Applications of unmanned aerial vehicles in Antarctic environmental research

Identification and characterization of biotic and abiotic components in a penguin colony using RGB and multispectral cameras

Figure 1 shows two image mosaics of a Chinstrap penguin (Pygoscelis antarcticus) colony, composed of 3800 pictures taken during a 29-min flight at 100 m altitude with a multispectral camera (MicaSense RedEdge-MX) using RGB bands (i.e., Red-668, Green-560 and Blue-475) (Fig. 1A) and the 10 wavelength bands covering the spectrum from visible to near-infrared light (Fig. 1B). With a resolution of 6 cm/pixel, penguin nests are clearly visible in the RGB mosaic, which are characterized by the absence of vegetation and with a predominant pink/brown color due to the abundance of guano deposition. The RGB mosaic also shows snow patches (white color), moss beds (green color) and one small lagoon with a bloom of red-pigmented greenalgae (Chlorophyceae) (Fig. 1A, upper right corner). Red algae (Chlamydomonas nivalis) patches on snow and ice are visible by zooming into a region of ice (Fig. 1A). More detailed information is obtained when the light spectrum from visible to near-infrared is used. Using the 10 wavelength bands, a thematic map was generated with the QGIS software and using a non-supervised classification method (Fig. 1B). Here it is possible to distinguish up to four species of mosses and three types of penguin guano that was verified with field observations.

Deception Island harbors up to 54 species of mosses, of which 13 species (including two endemics) have not been recorded elsewhere in the Antarctic. This, together with eight species of liverwort and 75 species of lichen, makes Deception Island an exceptional and unique place in Antarctica with legal protection under the Antarctic Treaty³. The use of a multispectral sensor onboard the UAV provides unique information to detect, classify and monitor moss beds without anthropogenic impacts. Antarctic moss bed health has already been assessed using multispectral sensors onboard UAVs¹². Taxonomic identification would be feasible by indentifying previously each species in the field and later assigning the spectral signature using the UAV, as recently suggested by Miranda et al. (2020), who monitored lichens and mosses in the Antarctic using a combination of satellite imagery and UAVs¹³.

Penguin guano has been suggested to be an important source of bioactive metals (e.g. Cu, Fe, Mn, Zn) for the sea surface waters, potentially fueling primary production of the Southern Ocean¹⁴. It has been suggested that the penguin species that feed mainly on Antarctic krill (Euphausia superba) (i.e., Chinstrap: Pygoscelis antarcticus, Adélie: Pygoscelis adeliae and Gentoo: Pygoscelis papua) excrete the highest concentrations of these bioactive metals¹⁵. Guano from these three congeneric penguin species has revealed the presence of microplastics across the Antarctic⁵. However, in order to estimate the magnitude of penguin fecal products that reach the sea, it is necessary to quantify the amount of guano excreted by the penguin colonies on land. This is possible with the multispectral reflectance data obtained from the UAV, which not only identify the guano coverage but also distinguishes different types of guano. Guano color is the result of diet, which, in turn, is related to the phase of the breeding cycle; therefore, a diet rich in krill is characterized by an excretion of pink guano, while a diet predominantly based on fish implies white guano¹⁶. Dark guano is the result of the mixture of guano with the soils that produce mud during wet precipitation.

It is increasingly common in the Arctic and Antarctic to find well-developed algae blooms as highly visible red patches on the snow surface caused by red-pigmented green algae (Chlorophyceae), and that produce the phenomenon commonly-known as red snow¹⁷. These algal blooms play a crucial role in decreasing the snow-surface albedo and, consequently, accelerating the melt rate, as well as in nutrient and carbon cycling^18,19. Mapping and monitoring the extent of snow algal blooms have so far been focused on satellite remote sensing; however, the spectral, temporal and spatial resolution of multi-spectral satellite imagery limits the study of most snow and ice algae¹⁸. Images taken from our UAV can enable the detection of patches of red snow on the surface snow with centimetric resolution (Fig. 1A). In addition, the image mosaic reveals the existence of a red snow bloom in a small pond located in a valley inside the colony (Supplementary Fig. S1). To the best of our knowledge, the existence of this bloom has not been previously reported and its monitoring could provide relevant information about the formation and proliferation of this bloom and its impact on cryospheric environments.

As a whole, the image mosaic of the Chinstrap penguin colony in Vapour Col (the second largest breeding colony in the island with about 12,000 pairs of penguins²⁰) may provide unique information about the different ecological niches linked to a penguin colony and their interactions. For example, the distribution and type of guano as nutrient and metal sources could be influencing the distribution and speciation of the flora in the area.

3D geological formation using RGB camera

Deception Island is a complex volcanic system formed as a result of the explosive eruption of basaltic-to-andesitic magmas²¹. Among its multiple structures and stratigraphy, we surveyed the Murature formation, a consolidated andesitic lapilli tuff²². Using the quadcopter with a RGB camera and the software Pix4D we created a 3D photogrammetry of the Murature formation (Fig. 2; Supplementary Movie S1). The software uses a Structure from Motion photogrammetry algorithm, where obtained 3D points are interpolated to form a triangulated irregular network in order to obtain digital Surface model (DSM). This DSM is then used to project every image pixel and to calculate the georeferenced orthomosaic. For the Murature formation, the photogrammetry was generated with 843 pictures obtained from three 20-min flights at an altitude of 40 meters, taking pictures from two different angles to obtain the heights of the features (60° and 90°). With 1.4 cm/pixel resolution the resulting mosaic provides a unique view of the geological formation that will support the study of how the rocks were formed and its evolution in relation to the various geological processes that occurred on the island. 3D photogrammetry is also useful in geomorphological research. Specifically, in Deception Island morphometrics studies of landform (e.g. Crater and cone diameters, depths, slopes, heights, etc.) are useful to estimate the eruptive recurrence of the island, and in turn, for advising volcanic hazards²³.

