Engineering topics
Publications and source records attributed to Tilton, James C..
The presentation stressed the importance of a pre-launch test plan and a post-launch CalVal plan in the geometric aspects of instrument focal length, spatial responses, band-to-band registration, pointing, geolocation and long-term monitoring.
Two Moderate Resolution Imaging Spectroradiometer (MODIS) sensors have been in operations for more than 19 and 17 years (thus 36 combined years) as part of NASA's Earth Observing System (EOS) on the Terra platform that was launched in December 1999 and on the Aqua platform that was launched in May 2002, respectively. Accurate geolocation is a critical element needed for accurate retrieval of global biogeophysical parameters. In this paper, we describe the latest trends in the continuously improved MODIS geolocation accuracy in Collection-5 (C5), C6 and C6.1 re-processing and forward-processing data streams. We improved geolocation accuracy in the re-processed data and corrected for geolocation biases found in forward-processed data, including those caused by operations such as the stop-go-stop status of the Advanced Microwave Scanning Radiometer for EOS (AMSR-E) instrument on the Aqua platform. We discuss scan-toscan underlaps near nadir over the equator regions that was discovered in checking the non-underlapping requirement in the Visible Infrared Imaging Radiometer Suite (VIIRS) based on trending parameters from the actual Suomi National Polar-orbiting Partnership (S-NPP) satellite orbit. The underlaps are closely tied to instrument effective focal length that is measured from on-orbit data using a technique we recently developed. We also discuss potential improvements for the upcoming C7 re-processing.
Accurate information on cropland spatial distribution is required for global-scale assessments and agricultural land use policies. Cloud computing platforms such as Google Earth Engine (GEE) provide unprecedented opportunities for large-scale classifications of Landsat data. We developed a novel method to fuse pixel-based random forest classification of continental-scale Landsat data on GEE and an object-based segmentation approach known as recursive hierarchical segmentation (RHSeg). Using our fusion method, we produced a continental-scale cropland extent map for North America at 30m spatial resolution for the nominal year 2010. The total cropland area for North America was estimated at 275.18 million hectares (Mha). The overall accuracies of the map are>90% across the continent. This map also compares well with the United States Department of Agriculture (USDA) cropland data layer (CDL), Agriculture and Agri-food Canada (AAFC) annual crop inventory (ACI), and the Mexican government agency Servicio de Informacion Agroalimentaria y Pesquera (SIAP)'s agricultural boundaries. Furthermore, our map compared well with sub-country statistics including state-wise and county-wise cropland statistics in regression models resulting in R2 > 0.84. This key contribution paves the way for more detailed products such as crop intensity, crop type, and crop irrigation, and provides a method for creating high-resolution cropland extent maps for other countries where spatial information about croplands are not as prevalent.
Hyperspectral data are becoming more widely available via sensors on airborne and unmanned aerial vehicle (UAV) platforms, as well as proximal platforms. While space-based hyperspectral data continue to be limited in availability, multiple spaceborne Earth-observing missions on traditional platforms are scheduled for launch, and companies are experimenting with small satellites for constellations to observe the Earth, as well as for planetary missions. Land cover mapping via classification is one of the most important applications of hyperspectral remote sensing and will increase in significance as time series of imagery are more readily available. However, while the narrow bands of hyperspectral data provide new opportunities for chemistry-based modeling and mapping, challenges remain. Hyperspectral data are high dimensional, and many bands are highly correlated or irrelevant for a given classification problem. For supervised classification methods, the quantity of training data is typically limited relative to the dimension of the input space. The resulting Hughes phenomenon, often referred to as the curse of dimensionality, increases potential for unstable parameter estimates, overfitting, and poor generalization of classifiers. This is particularly problematic for parametric approaches such as Gaussian maximum likelihood–based classifiers that have been the backbone of pixel-based multispectral classification methods. This issue has motivated investigation of alternatives, including regularization of the class covariance matrices, ensembles of weak classifiers, development of feature selection and extraction methods, adoption of nonparametric classifiers, and exploration of methods to exploit unlabeled samples via semi-supervised and