The LANDHYPERNET and WATERHYPERNET networks (which together make up the HYPERNETS network) consist of a set of autonomous hyperspectral spectroradiometers (HYPSTAR^®) acquiring fiducial reference measurements of surface reflectance at various sites covering a wide range of surface types (both land and water) for use in satellite Earth observation validation and remote sensing applications. This paper describes the processing algorithm for the HYPSTAR^® data products. The hypernets_processor is a Python software package to process the LANDHYPERNET and WATERHYPERNET in-situ hyperspectral raw data, collected from the measurement network under the standard measurement protocols, to the designated products, through data transmission and conversion, application of calibration, evaluation of reflectance and other variables, and, archiving for distribution to users. In order to achieve fiducial reference measurement quality, uncertainties are propagated through each step of the processing chain, taking into account temporal and spectral error-covariance. Such detailed uncertainty information is unique for any satellite validation network. We also describe the HYPSTAR^® products acquired until 2023–04–31, consisting of 12,190 LANDHYPERNET sequences and 55,514 WATERHYPERNET sequences (of which respectively 11,802 and 44,412 were successfully processed to surface reflectance).

Satellites are now routinely used for measuring water and land surface reflectance and hence environmentally relevant parameters such as aquatic chlorophyllaconcentration and terrestrial vegetation indices. For each satellite mission, radiometric validation is needed at bottom of atmosphere for all spectral bands and covering all typical conditions where the satellite data will be used. Existing networks such as AERONET-OC for water and RadCalNet for land provide vital information for validation, but (AERONET-OC) do not cover all spectral bands or (RadCalNet) do not cover all surface types and viewing angles. In this Perspective Article we discuss recent advances in instrumentation, measurement methods and uncertainty estimation in the field of optical radiometry and put forward the viewpoint that a new network of automated hyperspectral radiometers is needed for multi-mission radiometric validation of water and land surface reflectance. The HYPERNETS federated network concept is described, providing a context for research papers on specific aspects of the network. This network is unique in its common approach to both land and water surfaces. The common aspects and the differences between land and water measurements are explained. Based on early enthusiasm for HYPERNETS data from validation-oriented workshops, it is our viewpoint that this new network of automated hyperspectral radiometers will be useful for multi-mission radiometric validation of water and multi-angle land surface reflectance. The HYPERNETS network has strong synergy with other measurement networks (AERONET, AERONET-OC, RadCalNet, FLUXNET, ICOS, skycam, etc.) and with optional supplementary measurements, e.g., water turbidity and fluorescence, land surface temperature and soil moisture, etc.

Atmospheric corrections introduce uncertainties in bottom-of-atmosphere Ocean Colour (OC) products. In this paper, we analyse the uncertainty budget of the SeaDAS atmospheric correction algorithm. A metrological approach is followed, where each of the error sources are identified in an uncertainty tree diagram and briefly discussed. Atmospheric correction algorithms depend on ancillary variables (such as meteorological properties and column densities of gases), yet the uncertainties in these variables were not studied previously in detail. To analyse these uncertainties for the first time, the spread in the ERA5 ensemble is used as an estimate for the uncertainty in the ancillary data, which is then propagated to uncertainties in remote sensing reflectances using a Monte Carlo approach and the SeaDAS atmospheric correction algorithm. In an example data set, wind speed and relative humidity are found to be the main contributors (among the ancillary parameters) to the remote sensing reflectance uncertainties.

The HYPERNETS project developed a new hyperspectral radiometer (HYPSTAR^®) integrated in automated networks of water (WATERHYPERNET) and land (LANDHYPERNET) bidirectional reflectance measurements for satellite validation. In this paper, the feasibility of using LANDHYPERNET surface reflectance data for vicarious calibration of multispectral (Sentinel-2 and Landsat 8/9) and hyperspectral (PRISMA) satellites is studied. The pipeline to process bottom of atmosphere (BOA) surface reflectance HYPERNETS data to band-integrated top of atmosphere (TOA) reflectances and compare them to satellite observations is detailed. Two LANDHYPERNET sites are considered in this study: the Gobabeb HYPERNETS site in Namibia (GHNA) and Princess Elizabeth Base in Antarctica (PEAN). 36 near-simultaneous match-ups within 1 h are found where HYPERNETS and satellite data pass all quality checks. For the Gobabeb HYPERNETS site, agreement to within 5% is found with Sentinel-2 and Landsat 8/9. The differences with PRISMA are smaller than 10%. For the HYPERNETS Antarctica site, there are also a number of match-ups with good agreement to within 5% for Landsat 8/9. The majority show notable disagreement, i.e., HYPERNETS being over 10% different compared to satellite. This is due to small-scale irregularities in the wind-blown snow surface, and their shadows cast by the low Sun. A study comparing the HYPERNETS measurements against a bidirectional reflectance distribution function (BRDF) model is recommended. Overall, we confirm data from radiometrically stable HYPERNETS sites with sufficient spatial and angular homogeneity can successfully be used for vicarious calibration purposes.

Correcting atmospheric effects on hyperspectral optical satellite scenes is paramount to ensuring the accuracy of derived bio-geophysical products. The open-access benchmark Atmospheric Correction Inter-comparison eXercise (ACIX) was first initiated in 2016 and has now been extended to provide a comprehensive assessment of atmospheric processors of space-borne imaging spectroscopy missions (EnMAP and PRISMA) over land surfaces. The exercise contains 90 scenes, covering stations of the Aerosol Robotic Network (AERONET) for assessing aerosol optical depth (AOD) and water vapour (WV) retrievals, as well as stationary networks (RadCalNet and HYPERNETS) and ad hoc campaigns for surface reflectance (SR) validation. AOD, WV, and SR retrievals were assessed using accuracy, precision, and uncertainty metrics. For AOD retrieval, processors showed a range of uncertainties, with half showing overall uncertainties of <0.1 but going up to uncertainties of almost 0.4. WV retrievals showed consistent offsets for almost all processors, with uncertainty values between 0.171 and 0.875 g/cm². Average uncertainties for SR retrievals depend on wavelength, processor, and sensor (uncertainties are slightly higher for PRISMA), showing average values between 0.02 and 0.04. Although results are biased towards a limited selection of ground measurements over arid regions with low AOD, this study shows a detailed analysis of similarities and differences of seven processors. This work provides critical insights for understanding the current capabilities and limitations of atmospheric correction algorithms for imaging spectroscopy, offering both a foundation for future improvements and a first practical guide to support users in selecting the most suitable processor for their application needs.

Environmental sciences are supported by multi-decadal observations from instruments on land, in the oceans, in the air, and from space. For such observations to have sufficient quality to underpin decisions, they need the long-term stability and global equivalence achieved through traceability to an internationally agreed reference, preferably the international system of units, the SI. Such traceability is achieved through metrological principles of linking measurements through an unbroken chain of comparisons to suitable references, with rigorous uncertainty analysis following internationally accepted approaches, and validated through comparisons with independent approaches at key steps of the chain. Increasingly, communities responsible for coordinating observational networks and transforming their data into products and services, whether from satellites or other platforms, are calling for metrological principles to be applied across the entire value chain. However, it is not always straightforward to apply these principles to observations of the Earth’s environmental system. This paper discusses what these metrological principles are, how they differ from other approaches in use in the observational sciences, and how the modern, evolving metrological guides can be used by those observational communities, to meet the challenging requirements of climate-quality observations and the data products derived from them.