Conference Agenda

Global imaging spectrometer missions such as PRISMA, DESIS, EnMAP, EMIT, CHIME, and SBG-VSWIR must provide high-quality spectra over a wide variety of targets, from dark aquatic regions to bright desert playas. In order to have cohesive and interoperable data products, instruments will need to be calibrated to the same standard. One elusive yet important aspect of calibration is detector nonlinearity characterization, which is especially important for high performance for targets with high spectral contrast, including atmospheric absorption features. Since measuring the nonlinearity of a detector requires leveraging the linearity of another process, such as the linearity of increasing integration time or illumination intensity, it is a difficult characterization to make, as it requires high knowledge of the imaging system and test equipment. This is compounded for imaging spectrometers, as the signal response across wavelengths is nonuniform by design, and full-spectrum illumination is more difficult to characterize than narrow-band illumination, such as lasers. To overcome this challenge, we will present nonlinearity calibration measurements tested on the Mapping Imaging Spectrometer for Europa (MISE) instrument. MISE has a HgCdTe array with higher nonlinearity than expected for SBG-VSWIR due to differences in the detector hardware. MISE’s relatively high nonlinearity allowed us to compare four nonlinearity calibration measurements, including a ramped spectral source, integration time sweep, controlled source changes, and the double aperture method. By combining multiple measurements, we were able to distinguish the detector nonlinearity from the nonlinearities inherent in the test configurations. We will test these methods on the upcoming AVIRIS-5 airborne instrument, which has a similar detector to SBG-VSWIR, and we will compare to a pulsed laser nonlinearity measurement of a SBG-VSWIR non-integrated engineering model detector. We hope to establish a best practice for nonlinearity characterization of imaging spectrometers which can be shared across SBG-VSWIR, CHIME, and other upcoming imaging spectrometer missions. This type of collaboration in calibration standards will allow for the most accurate, most interoperable measurements possible for a wide range of scenes across the Earth’s surface.

The HYPERNETS project developed an innovative hyperspectral radiometer (HYPSTAR®) integrated in automated networks of water (WATERHYPERNET) and land (LANDHYPERNET) bidirectional reflectance measurements for satellite validation. This new network of automated hyperspectral radiometers will be invaluable for radiometric validation of water and multi-angle land surface reflectance for hyperspectral (and multi-spectral) satellite sensors. We here present the LANDHYPERNET network, including its measurement protocol, processing to surface reflectance and examples of satellite validation of various products.

The HYPSTAR®-XR instrument used in the LANDHYPERNET network features both a VNIR and SWIR sensor. It provides VNIR data with a wavelength range of 380–1000 nm with 0.5 nm sampling and 3 nm resolution, and SWIR data ranging from 1000–1680 nm with 3 nm sampling and 10 nm resolution. The raw data is automatically transferred to a central server, processed in near real-time to reflectance and other variables and archived for web distribution. Furthermore, to achieve fiducial reference measurement quality, measurement uncertainty is propagated through the full processing chain, including treatment of temporal and wavelength error-covariance, a level of detail unique for such satellite validation network.

We also highlight some results comparing the LANDHYPERNET network data to satellite data. These include a study looking at the feasibility of using LANDHYPERNET surface reflectance data for vicarious calibration of multispectral (Sentinel-2 and Landsat 8/9) and hyperspectral (PRISMA) satellites and a study utilising LANDHYPERNET data products to understand the seasonal dynamics of the reflectance’s of a deciduous broadleaf forest in the UK.

Over the course of more than two and a half years, the hyperspectral EnMAP (Environmental Mapping and Analysis Program) satellite has been orbiting, and the EnMAP science segment at the GFZ has been conducting independent EnMAP data product validation monitoring activities. These efforts have provided insights into EnMAP product quality and paved the way for in-depth sensor cross-validation. This independent EnMAP product validation is a crucial component of the EnMAP mission, as it continuously monitors the radiometric, geometric, and spectral product quality during the entire operational phase. Moreover, the cross-validation between the other in-orbit imaging sensor systems is a significant step towards product harmonization, quantitative comparability of the spectral signal, and classification of the EnMAP data quality itself. The need for cross-sensor-based time series further underscores the importance of this research, as it challenges the complementarity between the different missions.

To validate the radiometric, geometric, and spectral EnMAP product quality, EnMAP image-only and in-situ-based validation scenarios are realized with specially developed algorithms and workflows. In particular, for cross-validation, EnMAP L1B and L2A data are systematically compared with the respective PRISMA and EMIT products. Decoupling and quantification of radiometric, geometric & spectral error contributions are realized. All close-in-time and spatially overlapping acquisitions are identified globally in the first step. Different acquisition time offsets of up to four days are allowed depending on the surface type. There must be a spatial overlap of at least 1/3 of a tile and no clouds. In the second step, homogeneous segments in the overlapping regions are identified. Proper spatial co-alignment of the homogeneous patches and normalizing viewing and illumination conditions are guaranteed. Well-known and stable sites like the PICS, RadCalNet sites, and in-situ field measurements are used as anchor points to realize, besides the relative, also an absolute validation.

For both product levels, globally distributed and parameter-wise well-balanced wavelength- but also signal-level-based disparity statistics (MEAN, STD, and RSME) are presented, which allow the decoupling of systematic (accuracy) and random (precision) disparity contributions. With the growing number of spatial and temporal matching acquisitions within the different missions, the balancing and covering of parameter ranges are improved, and the disparities statistics are stabilized. This will allow gain- and offset-based product harmonization between the missions in the future.

.