active learning. Data sets are also quite large, motivating computationally efficient algorithms and implementations. This chapter provides an overview of the recent advances in classification methods for mapping vegetation using hyperspectral data. Three data sets that are used in the hyperspectral classification literature (e.g., Botswana Hyperion satellite data and AVIRIS airborne data over both Kennedy Space Center and Indian Pines) are described in Section 3.2 and used to illustrate methods described in the chapter. An additional high-resolution hyperspectral data set acquired by a SpecTIR sensor on an airborne platform over the Indian Pines area is included to exemplify the use of new deep learning approaches, and a multiplatform example of airborne hyperspectral data is provided to demonstrate transfer learning in hyperspectral image classification. Classical approaches for supervised and unsupervised feature selection and extraction are reviewed in Section 3.3. In particular, nonlinearities exhibited in hyperspectral imagery have motivated development of nonlinear feature extraction methods in manifold learning, which are outlined in Section 3.3.1.4. Spatial context is also important in classification of both natural vegetation with complex textural patterns and large agricultural fields with significant local variability within fields. Approaches to exploit spatial features at both the pixel level (e.g., co-occurrence–based texture and extended morphological attribute profiles [EMAPs]) and integration of segmentation approaches (e.g., HSeg) are discussed in this context in Section 3.3.2. Recently, classification methods that leverage nonparametric methods originating in the machine learning community have grown in popularity. An overview of both widely used and newly emerging approaches, including support vector machines (SVMs), Gaussian mixture models, and deep learning based on convolutional neural networks is provided in Section 3.4. Strategies to exploit unlabeled samples, including active learning and metric learning, which combine feature extraction and augmentation of the pool of training samples in an active learning framework, are outlined in Section 3.5. Integration of image segmentation with classification to accommodate spatial coherence typically observed in vegetation is also explored, including as an integrated active learning system. Exploitation of multisensor strategies for augmenting the pool of training samples is investigated via a transfer learning framework in Section 3.5.1.2. Finally, we look to the future, considering opportunities soon to be provided by new paradigms, as hyperspectral sensing is becoming common at multiple scales from ground-based and airborne autonomous vehicles to manned aircraft and space-based platforms.
The first NOAA/NASA Join Polar Satellite System (JPSS-1) satellite was successfully launched on November 18, 2017,becoming NOAA-20. Instruments on-board the NOAA-20 satellite include the Visible Infrared Imaging RadiometerSuite (VIIRS). This instrument is the second build of VIIRS, with the first flight instrument on-board NASA/NOAASuomi National Polar-orbiting Partnership (SNPP) satellite operating since October 2011. The purpose of these VIIRSinstruments is to continue the long-term measurements of biogeophysical variables for multiple applications includingweather forecasting, rapid response and climate research. The geometric performance of VIIRS is essential to retrievingaccurate biogeophysical variables. This paper describes the early on-orbit geometric performance of the JPSS-1/NOAA-20 VIIRS. It first discusses the on-orbit orbit and attitude performance, a key input needed for accurate geolocation. Itthen discusses the on-orbit geometric characterization and calibration of VIIRS and an initial assessment of thegeometric accuracy. It follows with a discussion of an improvement in the instrument geometric model that correctssmall geometrical artifacts that appear in the along-scan direction. Finally, this paper discusses on-orbit measurements ofthe focal length and the impact of this on the scan-to-scan underlap/overlap.
The Visible Infrared Imaging Radiometer Suite (VIIRS) instrument is a whiskbroom system with 22 spectral bands split between 16 moderate resolution bands (M-bands), five imagery resolution bands (I-bands) and a panchromatic day-night band. Latitude and Longitude geolocation data are generated for each pixel at the M-band, I-band and day-night band spatial resolutions based upon various instrument parameters including focal length. In this study we measure the focal length of the VIIRS instrument from on-orbit data. This is achieved by simulating VIIRS band I2 using Landsat 8 OLI band 5 utilizing the VIIRS instrument system point spread function (PSF) and geolocation data generated with varying values of focal length. The focal length value that produces the highest spatial correlation between the original and simulated VIIRS data is taken to be the measured instrument focal length.
