ASPRS

PE&RS April 2000

VOLUME 66, NUMBER 4
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
JOURNAL OF THE AMERICAN SOCIETY FOR PHOTOGRAMMETRY AND REMOTE SENSING

Peer-Reviewed Article Abstracts

385 Rigorous Mathematical Modeling of Airborne Pushbroom Imaging Systems
Changno Lee, Henry J. Theiss, James S. Bethel, and Edward M. Mikhail

Abstract
Download Full Article
Airborne pushbroom imaging systems are becoming increasingly of interest for panchromatic, multispectral, and hyperspectral applications. The source imagery for this paper is from the HYDICE  (Hyperspectral Digital Imagery Collection Experiment) sensor, developed by the Naval Research Laboratory at Stennis Space Center. The primary focus of this paper is to present rigorous mathematical models that rectify a single band of HYDICE  imagery so that each pixel in the image can be accurately transformed to its ground location. Equations are presented for a piecewise polynomial model and a Gauss-Markov model. The implementation of the mathematical modeling of straight linear features into the rectification using either model is discussed. The valuable contribution of linear features and the superior performance of the Gauss-Markov model are shown in tables as well as in orthorectified images that compare the results to those from simpler models.

393 Accurate Orientation for Airborne Mapping Systems
Jan Skaloud and Klaus-Peter Schwarz

Abstract
Download Full Article
Stringent requirements on the accuracy of attitude determination are currently a major challenge for strapdown INS/GPS integration, which is at the core of self-contained airborne remote sensing and mapping systems. This paper reviews the error models for INS/GPS  integration and focuses in detail on designing filtering methods for improving attitude accuracy in the bandwidth in which an inertial system does not benefit from frequent GPS  position/velocity updates. Several filtering methods are designed based on the spectral analysis of the raw inertial signal in a dynamic environment. These include a spectral technique for dither spike removal and a class of low-pass finite-impulse-response (FIR) filters operating in forward/backward manner for achieving zero phase distortion. The orientation performance of the whole system with different filters is evaluated by comparing it to the "true" attitude information provided by a photogrammetric block adjustment. Results show clearly that the choice of an appropriate filter is decisive for attitude accuracy. Overall, the INS/GPS  integration combined with the most suitable pre-filtering method agrees with the external orientation reference to 0.005º (19") RMS over the whole test period, while the flight-line consistency is typically 0.003º (10") RMS. The best filter in the comparison has an RMS seven times smaller than the Butterworth filter which is frequently applied in the industrial designs of INS.

403 Contributions of Differential GPS and GLONASS Observations to Point Accuracy  under Forest Canopies
Eric Næsset, Trygve Bjerk, Ola Øvstedal, and  Lorentz H. Ryan

Abstract
Download Full Article
A 24-channel, single-frequency receiver observing the C/A code and carrier phase of both GPS  and GLONASS  was used to determine the positional accuracy of 27 points under tree canopies. The mean positional accuracy based on combined differential postprocessing of GPS C/A  code and carrier phase observations ranged from 0.20 m to 5.72 m for 2.5 min to 30 min of observation at points with basal area ranging from <15 m²/ha to >25 m²/ha. The mean positional accuracy of combined differential postprocessing of GPS+GLONASS C/A  code and carrier phase observations ranged from 0.09 m to 2.85 m. The accuracy increased with decreasing density of forest and improving geometric satellite distribution as observed at the base station. Static differential postprocessing of 30-min observations of GPS+GLONASS  carrier phase gave fixed solutions for 13 of the 27 points. The accuracy of these solutions ranged from 0.01 m to 0.09 m. The probability of obtaining a fixed solution under tree canopies was modeled by logistic regression. The regression model classified correctly 81.5 percent of the processed solutions.

409 Accuracy Assessment of Wetland Boundary Delineation Using Aerial Photography and Digital Orthophotography
Jeffrey Barrette, Peter August, and Francis Golet

Abstract
Download Full Article
We compared the horizontal accuracy of forested wetland boundary delineations obtained from large-scale, natural color aerial photography against delineations obtained from "heads-up" digitizing of digital orthophotography. The absolute value of the mean (± SD) distance between the field-derived "true" wetland boundary location and the orthophotograph-derived wetland boundaries was 3.4 ± 3.4 m and was slightly more accurate than the mean distance between the "true" boundary and the aerial-photo-derived wetland boundaries (4.5 ± 4.0 m). Field visitation increased the accuracy of delineating wetland boundaries, and the majority of wetland edges-both orthophotograph- and photo-derived-were placed on the wetland side relative to the `true' boundary.

