Amy L. Neuenschwander

Fusion of airborne polarimetric and interferometric SAR for classification of coastal environments

AIRSAR and TOPSAR data were acquired over the wetlands of Bolivar Peninsula along the Gulf coast of Texas for mapping land cover types and topographic features such as beach ridges, dunes, and relict storm features. Classification of land cover over this wetlands and uplands environment is difficult because of the similarity of spectral signatures of the vegetation types. In addition, because the distribution of vegetation communities in coastal marshes is strongly related to salinity, which in turn is largely dictated by frequency and duration of inundation, surface topography is critical to determination of the vegetation characteristics at any location. The potential advantages of multisensor classification, including, in particular, topographic information from a TOPSAR DEM are investigated. An approach which employs a class dependent feature selection procedure in conjunction with pairwise Bayesian classifiers is proposed and applied to the polarimetric and interferometric SAR data.

Landcover classification of small-footprint, full-waveform lidar data

Full-waveform lidar data are emerging into the commercial sector and provide a unique ability to characterize the landscape. The returned laser waveforms indicate specific reflectors within the footprint (vertical structure), while the shape of the return convolves surface reflectance and physical topography. These data are especially effective in vegetative regions with respect to canopy structure characterization. The objective of this research is to evaluate the performance of waveform-derived parameters as input into a supervised classifier. Extracted waveform metrics include Gaussian amplitude, Gaussian standard deviation, canopy energy, ground energy, total waveform energy, ratio between canopy and ground energy, rise time to the first peak, fall time of the last peak, and height of median energy (HOME). The classifier utilizes a feature selection methodology which provides information on the value of waveform parameters for discriminating between class pairs. For this study area, energy ratio and Gaussian amplitude were selected most frequently, but rise time and fall time were also important for discriminating different tree types and densities. The lidar classification accuracy for this study area was 85.8% versus 71.2% for Quickbird imagery. Since the lidar-based input data are structural parameters derived from the waveforms, the classification is improved for classes that are spectrally similar but structurally different.

Interactions between fire and flooding in a southern African floodplain system (Okavango Delta, Botswana)

A series of 98 satellite images was analysed to reconstruct the fire and flood history of a floodplain system in southern Africa (Okavango Delta, Botswana). The data was used to investigate interactions between fire and flooding, and to determine the relevance of rainfall and flood-events for fire occurrences on floodplains and on drylands. The aims of the study are (1) to analyse and compare the fire frequency on floodplains and on adjacent drylands, (2) to investigate the influence of rainfall and flooding on the fire occurrence and (3) to determine correlations between fire frequency and flood frequency. The analyses show higher fire frequencies on floodplains than on drylands because of higher biomass production and fuel loads. The fire occurrence on drylands shows a correlation with annual rainfall events, while the fire frequency on floodplains is in principle determined by the flood frequency. Between floodplain types, clear differences in the susceptibility to fire where shown by analysing flood frequency vs. fire frequency. Here, the highest potential to burn was found for floodplains that get flooded about every second year. By calculating mean fire return intervals, the potential to burn could be specified for the different floodplain types.

Damage Patterns from Satellite Images of the 2003 Bam, Iran, Earthquake

High-resolution (0.6m) commercial satellite images contain a wealth of information for mapping earthquake damage. Satellite images of the city of Bam, acquired on 30 September 2003 (pre-earthquake) and 03 January 2004 (post-earthquake), were obtained and used to distinguish damage patterns across the city. Comparisons between pre- and post-earthquake images clearly show structural damage and collapse. Using spectral (color) and textural information from the post-earthquake image, regions of damage were identified using a semi-automated computer-based algorithm. This analysis indicates that the damage within the city of Bam was concentrated in the eastern sections of the city. The extent of damage in some sections of the city reached 100%. The results from this study not only provide information regarding damage patterns for the city of Bam, but they also illustrate the potential for using satellite images to understand and document earthquake effects during future earthquakes.

Results from the EO‐1 experiment—A comparative study of Earth Observing‐1 Advanced Land Imager (ALI) and Landsat ETM+ data for land cover mapping in the Okavango Delta, Botswana

The Earth Observing‐1 (EO‐1) satellite acquired a sequence of data in 2001 and 2002 that highlighted the annual flooding of the lower Okavango Delta. The data were collected as part of the calibration/validation programme for the Advanced Land Imager (ALI) sensor on the NASA EO‐1 satellite. The primary purpose of this study was to compare the capability of ALI to that of Landsat ETM+ for large‐scale mapping applications in the Okavango Delta. While the extent and inaccessibility of many areas of the Delta make application of remote sensing attractive, the availability of data with adequate spatial and spectral resolution has limited the characterization of the complex patterns of land cover and geomorphology in the Delta. Initial analysis of the ALI data via supervised classification clearly showed macro‐flood features, delineation of downstream channel flow areas, and lateral‐downstream inundation of the floodplain. These patterns and the proportions of flooding of the channel compared to that of the floodplain (impoundment) varied annually, from the wetter seasonal swamps through the drier seasonal and occasional swamps. Consistently higher classification accuracies achieved using ALI data relative to ETM+ data are attributed to the higher signal‐to‐noise ratio and the increased dynamic range of the ALI data.