A satellite-derived cropland extent map at high spatial resolution (30-m or better) is a must for food and water security analysis. Precise and accurate global cropland extent maps, indicating cropland and non-cropland areas, is a starting point to develop high-level products such as crop watering methods (irrigated or rainfed), cropping intensities (e.g., single, double, or continuous cropping), crop types, cropland fallows, as well as assessment of cropland productivity (productivity per unit of land), and crop water productivity (productivity per unit of water). Uncertainties associated with the cropland extent map have cascading effects on all higher-level cropland products. However, precise and accurate cropland extent maps at high spatial resolution over large areas (e.g., continents or the globe) are challenging to produce due to the small-holder dominant agricultural systems like those found in most of Africa and Asia. Cloud-based Geospatial computing platforms and multi-date, multi-sensor satellite image inventories on Google Earth Engine offer opportunities for mapping croplands with precision and accuracy over large areas that satisfy the requirements of broad range of applications. Such maps are expected to provide highly significant improvements compared to existing products, which tend to be coarser in resolution, and often fail to capture fragmented small-holder farms especially in regions with high dynamic change within and across years. To overcome these limitations, in this research we present an approach for cropland extent mapping at high spatial resolution (30-m or better) using the 10-day, 10 to 20-m, Sentinel-2 data in combination with 16-day, 30-m, Landsat-8 data on Google Earth Engine (GEE). First, nominal 30-m resolution satellite imagery composites were created from 36,924 scenes of Sentinel-2 and Landsat-8 images for the entire African continent in 2015-2016. These composites were generated using a median-mosaic of five bands (blue, green, red, near-infrared, NDVI) during each of the two periods (period 1: January-June 2016 and period 2: July-December 2015) plus a 30-m slope layer derived from the Shuttle Radar Topographic Mission (SRTM) elevation dataset. Second, we selected Cropland/Non-cropland training samples (sample size 9791) from various sources in GEE to create pixel-based classifications. As supervised classification algorithm, Random Forest (RF) was used as the primary classifier because of its efficiency, and when over-fitting issues of RF happened due to the noise of input training data, Support Vector Machine (SVM) was applied to compensate for such defects in specific areas. Third, the Recursive Hierarchical Segmentation (RHSeg) algorithm was employed to generate an object-oriented segmentation layer based on spectral and spatial properties from the same input data. This layer was merged with the pixel-based classification to improve segmentation accuracy. Accuracies of the merged 30-m crop extent product were computed using an error matrix approach in which 1754 independent validation samples were used. In addition, a comparison was performed with other available cropland maps as well as with LULC maps to show spatial similarity. Finally, the cropland area results derived from the map were compared with UN FAO statistics. The independent accuracy assessment showed a weighted overall accuracy of 94, with a producers accuracy of 85.9 (or omission error of 14.1), and users accuracy of 68.5 (commission error of 31.5) for the cropland class. The total net cropland area (TNCA) of Africa was estimated as 313 Mha for the nominal year 2015.
The visible infrared imaging radiometer suite (VIIRS) instrument was launched 28 October 2011 onboard the Suomi National Polar-orbiting Partnership (SNPP) satellite. The VIIRS instrument is a whiskbroom system with 22 spectral and thermal bands split between 16 moderate resolution bands (M-bands), five imagery resolution bands (I-bands) and a day-night band. In this study we estimate the along-scan line spread function (LSF) of the I-bands and M-bands based on measurements performed on images of the Lake Pontchartrain Causeway Bridge. In doing so we develop a model for the LSF that closely matches the prelaunch laboratory measurements. We utilize VIIRS images co-geolocated with a Landsat TM image to precisely locate the bridge linear feature in the VIIRS images as a linear best fit to a straight line. We then utilize non-linear optimization to compute the best fit equation of the VIIRS image measurements in the vicinity of the bridge to the developed model equation. From the found parameterization of the model equation we derive the full-width at half-maximum (FWHM) as an approximation of the sensor field of view (FOV) for all bands, and compare these on-orbit measured values with prelaunch laboratory results.
