SEARCH · Search NASA
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.
Algorithms for the solution of spatio-temporal recognition and classification problems are known to require intense image data computation.
Pumice rafts are an occasional byproduct of explosive volcanic eruptions that occur near bodies of water. Rafts can be tens of kilometers across and are a hazard to shipping, as the pumice can block the seawater intake valves for ships’ engine cooling systems. In particular, this research focuses on pumice rafts that emanated from the eruptions near Tonga in 2006, Yemen in 2007, and the Chile-Argentina border in 2011 using data from Landsat 5 & 7, MODIS onboard the Aqua and Terra satellites, and ALI aboard the EO-1 satellite. False color composites were created to facilitate visual identification of pumice rafts in high resolution images. Additionally, a material of interest subpixel classification was implemented to automatically demarcate the extent of the pumice rafts. MODIS was found to be the most useful sensor for tracking larger rafts in the open ocean, while Landsat 5 & 7 and ALI were more effective in tracking smaller rafts in lakes and rivers. The pumice rafts generally took on a temperature between that of the ambient air and water temperature. The data collected for this study has been compiled to provide an effective tool for further investigation of pumice rafts.
The Mark Twain National Forest (MTNF) encompasses 1.5 million acres of public land in the Ozarks region of southeastern Missouri. The industrial boom between the 1880s and 1920s had devastating effects on the shortleaf pine (Pinus echinata),Missouri’s only native pine species. The combination of fire suppression on this fire-dependent species and timber harvest of mature pine stands inhibited the development of pine seedlings and promoted the establishment of hardwood stands. Partners at the US Forest Service’s MTNF are currently involved in restoration efforts in two ranger districts, which includes removing invasive eastern red cedar (Juniperus virginiana) and prescribed burning. To expand spatial coverage for the MTNF beyond in situ observation sites, the NASA DEVELOP team analyzed land cover change from 1986 through 2019 and forecasted changes based on a ‘business-as-usual' scenario out to 2040. The team incorporated remotely sensed data from Landsat 5 Thematic Mapper (TM) and Landsat 8 Operational Land Imager (OLI) into the random trees supervised classification tool in ArcGIS Pro. This tool spectrally separated pixels into five distinct land cover classes and produced classifications for 1986 and 2019, with kappa statistics of 0.87 and 0.81, respectively. Overall, there was a net decrease in conifer and meadow land cover between 1986 and 2019 along with a net increase in water, developed, and deciduous land cover. The team used TerrSet’s Land Change Modeler to forecast land cover through 2040. Results showed an increase in coniferous land cover and a decrease in deciduous cover, indicating a high probability that current restoration efforts will produce the intended effect.
The Mark Twain National Forest (MTNF) encompasses 1.5 million acres of public land in the Ozarks region of southeastern Missouri. The industrial boom between the 1880s and 1920s had devastating effects on the shortleaf pine (Pinus echinata),Missouri’s only native pine species. The combination of fire suppression on this fire-dependent species and timber harvest of mature pine stands inhibited the development of pine seedlings and promoted the establishment of hardwood stands. Partners at the US Forest Service’s MTNF are currently involved in restoration efforts in two ranger districts, which includes removing invasive eastern red cedar (Juniperus virginiana) and prescribed burning. To expand spatial coverage for the MTNF beyond in situ observation sites, the NASA DEVELOP team analyzed land cover change from 1986 through 2019 and forecasted changes based on a ‘business-as-usual' scenario out to 2040. The team incorporated remotely sensed data from Landsat 5 Thematic Mapper (TM) and Landsat 8 Operational Land Imager (OLI) into the random trees supervised classification tool in ArcGIS Pro. This tool spectrally separated pixels into five distinct land cover classes and produced classifications for 1986 and 2019, with kappa statistics of 0.87 and 0.81, respectively. Overall, there was a net decrease in conifer and meadow land cover between 1986 and 2019 along with a net increase in water, developed, and deciduous land cover. The team used TerrSet’s Land Change Modeler to forecast land cover through 2040. Results showed an increase in coniferous land cover and a decrease in deciduous cover, indicating a high probability that current restoration efforts will produce the intended effect.
Digitally processed SEASAT synthetic aperture raar (SAR) imagery of the Denver, Colorado urban area was examined to explore the potential of SAR data for mapping urban land cover and the compatability of SAR derived land cover classes with the United States Geological Survey classification system. The imagery is examined at three different scales to determine the effect of image enlargement on accuracy and level of detail extractable. At each scale the value of employing a simplistic preprocessing smoothing algorithm to improve image interpretation is addressed. A visual interpretation approach and an automated machine/visual approach are employed to evaluate the feasibility of producing a semiautomated land cover classification from SAR data. Confusion matrices of omission and commission errors are employed to define classification accuracies for each interpretation approach and image scale.
