Raymond Soffer

Detection of single graves by airborne hyperspectral imaging

Airborne hyperspectral imaging (HSI) was assessed as a potential tool to locate single grave sites. While airborne HSI has shown to be useful to locate mass graves, it is expected the location of single graves would be an order of magnitude more difficult due to the smaller size and reduced mass of the targets. Two clearings were evaluated (through a blind test) as potential sites for containing at least one set of buried remains. At no time prior to submitting the locations of the potential burial sites from the HSI were the actual locations of the sites released or shared with anyone from the analysis team. The two HSI sensors onboard the aircraft span the range of 408-2524nm. A range of indicators that exploit the narrow spectral and spatial resolutions of the two complimentary HSI sensors onboard the aircraft were calculated. Based on the co-occurrence of anomalous pixels within the expected range of the indicators three potential areas conforming to our underlying assumptions of the expected spectral responses (and spatial area) were determined. After submission of the predicted burial locations it was revealed that two of the targets were located within GPS error (10m) of the true burial locations. Furthermore, due to the history of the TPOF site for burial work, investigation of the third target is being considered in the near future. The results clearly demonstrate promise for hyperspectral imaging to aid in the detection of buried remains, however further work is required before these results can justifiably be used in routine scenarios.

Spectral Reflectance of Polar Bear and Other Large Arctic Mammal Pelts; Potential Applications to Remote Sensing Surveys

Spectral reflectance within the 350–2500 nm range was measured for 17 pelts of arctic mammals (polar bear, caribou, muskox, and ringed, harp and bearded seals) in relation to snow. Reflectance of all pelts was very low at the ultraviolet (UV) end of the spectrum ( 90%), gradually dropped to near zero at 1500 nm, and then fluctuated between zero and 20% up to 2500 nm. All pelts could be distinguished from clean snow at many wavelengths. The polar bear pelts had higher and more uniform averaged reflectance from about 600–1100 nm than most other pelts, but discrimination was challenging due to variation in pelt color and intensity among individuals within each species. Results suggest promising approaches for using remote sensing tools with a broad spectral range to discriminate polar bears and other mammals from clean snow. Further data from live animals in their natural environment are needed to develop functions to discriminate among species of mammals and to determine whether other environmental elements may have similar reflectance.

Quality Control Assessment of the Mission Airborne Carbon 13 (MAC-13) Hyperspectral Imagery from Costa Rica

Abstract. A data quality assessment of airborne hyperspectral imagery (HSI) from Mission Airborne Carbon 2013 (MAC13) is presented. Because data quality is fundamentally important for modeling landscape biophysical characteristics from HSI, this article presents an assessment related to spectral alignment, spectroradiometric calibration, and geocorrection for 2,700 km2 of imagery acquired with the CASI-1500 and SASI-644 systems (375 nm – 2523 nm, 2.5 m resampled pixel size). MODIS, in-situ and image-based estimations of aerosol optical depth are compared for calculations of visibility for atmospheric correction. Information content (dimensionality) across the 5 ecosystems and 2 developed areas are also compared to illustrate the benefit of the extensive spectral resolution of the data. New approaches to the offset corrections of the imagery improved the accuracy of the calibrated results (radiance and reflectance). Assessment of visibility values applied to the atmospheric correction adduced that apparent reflectance computed using in-scene modeled visibility produced the most similar results to ground spectra. Dimensionality analysis revealed increased information content for all ecosystems when both sensors were considered. While not every HSI issue can be completely compensated for, an appreciation of common artifacts allows users to make more informed decision about their impact on planned analysis.

