmap for understanding and quantifying uncertainties in a way that facilitates robust intercomparison of multiple space-based imaging spectroradiometer measurements.
A thorough understanding of the global carbon pools and cycle is essential for understanding and predicting the effects of climate change. Coastal waters play a key role in the global carbon cycle but remain poorly understood due to their optical complexity, high spatial variability, and sensitivity to climate change. Satellite remote sensing data can provide high spatial and temporal resolution for carbon monitoring at local, regional, and global scales. This study estimates the potential of Sentinel-2 MSI (S2) and Sentinel-3 OLCI (S3) for mapping carbon fractions in coastal waters. In situ hyperspectral reflectance measurements and concentrations of different carbon fractions, namely Total Organic Carbon (TOC), Dissolved Organic Carbon (DOC), Particulate Organic Carbon (POC), and Dissolved Inorganic Carbon (DIC), were used. Bio-optical and physical water parameters, as well as carbon fractions, were measured in the coastal areas of the eastern Baltic Sea from 2023 to 2024. In situ reflectance spectra were recalculated to S2 and S3 band reflectance using their respective spectral response functions. Regression models, including linear regression and machine learning methods, were developed and validated against in situ data to estimate carbon fractions. For S2, the combinations of green and red or red-edge bands achieved the best performance in carbon retrieval, with R2 values of 0.67, 0.64, and 0.56 for TOC, DOC, and DIC, respectively. The corresponding MAPE values ranged from 14% to 18%, and biases were close to zero. Slightly higher accuracies were obtained with S3 simulations. The S3 models that used reflectance at 490, 620, and 665 nm produced the most accurate estimates for both TOC (R2 = 0.71) and DOC (R2 = 0.67). For DIC, the best predictive model combined information from 510, 665, and 709 nm bands, achieving R2 of 0.62. Among the tested optical proxies, CDOM absorption at 443 nm showed the most consistent relationship with TOC, DOC, and DIC (R2 = 0.56–0.30). The study represents a significant step forward in the remote sensing of coastal waters, and if adopted, the proposed carbon-fraction products could enable substantial progress across various fields, from carbon-cycle research to monitoring and policymaking.