Understanding the impacts of urbanization requires accurate and updatable urban extent maps. Here we present an algorithm for mapping urban extent at global scale using Landsat data. An innovative hierarchical object-based texture (HOTex) classification approach was designed to overcome spectral confusion between urban and nonurban land cover types. VIIRS nightlights data and MODIS vegetation index datasets are integrated as high-level features under an object-based framework. We applied the HOTex method to the GLS-2010 Landsat images to produce a global map of human built-up and settlement extent. As shown by visual assessments, our method could effectively map urban extent and generate consistent results using images with inconsistent acquisition time and vegetation phenology. Using scene-level cross validation for results in Europe, we assessed the performance of HOTex and achieved a kappa coefficient of 0.91, compared to 0.74 from a baseline method using per-pixel classification using spectral information.
This is the first study to quantitatively assess and compare cross correlation and normalized mutual information methods used to register images in subpixel scale. The study shows that the normalized mutual information method is less sensitive to unaligned edges due to the spectral response differences than is cross correlation. This characteristic makes the normalized image resolution a better candidate for band to band registration. Improved band-to-band registration in the data from satellite-borne instruments will result in improved retrievals of key science measurements such as cloud properties, vegetation, snow and fire.
This paper describes trends in the Suomi National Polar-orbiting Partnership (SNPP) spacecraft ephemeris data over the four and half years of on-orbit operations. It then discusses the implications of these trends on the geometric performance of the Visible Infrared Imaging Radiometer Suite (VIIRS), one of the instruments onboard SNPP. The SNPP ephemeris data includes time stamped spacecraft positions and velocities that are used to calculate the spacecraft altitude and sub-satellite locations. Through drag make-up maneuvers (DMUs) the orbital mean altitude (spacecraft altitude averaged over an orbit) has been maintained at 838.8 kilometers to within plus or minus 0.2 kilometers and the orbital period at 101.5 minutes to within plus or minus 0.2 seconds. The corresponding orbital mean velocity in the terrestrial frame of reference has been maintained at 7524 meters per second to within plus or minus 0.5 meters per second. Within an orbit, the altitude varies from 828 kilometers near 15 degrees North latitude to 856 kilometers near the South Pole. Inclination adjust maneuvers (IAMs) have maintained the orbit inclination angle at 98.67 degrees to within plus or minus 0.07 degrees and the sun-synchronous local time at ascending node (LTAN) at 13:28 to within plus or minus 5 minutes. Besides these trends, it is interesting to observe that the orbit's elliptic shape has its major axis linking the perigee and apogee shorter than the line linking the ascending node and the descending node. This effect is caused by the Earth's oblate spheroid shape and deviates from a Keplerian orbit theory in which the two orbiting bodies are point masses. VIIRS has 5 imagery resolution bands, 16 moderate resolution bands and a day-night band, with 32, 16 and 16 detectors, respectively, aligned in the spacecraft flight (also known as track) direction. For each band's sample within a scan, the detectors sample the Earth's surface simultaneously in the track direction in the Earth Centered Inertial frame of reference. The distance between the center of the area sensed by the trailing detectors of one scan and the leading detectors of the next includes a component caused by earth rotation. This earth rotation component is relatively small (approximately 70 meters per second) for an orbit like SNPP, but must be taken into account in the design of low-Earth orbit scanning sensors similar to VIIRS to ensure contiguous coverage at nadir.