Thermal imagery to estimate animal abundance and to detect thermal anomalies

The combination of UAV technology with a thermal-imaging camera is very useful for studying and monitoring wildlife and thermal anomalies on Deception Island. Chinstrap penguin and fur seal (Arctocephalus gazella) heat signatures were detected at Vapour Col and Baily Head, respectively (Fig. 3A, E). Figure 3A shows a mosaic from a Vapour Col section composed of 336 images taken with a thermal camera (FLIR Vue Pro R) onboard the hexacopter during a 29-min flight at 100 m altitude, whereas Fig. 3C shows one thermal picture of fur seals at Baily Head. Penguins and fur seals, with a thermal signature of 15 °C and 26 °C, respectively, are clearly identified. Penguins are highly sensitive to climate change and are considered “marine sentinels” for quantifying environmental change in the Southern Ocean²⁴. However, the distribution and population dynamics of species such as the Chinstrap penguin are not well understood, mainly because they nest in remote and rugged areas, on-the-ground census work is difficult and sporadic²⁵. As demonstrated for Adelia penguins²⁶ the use of thermal imagery would allow reliable population estimates of Chinstrap penguins. Even, the use of RGB aerial images for animal counting would be far more accurate than from land-based surveys. Nevertheless, the scientific challenge is to develop a machine learning algorithm that can distinguish between animal species, based on their morphology and unique thermal fingerprint, which is only feasible using the high resolution provided by UAVs.

Other useful application of thermal cameras onboard UAVs on Deception Island is the easy and precise detection and monitoring of thermal anomalies. Figure 3B–D shows a thermal picture of one of the multiple fumaroles on the island, reaching temperatures above 90 °C. Seismic monitoring of volcanos on Deception Island has being ongoing since 1986, including many recorded volcano-tectonic earthquakes, long-period events and volcanic tremor²⁷. There have been six documented volcanic eruptions on the island between 1841 and 1971²⁸, nowadays volcanic and geothermal activities are limited to fumaroles and hot sands. Monitoring of these fumaroles using UAVs can provide a key in surveillance for early warming systems alerting of volcano activity on the island. UAVs not only accurately detect changes in temperature but also allow the increase in monitoring frequency when required.

Surface water sampling

UAVs provide unique opportunities for remote sample collection from surface waters, particularly in harsh or dangerous environments. Using a surface water sampling device described in the sampling and method sections we collected filtered fresh and saline surface waters at: (1) Three locations in Crater Lake (Fig. 4A). Crater Lake is part of the Antarctic Specially Protected Area (ASPA 140) due to its exceptional botanic and ecological value³. The use of drones for water sampling avoids human disturbance through the transportation and use of infrastructure, such as inflatable boats, and the risk that they pose to the natural ecological system. (2) One and six coastal locations in the Vapour Col and Baily Head penguin colonies, respectively (Fig. 4B, C). Access to the coastal zone inhabited by penguins requires approaches by boat (often assisted by an oceanographic vessel). The approaches do not only disturb the penguins that enter and exit the colony but, due to the coastal orography and waves, also dangerously hinders such an operation. The surface water sampling device onboard the UAV allowed in-situ water collection, minimizing the risk of impact on flora and fauna, limiting water disturbance and preventing contamination in the trace metal analysis. Attached to the sampling system we included a small multiparametric instrument referenced with time and GPS position to measure ancillary parameters, such as conductivity, temperature and depth (CastAway-CTD^®) (Fig. 4D). The aerial water sampling has been validated for trace metal analysis using ICP-MS by comparing metal concentrations of samples collected in a saline pond with the surface water sampling device onboard the UAV (i.e. average ± SD, n = 3; Ti: 0.20 ± 0.09; V: 1.92 ± 0.07; Cr: 1.5 ± 0.1; Mn: 19.4 ± 0.4; Fe: 11.6 ± 0.5; Cu: 1.9 ± 0.2; Zn: 0.5 ± 0.3; all values in ppb) and the traditional peristaltic pump system used from land or on boats²⁹ (i.e. average ± SD, n = 3; Ti: 0.20 ± 0.06; V: 1.93 ± 0.09; Cr: 1.3 ± 0.1; Mn: 19.1 ± 0.3; Fe: 11.8 ± 0.3; Cu: 2.1 ± 0.4; Zn: 0.4 ± 0.3; all values in ppb).

Deception Island is an example of the complexity of Antarctic environments, where environmental research studies need to deal with the inter- and multi-disciplinary analysis of processes, such as volcanic and geothermal activities, limnological process from its multiple lakes and ponds, sparse and exceptional flora and diverse fauna, among other. UAV surveys on Deception Island have demonstrated that this technology may substantially contribute to the progress in environmental biological, geological and chemical studies. UAVs permit researchers to study environmental processes at smaller spatial and temporal scales compared to other remote platforms (e.g. satellites), in a more cost-effective and safer way than on foot studies. Furthermore, they are less invasive and less disturbing to wildlife and the ecosystem. The simultaneous use of multi-sensors for multiple applications and the development of algorithms based on images obtained from the drone to detect, classify and count animals in real time are the new challenges that would significantly contribute to the study of the functioning of the Antarctic ecosystem and its ongoing environmental processes.

Source: Ecology - nature.com