534
July 2015
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
BOOK
REVIEW
Remote Sensing of Natural Resources
Guangxing Wang and Qihao Weng (eds.).
CRC Press, Boca Raton, FL. 2013. 580 p. Hardcover. ISBN
978-1-4665-5692-8.
Reviewed by:
Rudiger Gens, Remote Sensing
Scientist, Geophysical Institute, University of Alaska
Fairbanks, Fairbanks, AK.
Photogrammetric Engineering & Remote Sensing
Vol. 81, No. 7, July 2015, pp. 534–536.
0099-1112/15/534–442
© 2015 American Society for Photogrammetry
and Remote Sensing
doi: 10.14358/PERS.81.7.534
The book “Remote Sensing of Natural Resources” focuses on
remote sensing systems, algorithms, and their applications for
mapping, monitoring and modeling natural resources. It is meant
to serve as a supplementary text for undergraduate and graduate
students as well as the reference text for researchers, engineers
and decision makers. The book consists of seven sections and a
total of 26 chapters. While the book is generally well written, and
there is always something to be gleaned by the target audience,
the complexity of information presented fluctuates considerably
as one reads through the material. Some chapters provide an
extensive theoretical background, while others are limited to
reporting the results of specific case studies.
The
Remote Sensing Systems
section provides the technical
background of the systems used to acquire remote sensing data
for studying natural resources. The potential use of hyperspectral
and LiDAR data is emphasized in a number of chapters in the
book. Given the growing importance of these data for a large
variety of applications, it would have been nice to see more
information about these newer technologies in the introductory
chapter. Furthermore, a brief coverage of sensor capabilities
of new and future satellite missions, e.g. the recently launched
Soil Moisture Active Passive (SMAP) mission, would have
been useful for the readers. The authors do cover some future
missions, such as ICESat-2, in subsequent chapters.
The
Sampling Design and Product Quality Assessment
section reviews the progress and changing requirements in this
field of remote sensing. The integration of co-kriging (model-
based sampling) and stratification (design-based sampling)
greatly improved the cost efficiency of sampling design. The
demands on accuracy assessment of classification and modeling
have increased, particularly for large-area cover maps, change
detection and maps of continuous variables such as biomass and
leaf area index. The accuracy assessment is now extended from
crisp to soft classification maps. The extension is needed to more
accurately describe several classes within one pixel, e.g. natural
ecotones or gradations. The accuracy spatially varies depending
on the complexity of the landscape, soil properties, topographical
features, density of sample data and the increased accuracy of the
remotely sensed data. However, most of the current methods do
not meet the requirements of natural resource management yet.
In the area of
Land Use and Land Cover
(LULC)
Classification
multi- and hyper-spectral approaches are the most promising for
extracting information. The accuracy of the classification mostly
depends on the amount of effort spent on training, choice of
algorithm and processing capabilities. The section also devotes
two chapters to impervious surfaces, typically not considered
a natural resource. The higher resolution of the remote sensing
data available now severely limits the applicability of earlier
LULC classification techniques and significantly increases the
complexity of this task. Hyperspectral and LiDAR data show
promising results to address some of these difficulties.
For studying
Natural Landscapes, Ecosystems, and Forestry
the increased capabilities with respect to computer processing
power, spatial resolution and algorithm development has
been shown to have very positive effects. Increased computer
processing capabilities allow for greater complexity of modeling
forests and lake landscapes, where the remote sensing data is
used in initialization, validation and data integration. The higher
resolution data helps reducing the costs of forest monitoring and
now getting useful for modern forest management. The use of
airborne laser scanning has been especially emphasized. New
algorithms make increasing use of object-oriented approaches.
The authors point out that the use of remote sensing data for
Agriculture
studies, e.g. for crop monitoring and prediction, is not
as well developed as it could be. Availability of high-resolution
data increases the cost and complexity of the processing. While
there is great potential for precision farming, it is not widely
practiced yet. The data is also used for mapping of crop residue
cover, used to prevent erosion and conserve soil water resources.