We consider the image fusion problem involving remotely sensed data. We introduce cokriging as a method to perform fusion. We investigate the advantages of fusing Hyperion with ALI. The evaluation is performed by comparing the classification of the fused data with that of input images and by calculating well-chosen quantitative fusion quality metrics. We consider the Invasive Species Forecasting System (ISFS) project as our fusion application. The fusion of ALI with Hyperion data is studies using PCA and wavelet-based fusion. We then propose utilizing a geostatistical based interpolation method called cokriging as a new approach for image fusion.
The CCSDS-123.0-B Low-Complexity Lossless and Near-Lossless Multispectral & Hyperspectral Image Compression standard provides state-of-the-art compression for imaging spectrometer data. Issue 1 of this standard, published in 2012, provides only lossless compression capability. In 2019, Issue 2 was published, adding new features and in particular augmenting the compressor to provide also near-lossless compression, i.e., compression with a user-specified error limit in each spectral band. This presentation will describe the Issue 3 revision currently under development by the CCSDS Data Compression working group. This revision will provide region-of-interest compression capability. That is, a user-provided spatial classification map can be used to specify different data fidelity parameters in different image regions. This would, for example, allow an onboard classification algorithm to identify the most important regions of an image, or regions obscured by clouds, and adjust compression fidelity accordingly in each region to maximize value of imaging spectrometer data returned over constrained space communication links. The revision also aims to add an option that is unrelated to region-of-interest compression but can improve compression effectiveness in some cases.
The imaging of potato crops with satellites is described and evaluated in terms of the commercial application of the remotely sensed data. The identification and analysis of the crops is accomplished with multiple images acquired from the Landsat MSS and TM systems. The data are processed on a PC with image-procesing software which produces images of the seven 1024 x 1024 pixel windows which are subdivided into 21 512 x 512 pixel windows. Maximization of imaged data throughout the year aids in the identification of crop types by IR reflectance. The classification techniques involve the use of six or seven spectral classes for particular image dates. Comparisons with ground-truth data show good agreement; for example, potato fields are identified correctly 90 percent of the time. Acreage estimates and crop-condition assessments can be made from satellite data and used for corrective agricultural action.
The Department of Energy for decades has explored the earth system and atmosphere through research and deployment of in-situ and remote sensing platforms during field campaigns. Among these datasets exists a vast supply of cloud particle images that provide visual insight into the complex microphysics in the clouds that span our globe. The millions of images collected over decades of deployments provides a unique opportunity to further our understanding of our atmosphere down to the crystal size. This work over the past 5 years has sought to organize these images into digestible datasets that can then be used by scientists to further our understanding of microphysics. A machine learning model was developed that categorizes over 1.5 million images across 11 weather events with over 90% accuracy according to particle type. The database was then extended to include dimensional characteristics of the particle as well as co-location of environmental properties, such as temperature and water content. Then, to initialize the connection between these data and our understanding of how crystals form and grow, weather research and forecasting simulations were run to generate the growth histories of the classified crystals. This research culminates with 2 databases per event: (1) a database of all classified crystals and their dimensional and environmental properties and (2) simulated growth histories of each crystal. Finally, a user interface was created to allow researchers to explore data statistics.
The Center for Remote Sensing Research has conducted studies designed to evaluate the potential application of ERTS data in performing agricultural inventories, and to develop efficient methods of data handling and analysis useful in the operational context for performing large area surveys. This work has resulted in the development of an integrated system utilizing both human and computer analysis of ground, aerial, and space imagery, which has been shown to be very efficient for regional crop acreage inventories. The technique involves: (1) the delineation of ERTS images into relatively homogeneous strata by human interpreters, (2) the point-by-point classification of the area within each strata on the basis of crop type using a human/machine interactive digital image processing system; and (3) a multistage sampling procedure for the collection of supporting aerial and ground data used in the adjustment and verification of the classification results.
Automatic Image-100 analysis of LANDSAT data was performed using the MAXVER classification algorithm. In the pilot area, four vegetation units were mapped automatically in addition to the areas occupied for agricultural activities. The Image-100 classified results together with a soil map and information from RADAR images, permitted the establishment of the final legend with six classes: semi-deciduous tropical forest; low land evergreen tropical forest; secondary vegetation; tropical forest of humid areas, predominant pastureland and flood plains. Two water types were identified based on their sediments indicating different geological and geomorphological aspects.