The Correlation Coefficient as a Simple Tool for the Localization of Errors in Spectroscopic Imaging Data

The correlation coefficient (CC) was substantiated as a simple, yet robust statistical tool in the quality assessment of hyperspectral imaging (HSI) data. The sensitivity of the metric was also characterized with respect to artificially-induced errors. The CC was found to be sensitive to spectral shifts and single feature modifications in hyperspectral ground data despite the high, artificially-induced, signal-to-noise ratio (SNR) of 100:1. The study evaluated eight airborne hyperspectral images that varied in acquisition spectrometer, acquisition date and processing methodology. For each image, we identified a uniform ground target region of interest (ROI) that was comprised of a single asphalt road pixel from each column within the sensor field-of-view (FOV). A CC was calculated between the spectra from each of the pixels in the ROI and the data from the center pixel. Potential errors were located by reductions in the CCs below a designated threshold, which was derived from the results of the sensitivity tests. The spectral range associated with each error was established using a windowing technique where the CCs were recalculated after removing the spectral data within various windows. Errors were isolated in the spectral window that removed the previously-identified reductions in the CCs. Finer errors were detected by calculating the CCs across the ROI in the spectral range surrounding various atmospheric absorption features. Despite only observing deviations in the CCs from the 3rd–6th decimal places, non-trivial errors were detected in the imagery. An error was detected within a single band of the shortwave infrared imagery. Errors were also observed throughout the visible-near-infrared imagery, especially in the blue end. With this methodology, it was possible to immediately gauge the spectral consistency of the HSI data across the FOV. Consequently, the effectiveness of various processing methodologies and the spectral consistency of the imaging spectrometers themselves could be studied. Overall, the research highlights the utility of the CC as a simple, low monetary cost, analytical tool for the localization of errors in spectroscopic imaging data.

Spectro-spatial relationship between UAV derived high resolution DEM and SWIR hyperspectral data: application to an ombrotrophic peatland

Peatlands cover ~3% of the globe and are key ecosystems for climate regulation. To better understand the potential effects of climate change in peatlands, a major challenge is to determine the complex relationship between hydrology, microtopography, vegetation patterns, and gas exchange. Here we study the spectral and spatial relationship of microtopographic features (e.g. hollows and hummocks) and near-surface water through narrow-band spectral indices derived from hyperspectral imagery. We used a very high resolution digital elevation model (2.5 cm horizontal, 2.2 cm vertical resolution) derived from an UAV based Structure from Motion photogrammetry to map hollows and hummocks in the peatland area. We also created a 2 cm spatial resolution orthophoto mosaic to enhance the visual identification of these hollows and hummocks. Furthermore, we collected SWIR airborne hyperspectral (880-2450 nm) imagery at 1 m pixel resolution over four time periods, from April to June 2016 (phenological gradient: vegetation greening). Our results revealed an increase in the water indices values (NDWI1640 and NDWI2130) and a decrease in the moisture stress index (MSI) between April and June. In addition, for the same period the NDWI2130 shows a bimodal distribution indicating potential to quantitatively assess moisture differences between mosses and vascular plants. Our results, using the digital surface model to extract NDWI2130 values, showed significant differences between hollows and hummocks for each time period, with higher moisture values for hollows (i.e. moss dominated). However, for June, the water index for hummocks approximated the values found in hollows. Our study shows the advantages of using fine spatial and spectral scales to detect temporal trends in near surface water in a peatland.

A geospatial database model for the management of remote sensing datasets at multiple spectral, spatial, and temporal scales

In this study the development and implementation of a geospatial database model for the management of multiscale datasets encompassing airborne imagery and associated metadata is presented. To develop the multi-source geospatial database we have used a Relational Database Management System (RDBMS) on a Structure Query Language (SQL) server which was then integrated into ArcGIS and implemented as a geodatabase. The acquired datasets were compiled, standardized, and integrated into the RDBMS, where logical associations between different types of information were linked (e.g. location, date, and instrument). Airborne data, at different processing levels (digital numbers through geocorrected reflectance), were implemented in the geospatial database where the datasets are linked spatially and temporally. An example dataset consisting of airborne hyperspectral imagery, collected for inter and intra-annual vegetation characterization and detection of potential hydrocarbon seepage events over pipeline areas, is presented. Our work provides a model for the management of airborne imagery, which is a challenging aspect of data management in remote sensing, especially when large volumes of data are collected.