This PDF file contains the front matter associated with SPIE Proceedings Volume 12688, including the Title Page, Copyright information, Table of Contents, and Conference Committee information.

Diffusion learning is a generative technique commonly applied to create new images or audio directly from sampled noise. The machine learning approach works by applying degrading signals, such as noise, continuously and learning the denoising process with a neural network. In place of noise, other operations can be performed, such as the addition of atmosphere effects using a physics-based radiative transport code. In this paper, we explore coupling the MODTRAN software to a diffusion learning framework. The goal is to apply atmosphere systematically for a variety of reflective surfaces and use diffusion learning to train models for atmospheric correction. To achieve this, we generate a scoped dataset containing randomized Lambertian surfaces with differing solar illumination and surface angles.

This paper examines target detection statistics when scene illumination changes from full solar illumination to scenes which are topographically shadowed as well as scenes under twilight conditions. The impact of scene shadowing is examined using forward simulations of hyperspectral scenes. The reference reflectance scene contains man-made elements as well as significant areas of vegetation, roads, and bare dirt. The digital elevation map of the area has been modified with the addition of a tall mountain ridge placed to the west of the reference scene. For the detection study a spectral signature from a blue roof retrieved from the scene was randomly embedded into the scene at subpixel levels. The scene radiance is then simulated for various solar zenith angles which produce fully shadowed, partly shadowed, and fully sunlit images. Target detection is performed on these simulated scenes after in-scene atmospheric correction. Standard detection methods yield degraded performance as portions of the scene become shadowed. Approaches that segregate the sunlit and shadowed areas are investigated and shown to improve detection results.

In hyperspectral imaging, the integrity of the spectral signal is critical in order to satisfy the requirements inherent in spectral image processing. There is a need to establish methods for measuring and quantifying band-to-band coregistration errors in the point spread function, as a supplement to customary resolution specifications. The IEEE P4001 working group is creating a standard for hyperspectral imaging which will include metrics for such imperfections. In this work, we discuss a measurement technique that characterizes a hyperspectral camera by creating an image of the point spread function in each band, building on earlier results [H. E. Torkildsen and T. Skauli, Opt. Lett. Vol. 43, No. 16, 2018]. We discuss the implementation of this technique using a collimator suitable for routine camera testing. A line source is projected through the collimator. The source is scanned across a pixel in many different directions. Each scan produces a line spread function, and the set of line spread functions can be transformed into a 2D image of the point spread function via the Radon transform. We argue that this way of characterizing the spatial resolution of a hyperspectral camera is feasible without extraordinary experimental effort, and that it can be considered for inclusion in future versions of the standard. A simplified method requiring only two scans is also explored.

Hyperspectral imaging (HSI) is a technique that reveals hidden information that cannot be seen by the human eye or conventional cameras. However, HSI also poses many challenges such as high data dimensionality, complex signal processing and interpretation, and expensive hardware and software requirements. cuvis.ai is a new open source platform and scientific community that aims to make HSI more accessible, affordable and applicable for everyone. cuvis.ai leverages the power of modern AI concepts to unleash the full potential of HSI. We demonstrate the usefulness of cuvis.ai with three state-of-the-art applications of machine learning in HSI: detecting Alzheimer’s disease, providing surgical guidance, and diagnosing melanoma.

Live fuel moisture and chlorophyll content plays an important role in predicting wildfire ignition and propagation probabilities. We combine geometric and radiative transfer models to characterize the spectral properties of vegetation regions that are observed in visible through short-wave infrared (VNIR/SWIR) hyperspectral images over a wide range of conditions. Leaf spectral reflectance models that depend on moisture and chlorophyll content are combined with MODTRAN atmospheric models to predict sensor radiance spectra. These spectra are used as input to machine learning methods to generate algorithms for estimating fuel characteristics. We demonstrate properties of the models and algorithms using a database of model remote sensing data. We show that this approach provides accurate estimates of vegetation moisture and chlorophyll content.