With the advent of the EOS era and of configurable sensors, users of these instruments are faced with the twin problems of specifying data acquisition parameters and extracting desired information from the voluminous data. An application of a system model is made to explore system parameter trade-offs for a model sensor based on the High Resolution Imaging Spectrometer. Radiometric performance was studied, along with the effect on classification accuracy of several system parameters. Using a model scene based on typical agricultural reflectance and atmospheric conditions, the atmosphere and sensor are seen to have significant effects on the mean received signal and noise performance. The effect of random uncorrelated errors in the radiometric calibration of the detector array is seen to degrade system performance, especially in the spectral bands below 1 micron. Accurate pixel-to-pixel relative radiometric calibration and the use of the Image Motion Compensation option are seen to improve classification accuracy, especially at high solar zenith angles. Feature sets chosen from characteristics of the scene performed best overall, but ones chosen based on signal-to-noise ratios were seen to be more robust.
The anticipated expansion of the nuclear industry and the deployment of new nuclear reactors (200 + GW of new nuclear capacity by 2050) require the development of monitoring systems that align with safety and security concerns, providing enhanced evaluation capabilities. A remote monitoring system using satellites and deep learning techniques was evaluated for its ability to detect anomalies and capture various features of nuclear reactors independently of the conditions on the ground. Satellite images of current operational and under-construction nuclear power plants were collected from Google Earth Pro as a surrogate database. Subsequently, five datasets were created from the collected images. Transfer learning technique was used for several classification tasks utilizing VGG16, ResNet50V2, Xception, DenseNet121, and MobileNetV2 pre-trained models. In the first task, the capability of the monitoring system to detect abnormal conditions or processes in a nuclear power plant was investigated. In the second task, the ability to capture operational features remotely was examined. As an example, for the purposes of this study, these features included classifying reactors based on type, power range, or onsite condition. Several evaluation metrics were used to compare the performance of the pre-trained models and the overall monitoring system. Here, the evaluation results demonstrated that deep learning techniques and pre-trained models applied to satellite images have the potential to facilitate further and expand capabilities in monitoring systems to assess plant operation details.
Tidal marshes are highly productive ecosystems that support migratory birds as roosting and over-wintering habitats on the Pacific Flyway. Microphytobenthos, or more commonly 'biofilms' contribute significantly to the primary productivity of wetland ecosystems, and provide a substantial food source for macroinvertebrates and avian communities. In this study, biofilms were characterized based on taxonomic classification, density differences, and spectral signatures. These techniques were then applied to remotely sensed images to map biofilm densities and distributions in the South Bay Salt Ponds and predict the carrying capacity of these newly restored ponds for migratory birds. The GER-1500 spectroradiometer was used to obtain in situ spectral signatures for each density-class of biofilm. The spectral variation and taxonomic classification between high, medium, and low density biofilm cover types was mapped using in-situ spectral measurements and classification of EO-1 Hyperion and Landsat TM 5 images. Biofilm samples were also collected in the field to perform laboratory analyses including chlorophyll-a, taxonomic classification, and energy content. Comparison of the spectral signatures between the three density groups shows distinct variations useful for classification. Also, analysis of chlorophyll-a concentrations show statistically significant differences between each density group, using the Tukey-Kramer test at an alpha level of 0.05. The potential carrying capacity in South Bay Salt Ponds is estimated to be 250,000 birds.
Astronaut photography acquired from the International Space Station (ISS) using commercial off-the-shelf cameras offers a freely-accessible source for high to very high resolution (4-20 m/pixel) visible-wavelength digital data of Earth. Since ISS Expedition 1 in 2000, over 373,000 images of the Earth-Moon system (including land surface, ocean, atmospheric, and lunar images) have been added to the Gateway to Astronaut Photography of Earth online database (http://eol.jsc.nasa.gov ). Handheld astronaut photographs vary in look angle, time of acquisition, solar illumination, and spatial resolution. These attributes of digital astronaut photography result from a unique combination of ISS orbital dynamics, mission operations, camera systems, and the individual skills of the astronaut. The variable nature of astronaut photography makes the dataset uniquely useful for archaeological applications in comparison with more traditional nadir-viewing multispectral datasets acquired from unmanned orbital platforms. For example, surface features such as trenches, walls, ruins, urban patterns, and vegetation clearing and regrowth patterns may be accentuated by low sun angles and oblique viewing conditions (Fig. 1). High spatial resolution digital astronaut photographs can also be used with sophisticated land cover classification and spatial analysis approaches like Object Based Image Analysis, increasing the potential for use in archaeological characterization of landscapes and specific sites.
