IEEE Geoscience and Remote Sensing Magazine - December 2021 - 91

HR GLOBAL CHANGE DETECTION
Remote sensing imagery has long been considered an effective
data source for global change detection, due to its
large coverage area, convenient access, and frequent revisits.
Previous multitemporal global maps of land cover and
thematic change detection are often generated at a relatively
coarse resolution (i.e., >300 m), e.g., 8-km-resolution
global forest change based on Advanced Very High Resolution
Radiometer data for 1982-1999 [223], 500-m resolution
mapping of the global urban extent from MODIS data from
2005 and 2009 [224], [225], and the 300-m resolution annual
Climate Change Initiative Land Cover maps from 1992
to 2015 [226]. More recently, global-scale change detection
with fine spatial resolution (around 30 m) has been attempted
with open source Landsat imagery. Notable examples
include the Global Forest Cover database [227], GlobeLand30
global land cover product [228], Global Artificial
Impervious Area annual maps [229], Global Surface Water
data sets by the European Commission Joint Research Center
[230], and Global Human Settlement Layer framework
[231]. Please note that 30 m is not a high spatial resolution
in a common sense, but it should be regarded as high in
the case of intercontinental and global mapping. Recently,
Gong et al. [232] developed a 10-m resolution global land
cover map through Sentinel-2 images acquired in 2017.
It is a trend that global products are being developed in
finer spatial and temporal resolutions that can characterize
heterogeneous and mixed areas more accurately. For
instance, the Planet CubeSats are able to acquire images at
a 3-5-m spatial resolution with near-real-time daily global
coverage [233], which has potential for VHR global change
detection in the future. In addition, cloud computing platforms,
such as Google Earth Engine and Amazon Web
Services, can facilitate the processing of large volumes of
satellite images and speed the development of VHR global
mapping [234].
HYPERSPECTRAL CHANGE DETECTION
Hyperspectral data can distinguish more detailed land cover
types due to their rich spectral information. For a long
time, the data availability of hyperspectral images seemingly
limited real applications in precise change detection.
Recently, however, the development of hyperspectral satellites
with a relatively fine spatial resolution, e.g., Gaofen-5
(30 m, with 330 spectral bands), Tiangong-1 (10 m, with
128 spectral bands), and Zhuhai-1 (10 m, with 32 spectral
bands), and airborne hyperspectral sensors, e.g., HyMap
(3 m, with 126 spectral bands) and the Reflective Optics System
Imaging Spectrometer (ROSIS) (1.3 m, with 115 spectral
bands), has significantly increased the availability of multitemporal
hyperspectral images. However, studies related to
VHR hyperspectral change detection are very limited, and
even the existing methodologies were developed based on
synthetic data [235]. Moreover, advances in hyperspectral
image classification benefit from a set of widely used public
benchmark data sets, e.g., the ROSIS Pavia University and
DECEMBER 2021 IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE
Airborne Visible/Infrared Imaging Spectrometer Salinas
data sets [236]. Therefore, there is an urgent need for public
hyperspectral change detection data sets to promote the development
of the related research fields.
URBAN FUNCTIONAL ZONE CHANGE DETECTION
Currently, the classification of urban functional zones is one
of the important research areas in interpreting VHR remote
sensing images, as the urban functional zones can bridge
the semantic gap between land cover and human socioeconomic
activities. Current urban functional zone mapping
not only involves various image features, e.g., deep [237],
[238], angular [97], object based [239], and textural [240],
but it also refers to multisource geographic information,
such as points of interest (POIs) [241], social media [242],
and mobile phone positioning [100]. In rapidly urbanizing
regions, the timely and accurate monitoring of urban functional
zones is crucial for planning and management. However,
studies for change detection in urban functional zones
are lacking. Frankly, urban functional zone change detection
is a difficult task since land cover change does not necessarily
signify the conversion of a functional zone type. Meanwhile,
multisource geographic data, e.g., POIs, are widely used for
functional zone classification [230], but these data do not
provide a time tag, which hampers the dynamic monitoring
of urban functional zones. These issues should be overcome
to effectively monitor changes in cities.
CONCLUSIONS
With the increasing availability of VHR remote sensing
images, precise, frequent, and stereo change detection becomes
possible. To the best of our knowledge, a comprehensive
review of VHR change detection is lacking in the
current literature. Therefore, this article aimed to summarize
recent advances in VHR remote sensing image change
detection, including methods and applications. The review
of methods focused on feature extraction and change detectors
for multitemporal VHR images. Applications including
change detection for land cover and land use, impervious
surfaces, buildings, crops, vegetation, lakes and wetlands,
and ecosystem services were reviewed. Finally, some future
directions were suggested and discussed for this important
research area. Recommendations for future work include
focusing on change tracking, global change detection, hyperspectral
change detection, and urban functional zone
change detection to generate frequent and detailed semantic
change information on a global scale.
ACKNOWLEDGMENTS
The authors are grateful to the editor-in-chief, associate editor,
and reviewers for their insightful comments and suggestions.
This research was supported by the National Natural
Science Foundation of China, under grants 41901279,
41771360, and 41971295, and the Chinese Academy of
Sciences Interdisciplinary Innovation Team, under grant
JCTD-2019-04. (Corresponding author: Xin Huang.)
91
IEEE Geoscience and Remote Sensing Magazine - December 2021

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - December 2021
