Search NASA⌕ Search

SEARCH · Search NASA

Results for “MASKING”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 37 records · Page 2
All sourcesNASA NTRSDOE OSTI
NASA NTRSDate not supplied

Multi-Star Wavefront Control at the Occulting Mask Coronagraph Testbed: Monochromatic Laboratory Demonstration for the Roman Coronagraph Instrument

The Astro2020 decadal survey recommended a direct imaging flagship mission with a goal for mission yield of 25 or more potentially habitable exoplanets. A majority of Sun-like stars have a stellar companion that can introduce additional noise into the field of view of any high-contrast imaging instrument and enabling exoplanet discovery around binary stars represents a path to increased coronagraphic instrument efficiency. This includes both of the Alpha Centauri A and B stars which would represent the top science target for direct imaging if companion leakage can be suppressed. Multi-Star Wavefront Control (MSWC) is a technique that removes stellar leakage from both stellar components, enabling direct imaging of exoplanets in many binary star systems. We present the latest testbed results obtained with MSWC as part of the technology development effort focusing on demonstrations conducted on the Occulting Mask Coronagraph (OMC) testbed at JPL. OMC has a layout similar to the Roman Space Telescope coronagraph instrument (CGI) and is outfitted with a MSWC mask with the same design as the contributed mask for the Roman CGI. Our testbed results represent the first demonstrations of this technique using the recently installed full binary source for a geometry matching potential Alpha Centauri observations. Technical Review Abstract: The Astro2020 decadal survey recommended a direct imaging flagship mission with a goal for mission yield of 25 or more potentially habitable exoplanets. A majority of Sun-like stars have a stellar companion that can introduce additional noise into the field of view of any high-contrast imaging instrument and enabling exoplanet discovery around binary stars represents a path to increased coronagraphic instrument efficiency by increasing the available science target pool of bright nearby stars. This includes both of the Alpha Centauri A and B stars which would represent the top science target for direct imaging if companion leakage can be suppressed. Multi-Star Wavefront Control (MSWC) is a technique that removes stellar leakage from both stellar components, enabling direct imaging of exoplanets in many binary star systems . We present the latest testbed results obtained with MSWC as part of the technology development effort focusing on demonstrations conducted on the Occulting Mask Coronagraph (OMC) testbed at JPL during the completed first vacuum window and recent results from the ongoing second vacuum window. The MSWC mask consists of a shaped pupil mask similar to the one used for the Wide-Field of View mode but also includes a set of superimposed, regularly-spaced dots that serve as a diffraction grating. OMC has a layout similar to the Roman Space Telescope coronagraph instrument (CGI) and is configured with a binary imaging mode with a MSWC mask using same design as the contributed mask for the Roman CGI. Our testbed results represent the first demonstrations of this technique using the recently installed full binary source. We present results obtained in Super-Nyquist regime demonstrating suppression in the Super-Nyquist regime for the 3rd diffraction order reaching 8.7e-9 contrast with the Roman pupil. In addition, we present results obtained with the binary-star regime demonstrating 9.6e-8 contrast for a geometry matching potential Alpha Centauri observations in a monochromatic wavelength similar to Band 1. Planned demonstrations in the 2nd vacuum window will focus on Band 3d and Band 4 using the full MSWC mode for an Alpha Centauri geometry.

High-contrast imaging↗
NASA NTRS1990

Design criteria for small coded aperture masks in gamma-ray astronomy

Most theoretical work on coded aperture masks in X-ray and low-energy gamma-ray astronomy has concentrated on masks with large numbers of elements. For gamma-ray spectrometers in the MeV range, the detector plane usually has only a few discrete elements, so that masks with small numbers of elements are called for. For this case it is feasible to analyze by computer all the possible mask patterns of given dimension to find the ones that best satisfy the desired performance criteria. A particular set of performance criteria for comparing the flux sensitivities, source positioning accuracies and transparencies of different mask patterns is developed. The results of such a computer analysis for masks up to dimension 5 x 5 unit cell are presented and it is concluded that there is a great deal of flexibility in the choice of mask pattern for each dimension.

