IEEE Geoscience and Remote Sensing Magazine - March 2023 - 83

and airborne equipment to carry out specific tasks in the
windy and severe weather conditions, without interfering
with the normal life of the penguins [35], [57], [91]. The
monitoring of penguins mainly includes their count, body
type, community type, distribution range, and migration
process [22], [41], [66], [92], [93], [94], [95], [96]. A deep
neural network implemented in the open source software
Nvidia DIGITS was used to automatically identify and count
the number of occupied nests of Adélie penguins in UAV
orthomosaics [92]. Furthermore, the response of penguins
to UAVs at various flight altitudes and the vehicles' interference
with the activities of penguins have been evaluated;
penguin responses to UAVs with various engine types
were also compared [37], [97], [98]. An analysis of variance
(ANOVA) and Tukey's Honestly Significant Difference test
were used to establish the statistical significance of differences
among different types of behavioral responses of penguins
to UAVs [97]. The results showed that the behavior of
a penguin colony changes with the altitude and engine type
of UAVs. Adult king penguins showed almost no behavioral
stress in the breeding season, but the adults and fledglings
in nonbreeding crèches showed strong behavioral responses
to UAVs [99]. Two binomial generalized linear mixed models
were created to test for changes in sensitivity among
different stages of the penguins' breeding cycle to various
flight activities of a micro-UAV, and an ANOVA was then
used to test for statistically significant differences among
the models, indicating an influence of the observation day
[39]. It was also observed that different species have different
behavioral responses to UAVs, depending on the height
of a drone. For example, giant petrels and cormorants are
highly sensitive to UAVs. The sensitivity of petrels, seagulls,
and seals to three types of UAV missions at varying flying altitudes
have also been investigated [100] and demonstrated
that the behavioral response of the animals significantly increased
with a decrease in UAV altitude.
UAVs have recently been used to ecologically assess Antarctic
wildlife, which includes monitoring populations, the
phenology of seabirds and leopard and elephant seals [101],
[102], and the habitat and colony of shags [38], [103]. The
responses of chinstrap penguins (Pygoscelis antarcticus),
Antarctic fur seals (Arctocephalus gazella), and leopard seals
(Hydrurga leptonyx) to UAV overflights have has also been
observed [104], demonstrating that at different UAV flying
altitudes, the behavioral and physiological reactions of these
animals change throughout different breeding seasons. The
animals showed strong behavioral reactions with decreases
in the flying attitude of the UAVs. The reaction of penguins
grew during the breeding season (i.e., guard and molting
phases). Weddell seals (Leptonychotes weddellii) showed lowlevel
reactions to low-altitude (25-15-m) UAV flight missions
and modest wing speeds [105], and tiny UAV overflights at
25 m had a low influence on Weddell seals during the breeding
season. These studies show that high-resolution small
UAV photogrammetry is accurate and widely applicable for
assessing the ecological status (behavior and morphology) of
MARCH 2023 IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE
Antarctic animals without causing excessive interference, possibly
replacing more invasive data collection and providing
data support for the application of UAVs in Antarctic polar
research [66], [106].
UNMANNED AERIAL VEHICLES IN
ATMOSPHERE STUDIES
Almost all meteorological data were acquired from Antarctic
weather stations because of the severe environmental conditions
in the region. The advent of UAVs has significantly
eased meteorological data collection in some regional inaccessible
locations. UAVs equipped with meteorological
sensors can collect air temperature, relative humidity, pressure,
and wind speed data and use them to track changes
in the atmosphere above ice lakes and sea ice, temperature
of the atmospheric boundary layer (ABL), and interaction
between the atmosphere and ocean. Currently, relevant research
is mostly conducted in Terra Nova Bay, near the McMurdo
and Shyowa Stations on the Weddell Sea in Antarctica.
The " Aerosonde " UAV was used for the first time, in
2010, to study changes in the atmospheric layer above Terra
Nova Bay, local atmospheric and oceanic interactions, and
surface conditions of the adjacent ocean/sea ice [52]. This
was the first and longest (>17-h) UAV flight in Antarctica
during winter for atmospheric monitoring. Atmospheric
monitoring was used to collect air temperature, wind, pressure,
relative humidity, radiation, GPS, and operational aircraft
data and perform quality control for scientific purposes
[33]. The relationship among the air, sea ice, and ocean was
also studied and provided a 3D dataset of the atmospheric
conditions (such as air temperature, humidity, pressure,
and wind) and surface temperature of Terra Nova Bay [31].
The Small Unmanned Meteorological Observer (SUMO)
drone was used near the McMurdo station to observe the
temperature profiles of the ABL, which provided short-term
dynamic changes in the ABL temperature [53]. Japanese researchers
created a novel device mounted on a gliding UAV
for aerosol observation and sampling; the device was propelled
by a balloon to reach the sampling height [107]. It
could autonomously detach from the balloon at the target
height and glide back to the ground station, from where it
could be recovered using a parachute. At the Shyowa Station,
this UAV system was successfully tested at a height of
over 10,000 m, which is also the highest recorded height in
the Antarctic [32], [107]. The changes and flux of the ABL
above the sea ice in the Weddell Sea region were investigated
using combined fixed- and rotary-wing remotely piloted aircraft
systems [49], [108]. Using a UAV and radiosonde for
assimilation in the Polar Weather Research and Forecast
model mostly showed an improvement in the analysis of air
temperature, wind speed, and humidity at the observation
point but was limited by the vertical extent of the UAV observations.
The impact of the assimilation was limited to the
lowest 1-2-km layer, and radiosonde data assimilation was
more conducive to simulating sea level pressure and nearsurface
wind speed [34], [109].
83
IEEE Geoscience and Remote Sensing Magazine - March 2023

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - March 2023
