Peer-Reviewed Articles
857 Extraction of Impervious Surface Areas from High
Spatial Resolution Imagery by Multiple Agent
Segmentation and Classification
Yuyu Zhou and Y.Q. Wang
Abstract Download
Full Article
In recent years impervious surface areas (ISA) have emerged
as a key paradigm to explain and predict ecosystem health
in relationship to watershed development. The ISA data are
essential for environmental monitoring and management in
coastal State of Rhode Island. However, there is lack of
information on high spatial resolution ISA. In this study, we
developed an algorithm of multiple agent segmentation and
classification (MASC) that includes submodels of segmentation,
shadow-effect, MANOVA-based classification, and postclassification.
The segmentation sub-model replaced the
spectral difference with heterogeneity change for regions
merging. Shape information was introduced to enhance the
performance of ISA extraction. The shadow-effect sub-model
used a split-and-merge process to separate shadows and the
objects that cause the shadows. The MANOVA-based classification
sub-model took into account the relationship between
spectral bands and the variability in the training objects and
the objects to be classified. Existing GIS data were used in
the classification and post-classification process. The MASC
successfully extracted ISA from high spatial resolution
airborne true-color digital orthophoto and space-borne
QuickBird-2 imagery in the testing areas, and then was
extended for extraction of high spatial resolution ISA in the
State of Rhode Island.
869 Data Combination and Feature Selection for Multisource
Forest Inventory
Reija Haapanen and Sakari Tuominen
Abstract Download
Full Article
Both satellite images and aerial photographs are now
used operationally in Finland’s forestry for different
tasks; satellite images are used for national forest inventory
purposes and aerial images for forest management
planning. Due to the double coverage, it could be advantageous
to utilize the strengths of both image types.
The aim of this study was to evaluate the potential of the combination of Landsat ETM+ spectral and aerial photograph spectral and textural features for forest variable estimation. The studied stand variables were mean height, basal area per hectare, and the volume of the growing stock. Several approaches were tested when combining the image data sources: feature selection, feature weighting, satellite image-based stratification, and combination of individual estimates by weighting.
The highest accuracies were obtained when both data sources were used. There were several good ways to combine the data sources. Feature selection with genetic algorithm and subsequent feature weighting gave the lowest mean volume RMSE (63.7 m3/ha, 65.3 percent of the mean).
881 Comparison of Single-and Multi-date Landsat
Data for Mapping Wildfire Scars in Ocala
National Forest, Florida
Mary C. Henry
Abstract Download
Full Article
Remote sensing techniques have been widely used to map
fire scars in the western United States, but have not been
thoroughly tested in the eastern portion of the country. In this
study, a 1998 Landsat Thematic Mapper (TM) image and a
1999 Enhanced Thematic Mapper (ETM1) image were used to
test different image enhancements and classification algorithms
for mapping wildfire scars in Ocala National Forest, Florida.
Single-date analysis was conducted using the 1999 image,
while both images were used to complete multi-temporal
analysis. Both single- and multi-date datasets were classified
using a traditional method (maximum likelihood classification:
MLC) and a non-parametric technique (classification and
regression trees: CART). Comparison of all techniques showed
that MLC of a single image (1999) resulted in high accuracy
compared to the other methods and that principal components
analysis (PCA) and multitemporal PCA provided the best
spectral separability between burned and unburned areas.
893 Change Detection Techniques for Use in a Statewide
Forest Inventory Program
D.W. Wilkinson, R.C. Parker, and D.L. Evans
Abstract Download
Full Article
Eight change detection procedures and a hybrid forest type
classification procedure were tested for their ability to detect
forest land-cover change in east-central Mississippi. The
best performing method was change vector analysis using
vegetation indices with an image segmentation classification
that produced an overall accuracy (82.50 percent)
and overall Kappa (0.7900) calculated from error matrices.
The hybrid forest type classification had an overall accuracy
of 77.08 percent for 1997 and 71.25 percent for 2002.
The results of this study were compared to a prior pilot
inventory study for the same study area in east-central
Mississippi. There was considerable disagreement between
the two studies in terms of number of hectares in the age
classes and the forest type classes, most likely attributed to
the difference in methods for determining forest type classes.