Sembay, S.↗
NASA NTRS1999

Light masking of circadian rhythms of heat production, heat loss, and body temperature in squirrel monkeys

Whole body heat production (HP) and heat loss (HL) were examined to determine their relative contributions to light masking of the circadian rhythm in body temperature (Tb). Squirrel monkey metabolism (n = 6) was monitored by both indirect and direct calorimetry, with telemetered measurement of body temperature and activity. Feeding was also measured. Responses to an entraining light-dark (LD) cycle (LD 12:12) and a masking LD cycle (LD 2:2) were compared. HP and HL contributed to both the daily rhythm and the masking changes in Tb. All variables showed phase-dependent masking responses. Masking transients at L or D transitions were generally greater during subjective day; however, L masking resulted in sustained elevation of Tb, HP, and HL during subjective night. Parallel, apparently compensatory, changes of HL and HP suggest action by both the circadian timing system and light masking on Tb set point. Furthermore, transient HL increases during subjective night suggest that gain change may supplement set point regulation of Tb.

Non-NASA Center↗
NASA NTRS2008

The Sensitivity of Coded Mask Telescopes

Simple formulae are often used to estimate the sensitivity of coded mask X-ray or gamma-ray telescopes, but t,hese are strictly only applicable if a number of basic assumptions are met. Complications arise, for example, if a grid structure is used to support the mask elements, if the detector spatial resolution is not good enough to completely resolve all the detail in the shadow of the mask or if any of a number of other simplifying conditions are not fulfilled. We derive more general expressions for the Poisson-noise-limited sensitivity of astronomical telescopes using the coded mask technique, noting explicitly in what circumstances they are applicable. The emphasis is on using nomenclature and techniques that result in simple and revealing results. Where no convenient expression is available a procedure is given which allows the calculation of the sensitivity. We consider certain aspects of the optimisation of the design of a coded mask telescope and show that when the detector spatial resolution and the mask to detector separation are fixed, the best source location accuracy is obtained when the mask elements are equal in size to the detector pixels.

Skinner, Gerald K.↗
NASA NTRS2013

Achromatic Focal Plane Mask for Exoplanet Imaging Coronagraphy

Recent advances in coronagraph technologies for exoplanet imaging have achieved contrasts close to 1e10 at 4 lambda/D and 1e-9 at 2 lambda/D in monochromatic light. A remaining technological challenge is to achieve high contrast in broadband light; a challenge that is largely limited by chromaticity of the focal plane mask. The size of a star image scales linearly with wavelength. Focal plane masks are typically the same size at all wavelengths, and must be sized for the longest wavelength in the observational band to avoid starlight leakage. However, this oversized mask blocks useful discovery space from the shorter wavelengths. We present here the design, development, and testing of an achromatic focal plane mask based on the concept of optical filtering by a diffractive optical element (DOE). The mask consists of an array of DOE cells, the combination of which functions as a wavelength filter with any desired amplitude and phase transmission. The effective size of the mask scales nearly linearly with wavelength, and allows significant improvement in the inner working angle of the coronagraph at shorter wavelengths. The design is applicable to almost any coronagraph configuration, and enables operation in a wider band of wavelengths than would otherwise be possible. We include initial results from a laboratory demonstration of the mask with the Phase Induced Amplitude Apodization coronagraph.

Achromatic↗
DOE OSTI2026

Original images, ground truth annotations, and precited masks from U-NET for a fungal species x nitrogen experiment

This document describes the “MicroVision-MV003” dataset. This folder contains three subfolders: “originals”, “predicted_masks”, and “gt_masks”. The “originals” folder contains 720 original images acquired using two imaging modalities: 360 images from overhead imaging and 360 images from transmission (raw) imaging. Naming scheme:YYYY-MM-dd__plate_strain_nitrogen-level_replication_experiment_imaging-modalityYYYY-MM-dd: 2024-12-20, 2024-12-21, 2024-12-22,2024-12-23,2024-12-24,2024-12-25,2024-12-26,2024-12-27,2024-12-28,2024-12-29,2024-12-30,2024-12-31.plate: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30.strain: FG, LE.nitrogen-level: N-1, N-10, N-100.replication: 1, 2, 3, 4, 5.experiment: MV.003.imaging-modality: overhead, transmission (raw). For example, the image2024-12-20__1_FG_N-10_3_MV.003_overheadwas taken on 2024-12-20. The plate number is 1, the fungal strain is FG, the nitrogen treatment is N-10, and the replication number is 3 for experiment MV.003, acquired using the overhead imaging modality. The “predicted_masks” folder contains masks generated by a trained U-Net model for the transmission imaging modality.The “gt_masks” folder contains 326 human hand-traced masks that serve as ground truth.