No abstract available
Augmenting spectral data with spatial information for image classification has recently gained significant attention, as classification accuracy can often be improved by extracting spatial information from neighboring pixels. In this paper, we propose a new framework in which active learning (AL) and hierarchical segmentation (HSeg) are combined for spectral-spatial classification of hyperspectral images. The spatial information is extracted from a best segmentation obtained by pruning the HSeg tree using a new supervised strategy. The best segmentation is updated at each iteration of the AL process, thus taking advantage of informative labeled samples provided by the user. The proposed strategy incorporates spatial information in two ways: 1) concatenating the extracted spatial features and the original spectral features into a stacked vector and 2) extending the training set using a self-learning-based semi-supervised learning (SSL) approach. Finally, the two strategies are combined within an AL framework. The proposed framework is validated with two benchmark hyperspectral datasets. Higher classification accuracies are obtained by the proposed framework with respect to five other state-of-the-art spectral-spatial classification approaches. Moreover, the effectiveness of the proposed pruning strategy is also demonstrated relative to the approaches based on a fixed segmentation.
We have previously developed a best merge region-growing approach that integrates nonadjacent region object aggregation with the neighboring region merge process usually employed in region growing segmentation approaches. This approach has been named HSeg, because it provides a hierarchical set of image segmentation results. Up to this point, HSeg considered only global region feature information in the region growing decision process. We present here three new versions of HSeg that include local edge information into the region growing decision process at different levels of rigor. We then compare the effectiveness and processing times of these new versions HSeg with each other and with the original version of HSeg.
Acquiring current, accurate land-use information is critical for monitoring and understanding the impact of anthropogenic activities on natural environments.Remote sensing technologies are of increasing importance because of their capability to acquire information for large areas in a timely manner, enabling decision makers to be more effective in complex environments. Although optical imagery has demonstrated to be successful for land cover classification, active sensors, such as light detection and ranging (LiDAR), have distinct capabilities that can be exploited to improve classification results. However, utilization of LiDAR data for land cover classification has not been fully exploited. Moreover, spatial-spectral classification has recently gained significant attention since classification accuracy can be improved by extracting additional information from the neighboring pixels. Although spatial information has been widely used for spectral data, less attention has been given to LiDARdata. In this work, a new framework for land cover classification using discrete return LiDAR data is proposed. Pseudo-waveforms are generated from the LiDAR data and processed by hierarchical segmentation. Spatial featuresare extracted in a region-based way using a new unsupervised strategy for multiple pruning of the segmentation hierarchy. The proposed framework is validated experimentally on a real dataset acquired in an urban area. Better classification results are exhibited by the proposed framework compared to the cases in which basic LiDAR products such as digital surface model and intensity image are used. Moreover, the proposed region-based feature extraction strategy results in improved classification accuracies in comparison with a more traditional window-based approach.
Impervious surfaces, mainly artificial structures and roads, cover less than 1% of the world's land surface (1.3% over USA). Regardless of the relatively small coverage, impervious surfaces have a significant impact on the environment. They are the main source of the urban heat island effect, and affect not only the energy balance, but also hydrology and carbon cycling, and both land and aquatic ecosystem services. In the last several decades, the pace of converting natural land surface to impervious surfaces has increased. Quantitatively monitoring the growth of impervious surface expansion and associated urbanization has become a priority topic across both the physical and social sciences. The recent availability of consistent, global scale data sets at 30m resolution such as the Global Land Survey from the Landsat satellites provides an unprecedented opportunity to map global impervious cover and urbanization at this resolution for the first time, with unprecedented detail and accuracy. Moreover, the spatial resolution of Landsat is absolutely essential to accurately resolve urban targets such a buildings, roads and parking lots. With long term GLS data now available for the 1975, 1990, 2000, 2005 and 2010 time periods, the land cover/use changes due to urbanization can now be quantified at this spatial scale as well. In the Global Land Survey - Imperviousness Mapping Project (GLS-IMP), we are producing the first global 30 m spatial resolution impervious cover data set. We have processed the GLS 2010 data set to surface reflectance (8500+ TM and ETM+ scenes) and are using a supervised classification method using a regression tree to produce continental scale impervious cover data sets. A very large set of accurate training samples is the key to the supervised classifications and is being derived through the interpretation of high spatial resolution (approx. 2 m or less) commercial satellite data (Quickbird and Worldview2) available to us through the unclassified archive of the National Geospatial Intelligence Agency (NGA). For each continental area several million training pixels are derived by analysts using image segmentation algorithms and tools and then aggregated to the 30m resolution of Landsat. Here we will discuss the production/testing of this massive data set for Europe, North and South America and Africa, including assessments of the 2010 surface reflectance data. This type of analysis is only possible because of the availability of long term 30m data sets from GLS and shows much promise for integration of Landsat 8 data in the future.