903 Decision-Based Fusion for Improved Fluvial Landscape
Classification Using Digital Aerial Photographs and
Forward Looking Infrared Images
Kathy M. Smikrud, Anupma Prakash, and Jeff V. Nichols
Abstract Download
Full Article
Riverine landscapes associated with large dynamic floodplains
provide a complex array of habitats for fish. Mapping
and quantitative assessment of the habitats poses a major
challenge. This study uses high spatial resolution airborne
digital photographs and forward-looking infrared (FLIR)
images, simultaneously acquired in spring 2005 over a 12
river kilometer section of the Unuk River in Southeast Alaska
to map macro habitat indicators for Pacific salmon such as
large woody debris (LWD), water channels, sand/gravel bars,
and riparian vegetation. Image processing revealed that LWD
could best be extracted using contextual information from the
digital photos. River channel water had prominent shadows
cast from neighboring trees and could not be accurately
classified using digital photos, but could be well-delineated
using unsupervised classification of the FLIR images. All other
classes showed up well using supervised classification of
digital photos. Using a decision-based fusion approach,
the best individual classification results obtained from the
digital photos and FLIR images to generate an improved
fluvial landscape classification map (land-cover map). Using
a decision-based fusion method resulted in an overall
classification accuracy of the study area to 84.29 percent,
compared to 77.00 percent using supervised classification of
aerial photos alone. This appears to be first time that highresolution
airborne thermal images have been used for fluvial
landscape classification, and the study clearly demonstrates
the value of using thermal images and decision-based fusion
approach for improved land-cover classification.
913 Wal-Mart from Space: A New Source for Land Cover
Change Validation
David Potere, Neal Feierabend, Alan H. Strahler,
and Eddie E. Bright
Abstract Download
Full Article
We introduce an event data set of the location and opening
dates for 3,043 Wal-Mart stores as a means for validating
land-cover change-related products at medium (28.5 m) to
coarse (250 m to 1 km) resolutions throughout the conterminous
United States. Strengths of the Wal-Mart validation
data set include: construction within most U.S. ecoregions,
building footprints greater than a hectare in size, and
construction dates that span much of the remote sensing
record (1962 to 2004). We built the data set by geo-coding
Wal-Mart addresses, establishing opening dates, and geolocating
the footprints of 30 stores using online free highresolution
(4 m) imagery. Disturbance events were evident at
25 Wal-Marts within a single scene of the Landsat Ecosystem
Disturbance Adaptive Processing System (LEDAPS-beta) forest
disturbance product. In addition, we found clear disturbance
signals within two 16-day vegetation index time series:
the 250 m Moderate Resolution Imaging Spectroradiometer
normalized difference product (MOD44C) and the 1 km
enhanced vegetation index product (MOD13A2).
921 Neural Network Classification of Mangrove Species
from Multi-seasonal Ikonos Imagery
Le Wang, José L. Silván-Cárdenas, and Wayne P. Sousa
Abstract Download
Full Article
Tropical forests in many areas of Central and South
America experience strong seasonality in climatic variables
such as rainfall, solar radiation, wind speed, and relative
humidity. Such seasonality is typical of the mangrove
forests we study along the Caribbean coast of Panama. Tied
to this environmental variation are changes in leaf phenology
and physiology that can affect the spectral properties
of leaves and thus our ability to discriminate canopies of
differing species composition. The goals of this study were
two-fold. First, we compared the efficacy of three different
classification methods for discriminating mangrove
canopies, including a back-propagation, feed-forward neural
network classifier with two hidden layers of 24 and 12
neurons (hereafter, BP:24:12), a newly developed clusteringbased
neural network classifier (CBNN), and a maximum
likelihood classifier (MLC). Comparisons were made with
and without added textural information. Our second aim
was to compare the absolute and relative discrimination
abilities of these methods when applied to images of the
same forest acquired in different seasons.
Two sets of Ikonos images acquired in February (dry season) and May (early wet season) 2004 were analyzed in this study. When only spectral information was considered, MLC and CBNN discriminated differences in canopy species composition with higher accuracy than the BP:24:12 method. When second-order textural information was also taken into account, CBNN outperformed MLC and presented the best classification accuracy, i.e., kappa value equaled 0.93. Analyses of the wet season (May) image were consistently more accurate in discriminating mangrove canopies of differing species composition than analyses of the dry season (February) image, regardless of the classification method or the inclusion of textural information.
Color figure1. (Adobe PDF 18Kb) Color figure2. (Adobe PDF 16Kb)