09 BIOMASS FUELS↗
NASA NTRS1976

Influence of auditory fatigue on masked pure-tone thresholds

A description is presented of four related experiments involving conditions of 3-kHz low-intensity masking, a replication of experiment I with slight variations, 3-kHz high-intensity masking, and 6-kHz low-intensity masking. The frequencies of the tones which the observers detected were 3 and 6 kHz. The observed change in masked-tone threshold as a function of fatigue is discussed. It is found that masked-tone-detection thresholds remain essentially unchanged following fatigue if the masking-noise intensity is sufficiently great.

Parker, D. E.↗
NASA NTRS1985

Secondary gamma-ray production in a coded aperture mask

The application of the coded aperture mask to high energy gamma-ray astronomy will provide the capability of locating a cosmic gamma-ray point source with a precision of a few arc-minutes above 20 MeV. Recent tests using a mask in conjunction with drift chamber detectors have shown that the expected point spread function is achieved over an acceptance cone of 25 deg. A telescope employing this technique differs from a conventional telescope only in that the presence of the mask modifies the radiation field in the vicinity of the detection plane. In addition to reducing the primary photon flux incident on the detector by absorption in the mask elements, the mask will also be a secondary radiator of gamma-rays. The various background components in a CAMTRAC (Coded Aperture Mask Track Chamber) telescope are considered. Monte-Carlo calculations are compared with recent measurements obtained using a prototype instrument in a tagged photon beam line.

Owens, A.↗
NASA NTRS2005

Techniques for Producing Coastal Land Water Masks from Landsat and Other Multispectral Satellite Data

Coastal erosion and land loss continue to threaten many areas in the United States. Landsat data has been used to monitor regional coastal change since the 1970s. Many techniques can be used to produce coastal land water masks, including image classification and density slicing of individual bands or of band ratios. Band ratios used in land water detection include several variations of the Normalized Difference Water Index (NDWI). This poster discusses a study that compares land water masks computed from unsupervised Landsat image classification with masks from density-sliced band ratios and from the Landsat TM band 5. The greater New Orleans area is employed in this study, due to its abundance of coastal habitats and its vulnerability to coastal land loss. Image classification produced the best results based on visual comparison to higher resolution satellite and aerial image displays. However, density sliced NDWI imagery from either near infrared (NIR) and blue bands or from NIR and green bands also produced more effective land water masks than imagery from the density-sliced Landsat TM band 5. NDWI based on NIR and green bands is noteworthy because it allows land water masks to be generated from multispectral satellite sensors without a blue band (e.g., ASTER and Landsat MSS). NDWI techniques also have potential for producing land water masks from coarser scaled satellite data, such as MODIS.

Spruce, Joseph P.↗
NASA NTRS2005

Low-order aberration sensitivity of eighth-order coronagraph masks

In a recent paper, Kuchner, Crepp, and Ge describe new image-plane coronagraph mask designs that reject to eighth order the leakage of starlight caused by image motion at the mask, resulting in a substantial relaxation of image centroiding requirements compared to previous fourth-order and second-order masks. They also suggest that the new masks are effective at rejecting leakage caused by low-order aberrations (e.g., focus, coma, and astigmatism). In this paper, we derive the sensitivity of eighth-order masks to aberrations of any order and provide simulations of coronagraph behavior in the presence of optical aberrations.We find that the masks leak light as the fourth power of focus, astigmatism, coma, and trefoil. This has tremendous performance advantages for the Terrestrial Planet Finder Coronagraph.

planetary systems↗
NASA NTRS2005

Techniques for Producing Coastal Land Water Masks from Landsat and Other Multispectral Satellite Data

Coastal erosion and land loss continue to threaten many areas in the United States. Landsat data has been used to monitor regional coastal change since the 1970's. Many techniques can be used to produce coastal land water masks, including image classification and density slicing of individual bands or of band ratios. Band ratios used in land water detection include several variations of the Normalized Difference Water Index (NDWI). This poster discusses a study that compares land water masks computed from unsupervised Landsat image classification with masks from density-sliced band ratios and from the Landsat TM band 5. The greater New Orleans area is imployed in this study, due to its abundance of coastal habitats and ist vulnerability to coastal land loss. Image classification produced the best results based on visual comparison to higher resolution satellite and aerial image displays. However, density-sliced NDWI imagery from either near infrared (NIR) and blue bands or from NIR and green bands also produced more effective land water masks than imagery from the density-sliced Landsat TM band 5. NDWI based on NIR and green bands is noteworthy because it allows land water masks to be generated form multispectral satellite sensors without a blue band (e.g., ASTER and Landsat MSS). NDWI techniques also have potential for producing land water masks from coarser scaled satellite data, such as MODIS.

