IEEE Geoscience and Remote Sensing Magazine - December 2020 - 90

In addition to these images from multispectral sensors,
the scientific community has access to high-quality hyperspectral data from the Airborne Visible/Infrared Imaging
Spectrometer (AVIRIS), which began providing such data
in 1987. An important factor in the use of hyperspectral
images was the software produced under the leadership
of the Environment for Visualizing Images [3] to process
the data. In 1999, the Ikonos satellite began to capture the
first 1-m spatial resolution panchromatic images from
space. Ikonos was followed by the QuickBird, GeoEye, and
WorldView satellite series. The latest WorldView-4 satellite
is capable of delivering 30-cm panchromatic and 120-cm
MSIs, currently the highest spatial resolution a commercial
satellite can achieve.
In addition to developments related to optical passive
detection sensors (using solar energy), rapid progress has
been made regarding active systems in the field of remote
sensing. These systems, including lidar, radar, and synthetic aperture radar (SAR),
THE ABUNDANCE OF DATA
send energy to the target and
OF VARYING QUALITY
collect information about the
COLLECTED BY
object by analyzing and interMISCELLANEOUS SENSORS
preting the reflected energy.
In the 1990s, the first real
POSES A GREAT CHALLENGE
" operational " (nonexperimenFOR SCIENTISTS EXAMINING
tal) radar remote sensing satAND INTERPRETING
ellites were developed. From
THESE DATA.
1991 to 1995, the European
Space Agency and the space
agencies of Japan, Canada,
and the former Soviet Union helped increase the number
of radar imaging systems by launching radar satellites into
orbit [4]. After the NASA Shuttle Radar Topography Mission project, which collected the world's first 30-m-resolution digital terrain model (DTM) in 2000, the German
Aerospace Center (DLR) launched two satellites, TerraSAR-X
in 2007 and TanDEM-X in 2010, to create higher-spatialresolution DTMs using interferometric SAR observations.
The DLR announced that the DTM generated from these
two satellites has a spatial resolution of approximately 12 m
(0.4 arc seconds) with an absolute vertical accuracy of 10 m
(90% of sampling is within 10 m of truth), and its relative
vertical accuracy is fewer than 2 m in areas where the slope
is less than 20% and 4 m (90% of sampling is within 4 m of
truth) in other places [5].
In addition to the rapid developments in sensor technologies, progress has been made in recent years in incorporating the capabilities of different sensors into a single sensor.
In 2013, a new approach was developed that combines active and passive infrared imaging into a single chip. This
sensor enables lighter, simpler dual-mode active/passive
cameras that consume less power [6]. Meanwhile, many improvements have come along in data collection and in the
variety of applications conducted with low-cost and effective unmanned aerial vehicle (UAV) systems. Sensors with
90

different characteristics can be mounted on these platforms,
and many different projects can be carried out with highspatial-resolution images taken from low altitudes. These
systems are offered at reasonable prices and are easy to use
and available with an increasing number of options in the
market for the software required to process the captured
images. Thus, these systems are coming into widespread
use. Figure 1 shows in detail the current sensor types used
in remote sensing.
Figure 1 indicates that remote sensing data with different characteristics can be obtained using a wide variety of
platforms and different sensors employing different regions
of the electromagnetic spectrum. To make these different
types of data ready for use in different remote sensing projects, various preprocessing algorithms for topographic, geometric, atmospheric, and radiometric corrections should
be applied. While the processes to be performed for each
data type may vary, a number of approaches are also used
in the production of the final products for many applications, such as those related to environmental and natural
resources management, climate change, disaster management, forestry, archeology, geology, urban development,
and land use/land cover (LU/LC) mapping [8]-[11]. Analysis and interpretation of data consist of evaluating image
quality, enhancing images, preprocessing images, selecting
features, fusing images, filtering images, determining band
ratios and indexes, detecting change, classifying images,
and performing other tasks. The abundance of data of varying quality collected by miscellaneous sensors poses a great
challenge for scientists examining and interpreting these
data. Although most data collection and processing activities can be completed automatically, data analysis and interpretation for remote sensing data are complex tasks that
machine learning alone cannot yet perform adequately.
Furthermore, downloading global satellite data daily takes
more than a few terabytes, so analyzing and interpreting
these data are time-consuming challenges. Therefore, it
would be appropriate to use collective intelligence to cope
with such large data sets [12].
An average of 1-5 exabytes (EB; 1018 B) per day is generated, and 90% of the world's existing data has been generated in the last two years [13]. This large amount of data
doubles every two years. In 2010, data volume reached 1
zettabyte (ZB; 1021 B) for the first time; in 2011, approximately 1.8 ZB of data were produced [14]. By 2020, the volume of data in the world is projected to be 50 times higher
than in 2011 [15]. The producer of the most significant
amounts of data will be not a single person, company, or
organization, but crowds.
Today, the world's population has reached approximately 7.5 billion. According to 2018 statistics, the number of
Internet users increased by 7% annually to 4,021 billion.
Meanwhile, as of 2018, the number of smartphone users,
increasing at a rate of 4% annually, reached 5,135 billion.
As the number of people using the Internet is growing, so is
the time people spend on the Internet. The latest data from
IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE

DECEMBER 2020

IEEE Geoscience and Remote Sensing Magazine - December 2020

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - December 2020