We present a method for separating the effects of the illumination and atmosphere from spectral reflectance in visible through short-wave infrared (VNIR/SWIR) hyperspectral images collected several times per hour. The data is collected from a tower-based sensor which operates over a 360 degree field-of-view which includes many atmospheric paths in off-nadir viewing geometries. The frequency of collection allows physics-based atmospheric perturbation models to be exploited to constrain the degree of variation in the atmospheric spectral functions. The method makes use of distinctive structure in atmospheric and reflectance spectra that is uncovered using a large MODTRAN atmospheric database. The method is evaluated over a set of hyperspectral data acquired for a region of southern California. We show that this approach can be used to accurately separate spectral reflectance and the illumination/atmospheric conditions in frequent revisit VNIR/SWIR hyperspectral images.

Multispectral and hyperspectral sensors are used today in a wide range of applications. One approach to multispectral imaging uses mosaicked Fabry-Pérot filters on a per-pixel basis over an entire sensor array. They offer single-shot or snapshot spectral imaging over two-dimensional scenes without scanning over wavelengths or along a spatial axis. These multispectral imagers are an extension of the conventional Bayer 2 × 2 (red-green-green-blue) pattern, which means they offer customized spectral responses with an option for an increased number of channels compared to the classic Bayer-color imager. We designed and microfabricated prototype mosaic Fabry-Pérot filters for snapshot imagers where the spectral filters are directly bonded onto an image sensor. Our Fabry-Pérot filters are made of silicon oxide layers with different etalon heights, fabricated onto fused silica wafers, and sandwiched between two dielectric mirrors. The different heights of the oxides are fabricated by using grayscale lithography and reactive ion etching. Grayscale lithography allows for producing mosaic patterns with various heights using a single lithography step. We present results from our custom-designed prototype full-frame (centimeter-scale) 2×2 mosaicked etalon filter arrays. The topographies (height distribution and roughness) of the different mosaic elements were characterized using optical profilometry. The measured transmission curves from several designs offer a range of broad- and narrow-band optical (~400-800 nm) spectral responses. Improvements in uniformities of surface shapes, roughness, and sharper edges were achieved using “Advanced Diffuser-Based Gray-tone Lithography" as discussed in a companion contribution at this meeting.

Through an innovative public-private partnership, a new generation of high-fidelity hyperspectral imaging spectrometers has been designed to pinpoint, quantify, and track methane (CH4) and carbon dioxide (CO2) point-source emissions from super-emitters to help enable reduction of greenhouse gases in the Earth’s atmosphere. Two identical instruments, built concurrently at NASA Jet Propulsion Laboratory (referred to by JPL as the Carbon Plume Mapper project, CPM) and Planet Labs as part of the Carbon Mapper Coalition, feature an identical design which comprises a glass-ceramic, three-mirror anastigmat (TMA) telescope, held in place via a composite metering structure, and Dyson form spectrometer which reduces volume and mass for a fast (F/1.8) optical system. The telescope has a focal length and cross-track field of view (FOV) of 400 mm and 2.6 deg, respectively. Operating in the 400 – 2500 nm spectral range with 5.0 nm sampling, this spectrometer design has the sensitivity and resolution required to meet the demanding needs of space-based detection and quantification of CO2 and CH4 emissions. This work describes the instruments’ optomechanical configuration.

CARBO is a wide field of view imaging spectrometer for measuring CO, CO₂, CH₄ and solar-induced chlorophyll fluorescence (SIF). The key technological feature is an immersion grating that increases dispersion and reduces anamorphism for given design constraints, thus providing a more efficient sampling across the spectral band. Another novel design feature is polarization demultiplexing that provides additional information and improves grating efficiency. We present first data and atmospheric test results for a prototype of the 1598-1659 nm spectral band (used for CO₂ and CH₄ measurements).