The Curiosity Rover landed in a lithologically and geochemically diverse region of Mars. We present a recommended rock classification framework based on terrestrial schemes, and adapted for the imaging and analytical capabilities of MSL as well as for rock types distinctive to Mars (e.g., high Fe sediments). After interpreting rock origin from textures, i.e., sedimentary (clastic, bedded), igneous (porphyritic, glassy), or unknown, the overall classification procedure (Fig 1) involves: (1) the characterization of rock type according to grain size and texture; (2) the assignment of geochemical modifiers according to Figs 3 and 4; and if applicable, in depth study of (3) mineralogy and (4) geologic/stratigraphic context. Sedimentary rock types are assigned by measuring grains in the best available resolution image (Table 1) and classifying according to the coarsest resolvable grains as conglomerate/breccia, (coarse, medium, or fine) sandstone, silt-stone, or mudstone. If grains are not resolvable in MAHLI images, grains in the rock are assumed to be silt sized or smaller than surface dust particles. Rocks with low color contrast contrast between grains (e.g., Dismal Lakes, sol 304) are classified according to minimum size of apparent grains from surface roughness or shadows outlining apparent grains. Igneous rocks are described as intrusive or extrusive depending on crystal size and fabric. Igneous textures may be described as granular, porphyritic, phaneritic, aphyric, or glassy depending on crystal size. Further descriptors may include terms such as vesicular or cumulate textures.
An essential component in global climate research is accurate cloud cover and type determination. Of the two approaches to texture-based classification (statistical and textural), only the former is effective in the classification of natural scenes such as land, ocean, and atmosphere. In the statistical approach that was adopted, parameters characterizing the stochastic properties of the spatial distribution of grey levels in an image are estimated and then used as features for cloud classification. Two types of textural measures were used. One is based on the distribution of the grey level difference vector (GLDV), and the other on a set of textural features derived from the MaxMin cooccurrence matrix (MMCM). The GLDV method looks at the difference D of grey levels at pixels separated by a horizontal distance d and computes several statistics based on this distribution. These are then used as features in subsequent classification. The MaxMin tectural features on the other hand are based on the MMCM, a matrix whose (I,J)th entry give the relative frequency of occurrences of the grey level pair (I,J) that are consecutive and thresholded local extremes separated by a given pixel distance d. Textural measures are then computed based on this matrix in much the same manner as is done in texture computation using the grey level cooccurrence matrix. The database consists of 37 cloud field scenes from LANDSAT imagery using a near IR visible channel. The classification algorithm used is the well known Stepwise Discriminant Analysis. The overall accuracy was estimated by the percentage or correct classifications in each case. It turns out that both types of classifiers, at their best combination of features, and at any given spatial resolution give approximately the same classification accuracy. A neural network based classifier with a feed forward architecture and a back propagation training algorithm is used to increase the classification accuracy, using these two classes of features. Preliminary results based on the GLDV textural features alone look promising.
We previously presented Soil Property and Object Classification (SPOC), a machine learning-based terrain classifier for Mars rovers, for automatically segmenting rover images by its surface type such as sand and bedrock. This paper presents a number of practical improvements to pave the way for potential future onboard deployment. First, we achieved 97.0% overall pixel accuracy, evaluated against the classification generated by human experts on images from Mars Science Laboratory (MSL) missions. The substantial increase in accuracy was primarily enabled by the sheer volume of data used for training; we created a new large-scale dataset of Martian terrain labels, namely AI4Mars, which contains more than 400k labels contributed by citizen scientists for 50k images taken by the Mars Exploration Rovers (MER) and Mars Science Laboratory (MSL) rover. Second, we demonstrated that SPOC can quickly adapt to a new mission landed on a previously unseen site. Specifically, we pretrained a model with MER and MSL data from the AI4Mars dataset and then adapted to the Mars 2020 Rover (M2020) by feeding a small volume of data between Sol 0 and 157; the adapted model was tested on Sol 200-203 and resulted in 84.2% overall pixel accuracy and 93.4% reliability (recall) for detecting sand, the most concerning class for rover’s traversability. Third, we found that pretraining can substantially mitigate the decline of accuracy over time. We showed that the performance of a SPOC model pretrained with the ImageNet dataset and then trained by MSL images only up to Sol 390 remains comparable to a model trained by images up to Sol 1689 on the test data after Sol 1689. Fourth, we reimplemented SPOC with a light-weight convolutional neural network (CNN), MobileNetV2, which typically runs within tens of milliseconds (ms) on mobile processors such as Qualcomm’s Snapdragon. Finally, we released the AI4Mars dataset to the public to encourage open innovation.