Discriminating Saltcedar (
Tamarix ramosissima
)
from Sparsely Distributed Cottonwood
(
Populus euphratica
) Using a
Summer Season Satellite Image
Wenjie Ji and Le Wang
Abstract
Accurate mapping of saltcedar (Tamarix ramosissima) and
cottonwood (Populus euphratica) using remote sensing images
is required to study the dynamic relationship between these
two species. Our study used pixel-based and semi-object-based
methods to classify a high spatial resolution QuickBird image
acquired during the summer in northern China, where both
saltcedar and cottonwood are native species. The pixel-based
classification results revealed that spectral bands alone were
not sufficient to discriminate saltcedar from cottonwood trees
due to their similar foliage reflectance in the summer. Including
texture measures did not improve the result. The unique crown
shapes and shadows associated with sparsely distributed cot-
tonwood were used to facilitate the semi object-based method.
The overall accuracy of the object-based classification result
increased 15 percent compared to that of the pixel-based clas-
sification results and showed significant improvement in the
discrimination between saltcedar and cottonwood.
Introduction
Saltcedar (
Tamarix spp.
), a Eurasia native that was introduced
into the western United States more than a century ago, has
now colonized hundreds of thousands of acres of floodplains,
reservoir margins, and wetlands in all 17 US western states
(Nagler
et al
., 2011; Shafroth
et al
., 2005). During the pe-
riod of the rapid expansion of saltcedar from 1940s to 1960s
(Brotherson and Von Winkel, 1986), the population of many
native riparian species, such as cottonwood (
Populus spp.
)
and willows (
Salix spp.
), were reported to have declined
greatly (Nagler
et al
., 2004a; Stromberg, 1998).
Many researchers believe that the rapid spread of saltcedar
are preferred along regulated rivers while native pioneers,
such as cottonwood, will still dominate along free-flow rivers.
(Poff
et al
., 2007; Shafroth
et al.
, 1998; Shafroth
et al
., 2002;
Stromberg
et al
., 2007). However, recently it was found that
saltcedar were equally abundant on both regulated and unreg-
ulated rivers and would have naturalized and coexisted with
cottonwood trees even in the absence of river regulations (Mer-
ritt and Poff, 2010). While it still is unclear whether saltcedar
would replace cottonwood in the US, studying their relation-
ships in China, where saltcedar is native and poses no threats
to the others, could provide important insights and facilitate
understanding of the dynamics between these two species (Li
et al
., 2013). Very few studies have been conducted on this
topic (Cui
et al
., 2010; Wang and Zhang, 2014; Xie
et al
., 2011).
To better understand the relationship between cottonwood
and saltcedar, both in the US and in China, accurate distribu-
tion maps of both species along riparian corridors are required.
Remote sensing techniques have been applied extensively to
classify and map saltcedar distribution using aerial photos
(Akasheh
et al
., 2008; Ge
et al
., 2006; Nagler
et al
., 2005), mul-
tispectral satellite data (Evangelista
et al
., 2009 ; Lu and Wang,
2015; Silván-Cardenas and Wang, 2010; Wang
et al.
, 2013), and
hyperspectral images (Hamada
et al
., 2007; Narumalani
et al
.,
2006; Narumalani
et al
., 2009). The distinctive spectral char-
acteristic of the yellow-orange foliage of saltcedar during late
fall and early winter was used in many studies to discriminate
saltcedar from other riparian species (Everitt and Deloach,
1990; Ge
et al
., 2006; Silván-Cardenas and Wang, 2010). How-
ever, the duration of this unique coloration of saltcedar usually
is only several weeks. Consequently, ideal remote sensing im-
ages are not always available for saltcedar mapping. Although
the higher spectral resolution of hyperspectral data may
reduce this limitation, the cost of acquiring such data usually
is considerably higher. Therefore, methods are needed to iden-
tify saltcedar when they are spectrally more similar to other
riparian species using multispectral images. However, very
few studies have tried to address this issue probably because
of the high confusion among plant spectrums during these less
favorable times (Cochrane, 2000). In a pilot study comparing
spectral reflectance of saltcedar and cottonwood (resampled
using QuickBird’s spectrum-response function) collected
in the summer, we did not find any significant differences
between the spectral signatures of these two species (t-test,
p >0.1). Therefore, we believe that using spectral data alone
could be insufficient to distinguish saltcedar from cottonwood
in multispectral images acquired during the summer.
Despite the distance between the two continents, natural
environmental conditions, and even species compositions in
the arid and semi-arid areas of northern China are remarkably
similar to those in the western US, where saltcedar and cot-
tonwood are the most frequent and common species (Zhang
et al
., 2005). In addition, along many river corridors in both
China and the US, saltcedar often forms dense monocultural
State Key Laboratory Incubation Base of Urban Environmental
Processes and Digital Simulation; Key Laboratory of Resource
Environment and Geographic Information System; Key
Laboratory of 3-Dimensional Information Acquisition and
Application, Ministry of Education; College of Resources
Environment and Tourism, Capital Normal University,
Beijing, China; and the Department of Geography, University
at Buffalo, State University of New York, Buffalo, NY, United
States (
).
Photogrammetric Engineering & Remote Sensing
Vol. 81, No. 10, October 2015, pp. 795–806.
0099-1112/15/795–806
© 2015 American Society for Photogrammetry
and Remote Sensing
doi: 10.14358/PERS.81.10.795
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
October 2015
795