Through an innovative public-private partnership, a new generation of high-fidelity imaging spectrometers has been designed for the detection and measurement of methane (CH4) and carbon dioxide (CO2) plumes from super-emitters to help improve accounting and enable reduction of greenhouse gases in the Earth’s atmosphere. Two identical instruments, built concurrently at NASA Jet Propulsion Laboratory (referred to by JPL as the Carbon Plume Mapper project “CPM”) and Planet Labs as part of the Carbon Mapper Coalition, will measure the spectral range of 400 – 2500 nm with a spectral sampling of 5.0 nm. The identical optical design comprises a three-mirror anastigmat (TMA) telescope and Dyson form spectrometer which reduces volume and mass for a fast (F/1.8) optical system. The instruments will be integrated into Planet-built Tanager satellites and launched into low-Earth orbit (LEO). This work describes the assembly and alignment of the two identical instruments. At the subsystem level, both instruments follow the same procedure. For telescope alignment, the mirrors are first coarsely aligned with a coordinate measuring machine (CMM) and then finely aligned in a double-pass interferometer setup. The spectrometer subsystem is aligned onaxis using a commercial lens alignment instrument for precise, non-contact measurements. The telescope and spectrometer alignment results and performance are presented and compared. At the system level, the procedures deviate due to the separate and unique optical ground support equipment (OGSE) configurations utilized by JPL and Planet to implement the same instrument design. Both end-to-end optical alignment configurations are discussed, and the final CPM performance is shown with a focus on the five key and driving imaging spectrometer performance requirements.

The study of aquatic ecosystems is an important research area addressing diverse problems such as carbon sequestration in coastal margins and wetlands, kelp and seagrass studies, coral reefs, harmful algal blooms and hypoxia, and carbon cycling in this dynamic environment. The application of an imaging spectrometer to aquatic ecosystem study is particularly challenging due to low water-leaving radiance levels adjacent to the shore region with its higher values. The Committee on Earth Observation Satellites (CEOS) has established more stringent performance standards for the visible/near infrared wavelengths than are typically available in imaging spectrometer designs. We have recently developed a compact form imaging spectrometer, the Chrisp Compact VNIR/SWIR Imaging Spectrometer (CCVIS), that facilitates their modular usage with a wide field telescope without sacrificing performance. The CCVIS design and the operational concept have predicted performance that approaches the CEOS standards. The envisioned satellite implementation requires a pitchback maneuver where the imaging of the slit projected onto the surface is slowly scanned while recording focal plane array readouts at a higher rate thereby avoiding saturation over the land surface while obtaining a high signal-to-noise ratio over the water. The effective frame rate is determined by the time it takes to scan the projected slit one ground sample distance (GSD). This approach has the added benefit of measuring a range of angles during a single GSD acquisition, providing insight into the bidirectional reflectance distribution function (BRDF).

The HyperSpectral Imager for Climate Science (HySICS) is the sensor payload for the Climate Absolute Radiance and Refractivity Observatory (CLARREO) Pathfinder (CPF) mission. It is scheduled to be launched to the International Space Station for its operational mission. HySICS, an Offner-Chrisp imaging spectrometer, is designed to measure spectral reflectance of the Earth across the full reflective solar region of 350-2300 nm at a radiometric uncertainty requirement of 0.6% (k=2), representing a nearly threefold improvement over existing space-borne spectrometers. An independent calibration (IndCal) approach that relies on a pre-launch, detector- based, absolute radiometric calibration (RadCal) with a transfer to orbit using a high-fidelity instrument model will be used to demonstrate that HySICS meets this level of uncertainty. The activities, plans, procedures, personnel, facilities, and equipment needed for a successful pre-launch IndCal testing, in which the instrument's absolute spectral responses (ASRs) are measured by the Goddard Laser for Absolute Measurement of Radiance (GLAMR), are documented in this paper. A Monte-Carlo simulation of the GLAMR testing is run to predict the instrument's outputs, to identify the calibration error sources and verify the error budget. The trades between the various testing needs of the GLAMR calibration are presented. The resulting test plan is shown that was developed based on the modeling to fit within the testing schedule while optimizing the science needed for CPF0s inter-calibration, spectral line shape assessments, and accuracy of the independent calibration. The HySICS GLAMR testing marks the first application of the detector-based absolute radiometric calibration to an operational space-borne imaging spectrometer.