Spruce, Joe↗
NASA NTRS2020

Prediction and Evaluation of the Image of the WFIRST Coronagraph Pupil at the Shaped-Pupil Mask

To create a high contrast field of view with a shaped pupil coronagraph (SPC), the telescope pupil is imaged onto the shaped pupil mask, the shaped pupil mask is designed and fabricated through a design optimization based on a given pupil wavefront, and deformable mirrors (DM) in the coronagraph are modulated according to a nonlinear optimization procedure. Both the mask design optimization and the DM optimization require accurate knowledge of the image of the pupil at the shaped pupil plane for the calculation of Jacobians, and the evaluation of alignment tolerances. The shaped pupil coronagraphs demonstrated in the high contrast imaging testbed (HCIT) at JPL has heretofore assumed perfect imaging of the pupil onto the shaped pupil mask. Similarly, the design and performance modeling of the WFIRST CGI instrument makes the same assumption. In this paper we estimate and compare the image of the pupil mask at the shaped pupil plane in the case of the coronagraph testbed, and the image of the telescope pupil at the shaped pupil plane for the WFIRST CGI instrument. The estimated pupil images include field curvature and aberrations resulting from the layout of the parabolic mirror imaging system, which is optimized for the source image, not the image of the pupil. We also show the results of direct measurement of the image of the pupil mask at the shaped pupil plane in the coronagraph testbed.

Crowther, Blake↗
NASA NTRS2022

Initial Laboratory Demonstration of Multi-Star Wavefront Control at the Occulting Mask Coronagraph Testbed

Online Abstract: A majority of Sun-like stars, such as the A and B components of Alpha Centauri, have at least one stellar companion that can introduce additional noise into the field of view of any high-contrast imaging instrument. Multi-Star Wavefront Control (MSWC) is a wavefront-control technique that removes stellar leakage from both stellar components, enabling direct imaging of planets in many binary star systems. We present the latest experimental and modeling results obtained with MSWC as part of the technology development effort focusing on demonstrations conducted on the Occulting Mask Coronagraph (OMC) testbed at JPL. OMC has a layout similar to the Roman Space Telescope coronagraph instrument (CGI), and we used a MSWC mask similar to the one that was contributed to the Roman CGI. Our results represent the first demonstrations of this technique on the OMC testbed, with the ultimate goal of demonstrating full MSWC with validated models at contrast levels relevant to Roman CGI. Technical Review Abstract: A majority of Sun-like stars have at least one stellar companion that can introduce additional noise into the field of view of any high-contrast imaging instrument, limiting the achievable contrast. These include high-quality target stars such as the A and B components of Alpha Centauri, our nearest stellar neighbor. Enabling direct imaging of binary stars has the potential to increase the scientific yield for coronagraphic instruments planned on NASA's future space missions including the Roman Space Telescope and the next IR/O/UV Flagship recommended by Astro2020. Multi-Star Wavefront Control (MSWC) is a wavefront-control technique that simultaneously removes the (mutually incoherent) stellar leakage from both stellar components, enabling direct imaging of planets in many binary star systems. MSWC is an algorithmic technique and can be used with existing wavefront control systems on coronagraphic instruments (as well as starshades if a deformable mirror is available in the optical path). We summarize the latest experimental and numerical results obtained with MSWC as part of the technology development effort to demonstrate compatibility with existing high-contrast imaging platforms for this technique. The Super-Nyquist regime of MSWC was tested in vacuum at JPL's High Contrast Imaging Testbed (HCIT) on the Decadal Survey Testbed (DST) reaching 8.6e-9 contrast in a 10% band. Recently, the Occulting Mask Coronagraph (OMC) testbed at JPL is being prepared for demonstrations of Multi-Star Wavefront Control. A shaped pupil mask similar to the contributed MSWC mask on the Roman Space Telescope's coronagraph instrument has been recently manufactured including matching Lyot and focal plane masks and being installed on the OMC testbed. The goal of this experiment is a demonstration of MSWC using validated models on a testbed configuration and at contrast levels relevant to the Roman coronagraphic instrument.