417 Detecting Ice-Sheet Topography with AVHRR, RESURS-01, and Landsat TM  Imagery
Robert Bindschadler and Patricia Vornberger

Abstract
Download Full Article
The ability of different visible-band satellite imagers to detect subtle ice-sheet topographic features is examined by both qualitative and quantitative comparisons. The AVHRR  10-bit sensor is shown to have the best radiometric resolution, but its poorer spatial resolution limits this sensor's ability to discriminate smaller scale topography. RESURS-01  is a less well known 8-bit sensor, but this instrument's higher noise and poorer radiometric resolution restrict its utility for ice-sheet research. Landsat TM  applies 8-bit quantization over a narrower radiometric range, causing it to saturate for some conditions. This limitation allows TM  to achieve a radiometric resolution half that of AVHRR  while retaining a much higher spatial resolution. No SPOT  imagery was compared, but the sensor is shown to have less radiometric resolution than TM . However, SPOT  has better spatial resolution and is less likely to saturate.

423 Predicting the Distribution of Eurasian Badger (Meles meles) Setts over an Urbanized Landscape: A GIS Approach
Amanda Wright, Alan H. Fielding, and C. Philip Wheater

Abstract
The distribution of the Eurasian badger within an urbanized county (Greater Manchester, United Kingdom) was investigated through the development of a predictive model. Badger main sett distribution was taken as a surrogate of badger social group distribution, with the recognition of the importance of main setts as a limiting resource for badger distribution. The model utilized a range of environmental data layers within a GIS (land cover, soil, slope, and elevation). The data were transformed using principal components analysis to produce a reduced set of uncorrelated, normally distributed environmental variables. Predictive discriminant analysis was used to construct the model by discriminating main sett sample sites from random non-sett sites. The model classified over 80 percent of sites correctly and was validated using external analysis. A visual representation of the model was produced through interpolating within a GIS a discriminant surface from the discriminant scores of the model.

429 Multispectral Satellite Image and Ancillary Data Integration for Geological Classification
Evaristo Ricchetti

Abstract
Download Full Article
Digital classification of Landsat imagery for geological purposes often gives poor results. To improve classification accuracy, spectral data have been combined with ancillary data. These data have been used in pre-classification processing to enhance image quality, and as additional attribute information during the classification process. Several geological classifications were conducted using various levels of integrated spectral and topographic data. A slope map was used to add information about the geomorphologic nature of the different geological units. Logical channel and stratification methods were applied and compared to a spectral classification. The classification results demonstrate that a significant increase in overall accuracy can be achieved by combining topographic data with spectral data. The use of stratification, in addition to a logical channel, did not give an evident improvement in overall accuracy. The use of ancillary data in image classification must rely on in-depth knowledge of the target to select the attribute that best characterizes it.

437 Integrating Landsat, Geologic, and Airborne Gamma Ray Data as an Aid to Surficial Geology Mapping and Mineral Exploration in the Manitouwadge Area, Ontario
I.M. Kettles, A.N. Rencz, and S.D. Bauke

Abstract
Download Full Article
For surficial mapping and mineral prospecting purposes, Landsat Thematic Mapper, airborne gamma ray spectrometry, and geological data were integrated for a 5000-km² area near Manitouwadge, Ontario. Using characteristic Landsat data signatures for the main surficial units, the signatures for 6,250,000 30-by 30-m-pixel areas were evaluated, and each pixel area was assigned to a surficial unit category. When the predictive surficial geology map thus generated was compared to eight surficial geology units on a published surficial geology map, there were obvious visual similarities, with an overall pixel-by-pixel accuracy of 46 percent. Pixel areas with Landsat and gamma ray signatures comparable to those of training sites in base metal-enriched tills near the Manitouwadge area mines, commonly formed clusters or near-linear bands overlying Archean greenstone belts or in close contact with a carbonatite complex where Fe and Zn mineralization is known. Tests on these derivative pixel areas indicated that their distribution was controlled almost entriely by the gamma ray data signatures.
Top Home