Methane is a greenhouse gas that has a global warming potential (GWP) of 25 relative to carbon dioxide (CO2). In 2021 an estimated 36 billion-tons of CO2 and 640 million-tons (16 billion-tons GWP equivalent) of methane were emitted. Emission consists of anthropogenic and naturally occurring sources with natural emissions accounting for 35-50% of total emissions. Natural emission caused by decaying organic matter in wetlands and melting tundra, increase as global temperatures rise, thus creating a positive feedback loop increasing their significance and exacerbating global warming. Effective mapping of natural emissions requires high field-of-view (FOV) global satellite coverage. Detection of small weak sources and mapping of larger plume nonuniformity requires a small ground-sample-distance (GSD). Methane has several fine spectral features in the SWIR allowing for effective detection and quantification. In this paper we design an imaging spectrometer for single pixel methane quantification using a tradeoff study between spectral resolution and sensor SNR. Increases in spectral resolution increase spectral separability of methane from other gases, while decreasing SNR. Increases in GSD increase SNR and FOV, while reducing spatial resolution. Environmental factors such as water absorption and ground reflectance further affect performance. Retrieval performance was tested with simulated noisy at-aperture radiance spectra using a dry vegetation surface, atmospheres with varying water and methane concentrations, and the developed sensor model. Gas concentrations were retrieved by finding the atmosphere providing the smoothest retrieved reflectance. Retrieval errors were studied to find the optimal sensor parameters. Several future and current imaging spectrometers were used for comparison.

The Carbon Plume Mapper (CPM) instrument is a high-fidelity imaging spectrometer developed to pinpoint, quantify, and track methane (CH₄) and carbon dioxide (CO₂) point source emissions. CPM is an optically fast F/1.8 Dyson spectrometer that operates over the spectral range of 400 – 2500 nm with a spectral sampling of 5.0 nm. Three diffraction grating designs were measured in a testbed to provide a reliable prediction of grating performance in a Dyson system to inform CPM grating design. This paper will detail the gratings, testbed design, measurement process, and data used to assess grating efficiency through wavelength (500-1700 nm) of three grating designs, both full aperture and sub-aperture for two field angles, polarized and unpolarized.

The high spectral dimension of hyperspectral data justifies its great potential to accurately estimate fire severity. In this work, we validated spaceborne hyperspectral data from the Italian Precursore IperSpettrale della Missione Applicativa (PRISMA) mission to estimate fire severity in one of the largest forest fires ever recorded in the western Mediterranean, the megafire (28,046 ha) that occurred in the Sierra de la Culebra (northwestern Iberian Peninsula) between 15 and 19 June 2022. To take advantage of the high spectral dimensionality, Multiple Endmember Spectral Mixture Analysis (MESMA) was used to transform an original PRISMA Level 2D 1A scene into three fraction images of the basic components present in the image, namely CHAR, photosynthetic vegetation (PV), and non-photosynthetic vegetation and soil (NPVS). MESMA decomposed each pixel using different combinations of potential endmembers, overcoming the limitation of Linear Spectral Mixture Analysis of using the same number of endmembers to model all image pixels. We field measured initial fire severity in 70 plots using a slightly modified version of the Composite Burn Index (CBI). Shade-normalized char fraction image and CBI values showed a relatively strong linear relationship (R2 = 0.70). As a benchmark, we used the Delta Normalized Burn Ratio (dNBR) index from Sentinel 2A data. Linear regression between dNBR and CBI values obtained an R2 = 0.36. These results allow to conclude the suitability of the PRISMA hyperspatial data and the MESMA processing technique for the accurate assessment of fire severity in Mediterranean forest ecosystems.

The Ultra-Compact Imaging Spectrometer Moon (UCIS-Moon) instrument is a pushbroom shortwave infrared (SWIR) imaging spectrometer prototype developed at NASA’s Jet Propulsion Laboratory (JPL), California Institute of Technology under the Development and Advancement of Lunar Instrumentation (DALI) program. It is designed for integration with a lander or rover for lunar surface science missions. Operating over a 0.6 to 3.6 micron spectral range with 10 nm sampling and a 36 degree field of view, UCIS-Moon is capable of detecting spectral absorptions from common lunar materials, OH species, molecular H₂O, water ice, organics, and placing mineral identifications within an established geologic context at the cm to m scale. We discuss instrument assembly, alignment, and measured laboratory optical performance, which meets or exceeds the high-uniformity and high-resolution requirements while achieving a wide spectral range, field of view, and environmental tolerance, with limited mass and power resources. As such, the UCIS-Moon imaging spectrometer is well-suited to address key science questions about lunar geology, the abundance, sources, and sinks of volatiles at the Moon, and the distribution of possible in situ resources for future human exploration.