High-contrast imaging↗
NASA NTRS2013

High Contrast Internal and External Coronagraph Masks Produced by Various Techniques

Masks for high contrast internal and external coronagraphic imaging require a variety of masks depending on different architectures to suppress star light. Various fabrication technologies are required to address a wide range of needs including gradient amplitude transmission, tunable phase profiles, ultra-low reflectivity, precise small scale features, and low-chromaticity. We present the approaches employed at JPL to produce pupil plane and image plane coronagraph masks, and lab-scale external occulter type masks by various techniques including electron beam, ion beam, deep reactive ion etching, and black silicon technologies with illustrative examples of each. Further development is in progress to produce circular masks of various kinds for obscured aperture telescopes.

Phase Induced Amplitude Apodization Complex Mask C↗
NASA NTRS2022

Charging Experiments and Filtration Testing on Facemask Materials: Towards Goals of Full Performance Restoration of Used N95 Masks and Enhancement of Common-Fabric Facemasks

Different techniques have been developed for sanitizing N95 masks while retaining whatever charge remains on them. However, these sanitizing methods do not address the fact that the electrical charge, filtration efficacy, and breathability of such masks may be expected to degrade during long-term use based on mechanisms known in the literature. In this study, approaches to electrically (re)charging facemask filter materials were explored in response to the COVID-19 induced shortage of N95 masks. The effort consisted of charging experiments followed by filtration testing conducted at the coupon level during a month-long period (May 2020). The main focus of the work was to develop a recharging technique which could eventually be adapted and used as part of a broader full-performance restoration process for used N95 masks, e.g., which might employ simple washing by alcohol or soapy water as a preceding step for sanitization and declogging. Most tested coupons were therefore nonwoven meltblown fibrous polypropylene (PP) filter media extracted from decades old N95 masks. Keeping in mind commonly used cloth facemasks, limited trials were also performed with some commercially available fabrics. Charging was done at elevated temperature and was followed by cooling to freeze in the applied charges produced from corona discharges using different electrode configurations and parameter combinations, or from triboelectric rubbing. Charging was also focused on using potentially scalable equipment, e.g., Van de Graaff generators or consumer ionic hair products. Some degree of success was achieved with one of the fabrics and the PP filter media which was first purposefully discharged to have typical filtration efficiency (FE) of ~35%. Numerous attempts at corona charging discharged media resulted in filtration efficiency (FE) ≥ 80 – 85%. Some measure of charge robustness was also established via stable performance during loading testing. However, results where FE ≥ 90% were not reliably repeatable and no results met the FE ≥ 95% goal due to shortcomings of the charging process. The best three trials for PP media statistically met the FE ≥ 90% criteria, amounted to a small percentage of the trials relying on an ionic hair dryer, and are attributed to good embedding of charges via aerodynamic force. Triboelectric charging significantly enhanced the viability of one of the polyester fabrics –measured FE increased from 19% up to 46% while maintaining a reasonably low pressure drop.

N95↗
NASA NTRSDate not supplied

Development of a Consistent Cross-Platform GEO-Satellite Cloud Mask to Support CERES

Geostationary satellites provide continuous cloud and meteorological observations over a fixed portion of the Earth’s surface, allowing them to monitor the movement of storm systems and their diurnal variation. For climate studies, geostationary observations provide valuable insight of cloud formation and evolution and how they influence the Earth’s radiation budget. The Geo-Satellite Edition 4 cloud mask (GEO Ed4) is used operationally in NASA’s Cloud and Earth’s Radiant Energy System (CERES) project to help account for diurnal variations in cloudiness on Earth’s radiation budget. The Ed4 cloud mask was applied to imager data on MSG (MeteoSat Second Generation), Himawari, GOES-West, and GOES-East satellites using as much spectral information as possible. That approach led to some discontinuities when a more modern satellite replaced an older satellite with less spectral information. A different strategy is being investigated for the CERES GEO Ed5 cloud mask that only uses spectral channels common on every satellite. Thus, a 3-channel (0.6, 3.9, 11 µm) algorithm for daytime cloud detection, and a 2-channels (3.9 and 11 µm) algorithm for nighttime cloud detection have been implemented and tested for Ed5. The goal of Ed5 cloud mask is to achieve global cloud amount consistency across five geo-satellites, and a smooth transition from an old satellite to a new satellite over each geo-location. This paper compares cloud mask results between Ed4 and a preliminary Ed5 version over the GOES-East region, using GOES-8, GOES-13, and GOES-16 satellites, each with a different spectral channel complement. Instantaneous and monthly regional mean inter-comparisons are performed and evaluated with CALIPSO cloud products to assess the accuracy and consistency of the Ed5 approach relative to that taken in Ed4. Results will be presented and discussed at the conference.