The Visible Infrared Imager Radiometer Suite (VIIRS) sensor was launched 28 October 2011 on the Suomi National Polarorbiting Partnership (SNPP) satellite. VIIRS has 22 spectral bands covering the spectrum between 0.412 m and 12.01 m, including 16 moderate resolution bands (M-bands) with a spatial resolution of 750 m at nadir, 5 imaging resolution bands (I-bands) with a spatial resolution of 375 m at nadir, and 1 day-night band (DNB) with a near-constant 750 m spatial resolution throughout the scan. These bands are located in a visible and near infrared (VisNIR) focal plane assembly (FPA), a short- and mid-wave infrared (SWMWIR) FPA and a long-wave infrared (LWIR) FPA. All bands, except the DNB, are co-registered for proper environmental data records (EDRs) retrievals. Observations from VIIRS instrument provide long-term measurements of biogeophysical variables for climate research and polar satellite data stream for the operational communitys use in weather forecasting and disaster relief and other applications. Well Earth-located (geolocated) instrument data is important to retrieving accurate biogeophysical variables. This paper describes prelaunch pointing and alignment measurements, and the two sets of on-orbit correction of geolocation errors, the first of which corrected error from 1,300 m to within 75 m (20 I-band pixel size), and the second of which fine tuned scan angle dependent errors, bringing VIIRS geolocation products to high maturity in one and a half years of the SNPP VIIRS on-orbit operations. Prelaunch calibration and the on-orbit characterization of sensor spatial impulse responses and band-to-band co-registration (BBR) are also described.
The Visible Infrared Imager Radiometer Suite (VIIRS) instrument onboard the Suomi National Polar‐orbiting Partnership (SNPP) satellite was launched on 28 October 2011. The VIIRS has 5 imagery spectral bands (I-bands), 16 moderate resolution spectral bands (M-bands) and a panchromatic day/night band (DNB). Performance of the VIIRS spatial response and band-to-band co-registration (BBR) was measured through intensive pre-launch tests. These measurements were made in the non-aggregated zones near the start (or end) of scan for the I-bands and M-bands and for a limited number of aggregation modes for the DNB in order to test requirement compliance. This paper presents results based on a recently re-processed pre-launch test data. Sensor (detector) spatial impulse responses in the scan direction are parameterized in terms of ground dynamic field of view (GDFOV), horizontal spatial resolution (HSR), modulation transfer function (MTF), ensquared energy (EE) and integrated out-of-pixel (IOOP) spatial response. Results are presented for the non-aggregation, 2-sample and 3-sample aggregation zones for the I-bands and M-bands, and for a limited number of aggregation modes for the DNB. On-orbit GDFOVs measured for the 5 I-bands in the scan direction using a straight bridge are also presented. Band-to-band co-registration (BBR) is quantified using the prelaunch measured band-to-band offsets. These offsets may be expressed as fractions of horizontal sampling intervals (HSIs), detector spatial response parameters GDFOV or HSR. BBR bases on HSIs in the non-aggregation, 2-sample and 3-sample aggregation zones are presented. BBR matrices based on scan direction GDFOV and HSR are compared to the BBR matrix based on HSI in the non-aggregation zone. We demonstrate that BBR based on GDFOV is a better representation of footprint overlap and so this definition should be used in BBR requirement specifications. We propose that HSR not be used as the primary image quality indicator, since we show that it is neither an adequate representation of the size of sensor spatial response nor an adequate measure of imaging quality.