Applicability of an automatic surface detection approach to micro-pulse photon-counting lidar altimetry data: implications for canopy height retrieval from future ICESat-2 data

We develop and validate an automated approach to determine canopy height, an important metric for global biomass assessments, from micro-pulse photon-counting lidar data collected over forested ecosystems. Such a lidar system is planned to be launched aboard the National Aeronautics and Space Administration’s follow-on Ice, Cloud and land Elevation Satellite mission (ICESat-2) in 2017. For algorithm development purposes in preparation for the mission, the ICESat-2 project team produced simulated ICESat-2 data sets from airborne observations of a commercial micro-pulse lidar instrument (developed by Sigma Space Corporation) over two forests in the eastern USA. The technique derived in this article is based on a multi-step mathematical and statistical signal extraction process which is applied to the simulated ICESat-2 data set. First, ground and canopy surfaces are approximately extracted using the statistical information derived from the histogram of elevations for accumulated photons in 100 footprints. Second, a signal probability metric is generated to help identify the location of ground, canopy-top, and volume-scattered photons. According to the signal probability metric, the ground surface is recovered by locating the lowermost high-photon density clusters in each simulated ICESat-2 footprint. Thereafter, canopy surface is retrieved by finding the elevation at which the 95th percentile of the above-ground photons exists. The remaining noise is reduced by cubic spline interpolation in an iterative manner. We validate the results of the analysis against the full-resolution airborne photon-counting lidar data, digital terrain models (DTMs), and canopy height models (CHMs) for the study areas. With ground surface residuals ranging from 0.2 to 0.5 m and canopy height residuals ranging from 1.6 to 2.2 m, our results indicate that the algorithm performs very well over forested ecosystems of canopy closure of as much as 80%. Given the method’s success in the challenging case of canopy height determination, it is readily applicable to retrieval of land ice and sea ice surfaces from micro-pulse lidar altimeter data. These results will advance data processing and analysis methods to help maximize the ability of the ICESat-2 mission to meet its science objectives.

Development of laser waveform digitization for airborne LIDAR topographic mapping instrumentation

Airborne lidar (Light Detection and Ranging) topographic mapping has been the recent focus of both the research and commercial sectors in generation of high resolution surface models. While conventional airborne lidar systems that record the laser range and backscatter intensity information have made a revolutionary impact on three-dimensional imaging of the earths surface, full waveform digitization potentially provides excellent opportunities, both for improved mapping and new applications. Operating simultaneously with a conventional lidar system, a new module developed jointly by Optech, Inc. of Toronto, Canada, and the University of Texas at Austin records the analog waveform of the laser pulse and converts the samples to digital measurements. This paper focuses on discussion of the preliminary processing methodology and comparison of the information extracted from the conventional first/last return data and the waveform digitizer over several targets, including a residential area, conifer and hardwoods forest, and an airport runway. Additional information including surface roughness, terrain slope, canopy structure, and light transmittance through vegetation can be derived from the waveforms. Characterization of the canopy structure through waveform digitization can yield information such as fraction of photosynthetic active radiation (fPAR.) commonly used in ecosystem modeling.

Classification of LIDAR data using a lower envelope follower and gradient-based operator

A new, computationally efficient classification methodology was developed and implemented to classify Light Detection and Ranging (LIDAR) data as ground, vegetation, and man-made features (Weed 2001). The new procedure consists of several components that create ground, vegetation, and building surfaces, which are then used to classify the first and last reflection of each laser pulse. Ground and non-ground data are classified by adapting the concept of a lower envelope follower used to recover information in an amplitude modulated (AM) signal to the problem of extracting the ground surface from the LIDAR signal. The detected ground points include bare surface pixels and locations where the laser was able to penetrate the vegetation canopy, but exclude buildings and vegetation. Buildings are then classified by detecting the extended low gradient regions on their roofs. The first return LIDAR data points are used to accurately detect building edges distorted by multi-path errors in the last return LIDAR data. The combined roof and edge surfaces are then employed to threshold the first and last return LIDAR height values and detect the LIDAR points reflecting from buildings. Once the building points are classified, the vegetation points are extracted from the remaining LIDAR points using a mask of the regions where there were significant differences in the first and last return of the laser pulse. The technique is robust for classifying LIDAR data acquired over a range of terrains with different vegetation cover and types and sizes of buildings. It requires minimal user intervention for parameter selection.

Disturbance, Management, and Landscape Dynamics: Harmonic Regression of Vegetation Indices in the Lower Okavango Delta, Botswana

Focused on the Okavango Delta, Botswana, this research investigates (a) whether ecosystem signals derived from remotely sensed imagery can be decomposed using a harmonic regression, (b) if the deviations from the decomposed signal are correlated with observed flooding and fire regimes, and (c) the impact of explicitly including agriculture, settlement areas, and land management systems on the derived signals.

Earthquake damage identification using multi-temporal high-resolution optical satellite imagery

This paper uses preand post-earthquake high resolution (0.6 m) optical satellite images to identify damage patterns in the city of Bam, Iran during the 2003 Bam, Iran earthquake. The Quickbird satellite images were co-registered and used in a change detection algorithm to identify damage to the urban infrastructure. The change detection algorithm uses changes in texture, as computed by the correlation coefficient between the images, to identify damage. Additionally, a mask was applied to remove vegetation and shadow from the damage identification. The algorithm identified damage areas within the city of Bam that were consistent with results from field investigations. In the future, high resolution satellite images could play an important role in identifying earthquake damage patterns and assisting earthquake reconnaissance and response teams. Keywords-earthquake; change detection; urban damage.