clouds↗
NASA NTRSDate not supplied

Progress Towards a Common CERES Cloud Mask Algorithm for MODIS and VIIRS: Evaluation using CALIOP Data

For over two decades The Clouds and the Earth’s Radiant Energy System (CERES) project has endeavored to produce a long-term global climate data record for detecting changes in the Earth’s radiation budget and to improve understanding of how clouds contribute to those changes. CERES incorporates cloud information derived from passive narrowband satellite imaging radiometers, and over the course of the project different instruments have contributed to this effort, namely the Moderate Resolution Imaging Spectroradiometer (MODIS) and the Visible Infrared Imaging Radiometer Suite (VIIRS).Cloud properties have been derived from MODIS on the Aqua satellite for CERES since 2002,and VIIRS on the NOAA-20 satellite launched in 2017 will continue the cloud record once Aqua-MODIS reaches the end of its operational lifetime. However, MODIS and VIIRS have different spectral capabilities and characteristics which complicates our ability to seamlessly transition the record from MODIS to VIIRS. This study evaluates some of the recent algorithm modifications incorporated by the CERES Cloud Working Group towards developing a unified cloud mask algorithm for MODIS and VIIRS that utilizes a set of spectral bands common to both instruments. Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) data products are used to evaluate MODIS and VIIRS regional cloud fraction estimates and identify areas where improvements can be made. The cloud mask algorithm relies on computed estimates of cloud-free top-of-atmosphere radiances to differentiate cloudy and cloud-free imager pixels. The unified cloud mask algorithm currently under development uses a reduced number of spectral bands common to both MODIS and VIIRS and so relies on the accuracy of these simulated cloud-free radiances more heavily than predecessor algorithms, i.e., the CERES Edition 4 cloud mask, which use as much spectral information as available on each imager with less consideration for cross-platform consistency. Our validation strategy using CALIOP data is also described. The CALIOP observations are critical for assessing cloud detection accuracies and for independently confirming cloud-free conditions which enables a more robust evaluation of the simulated cloud-free radiances for biases due to factors such as view angle and the spectral dependence of water vapor absorption. These effects are known to differ for the two satellite instruments. Comparisons of MODIS and VIIRS cloud fractions are presented in context with estimates from CALIOP and from the CERES Edition 4 cloud mask algorithm to gauge current progress in developing accurate and consistent MODIS and VIIRS cloud properties for CERES.

CERES↗
NASA NTRS2023

Toward Physics-informed Neural Networks for 3D Multi-layer Cloud Mask Reconstruction

Three-dimensional (3D) cloud retrievals are critical for understanding their impact on climate and other applications such as aviation safety, weather prediction, and remote sensing. However, obtaining high-resolution and accurate vertical representation of clouds remains unsolved due to the limitations imposed by satellite instrumentation, viewing conditions, and the complexity of cloud dynamics. Cloud masks are essential for comprehending various cloud vertical properties, but deriving accurate 3D cloud masks from 2D satellite imagery data is a challenging task. To tackle these challenges, we introduce a physics-informed loss function for training deep learning models that can extend 2D cloud images into 3D cloud masks. The proposed loss, called CloudMask Loss, is composed of two domain knowledge-informed loss terms: one for evaluating cloud position and thickness, and the other for measuring the number of layers. By combining these loss terms, we improve the trainability of the deep learning models for more accurate and meaningful results. We apply the proposed loss function to different neural networks and demonstrate significant improvements in multi-layer cloud mask reconstruction. Utilizing the same neural network architecture, our proposed loss outperforms standard binary crossentropy loss in terms of multi-layer cloud classification accuracy, number of layers accuracy, and thickness mean absolute error (MAE). The proposed loss function can be readily integrated into various neural network architectures, resulting in substantial performance gains in 3D cloud mask generation.

multi-layer clouds↗