Evaluating the Use of Unoccupied Aircraft Systems (UASs) for Planetary Exploration in Mars Analog Terrain

We developed UAS operations based on conceptual designs for an MSH (W. Johnson et al. 2020; J. Bapst et al. 2021). The MSH would build on the Ingenuity technology demonstration (J. Balaram et al. 2021) by adding scientific instrumentation. As a result, the proposed MSH would carry considerably more payload than Ingenuity, with a potential flight time of approximately 6 minutes (before needing to recharge the battery) when carrying a 5 kg scientific payload (W. Johnson et al. 2020). To fly this mass, the future MSH would need to be much larger, approximately 31 kg unburdened, compared to Ingenuity's 1.8 kg. To simulate the capabilities of an MSH-like UAS, our simulation used instruments on board two commercial UASs along with handheld instruments for landed observations.

We selected scientific instrumentation for the UAS that would allow the UAS Science Operations Team to (1) address science questions motivated by the RAVEN Science Traceability Matrix (STM; Table 1 and Section 2.3) and (2) assess the benefits and limitations of each instrument for use on a future MSH. The instruments selected for the UAS included (1) a fixed focal length airborne imaging camera, (2) a fixed focal length microimaging camera for landed images of the ground, (3) a laser-induced breakdown spectrometer (LIBS) to measure the elemental composition of the surface, (4) a point spectrometer in the visible light to near-infrared (VISIR) wavelengths to determine mineralogy, and (5) sampling capabilities, which included a scoop and drill and the ability to transport one collected sample at a time. We assume that an airborne imaging camera on board a future MSH would be mounted on a single-axis gimbal underneath the front of the UAS body, allowing for look angles ranging from 0° (horizontal) to 90° (vertical, or nadir). A free axis for rotation (yaw) is considered unnecessary for near-term MSH designs because the pointing direction of the camera can be controlled by pivoting the orientation of the UAS. A third axis (roll) of rotation can help to orient the camera to the horizon but is not strictly necessary given other forms of camera stabilization (e.g., R. Ljubicic et al. 2021). The sampling apparatus for a future MSH would be mounted underneath the center of the UAS body and collect a sample from the surface upon which the UAS landed. The microimager, LIBS, and VISIR spectrometer would also be mounted underneath the UAS and aimed at the same location below the UAS to allow colocation of data collection and sampling.

Energy is a limiting factor for UAS operations on Mars. The MSH proposal includes a battery with 1.2–1.4 MJ (333.33–388.89 Wh) of usable energy (70% of total stored energy to allow for reserves and extend battery life; W. Johnson et al. 2020). We used the lower bound of 333.33 Wh sol^–1 as the power usage cap. A small amount of this energy is needed to keep the UAS warm throughout the sol. We apply the formula of W. Johnson et al. (2020), which scales sleep energy relative to Ingenuity by the cube root of the UAS mass. For flight, we set the power draw to use the available battery energy in ∼6 minutes to align with the hover-time estimates of W. Johnson et al. (2020) for a 5 kg payload. We determined the power usage for each instrument using estimates from analog instruments (e.g., Mastcam-Z and ChemCam) and created an activity dictionary (Table 2) and power budget calculators to plan each sol within the imposed limit.

We imposed two restrictions on data transmission during the mission. The first was a hard cap of 140 MB day^–1, and the second was a soft cap by applying an energy cost to transmission. The 140 MB day^–1 hard cap derives from data return rates from Mars 2020 ultra-high-frequency (UHF) communications through orbiting satellites such as the Mars Reconnaissance Orbiter, ⁹ assuming four to five satellite overpasses on any given sol. Unlike Ingenuity, which uses a lightweight, lower-power communications system reliant on Perseverance as a relay for data transmission (J. Balaram et al. 2021), our UAS must be capable of operating independently. Like Perseverance itself, the UAS requires a higher-power transmission system capable of communicating with an orbiting satellite. As a compromise for using a lower mass than Perseverance to meet the requirements of a 0.1 kg transmission system from W. Johnson et al. (2020), we lowered our bit rate to 420 kbps (compared to a 3 kg transceiver capable of 2 Mbps for Perseverance ¹⁰ ) while keeping the same power usage of 65 W.

Following an initial examination of potential mission start locations and exploration areas, a start location was chosen near the northern margin of the Holuhraun lava flow field (Figure 2) that balanced mission priorities with logistical constraints, trafficability constraints, and hazard assessments for the rover and UAS (see below). A separate hazard analysis (Section 2.3.1) and mapping of regions of interest for science investigations (Section 2.3.2) were performed for the stand-alone UAS mission simulation, separate from the rover mission simulation.

Zoom In Zoom Out Reset image size

The Science Operations Teams for the concurrent rover and UAS stand-alone simulated missions worked together to create shared science goals, operational goals, and practical considerations that would be followed by both vehicles to allow direct comparison of mission metrics (S. Gwizd et al. 2024). Shared operational constraints included that the missions should begin at the same location, each last at least 10 implemented sols, and include collection of at least three samples to be cached together in one location. The mission start location (Figure 2) was restricted to places that were accessible via road by the truck transporting the rover and close enough to the 2014–2015 Holuhraun lava flow field that the rover could reach the flow margin during a 10 sol mission (∼500 m; S. Gwizd et al. 2024). Once the mission simulations began, the rover and UAS Science Operations Teams operated in parallel but independently.

The science objectives for the missions were guided by the RAVEN project STM. Table 1 summarizes these objectives, which are linked to priorities outlined in the "Planetary Science and Astrobiology Decadal Survey 2023–2032" (National Academies of Sciences, Engineering, & Medicine 2023) and the Mars Exploration Program Analysis Group (MEPAG) Science Goals document (MEPAG 2020). The objective at the end of the mission was for the Science Operations Teams to use the data collected to develop an interpretation of the mission terrain, including a description of the chemical composition of the geologic units present, a geologic history of the region, and an assessment of the potential habitability of the area.

The rover and UAS Science Operations Teams conducted a premission analysis of the mission area to choose a mission start location, science targets, and an initial mission plan. For these tasks, the Science Operations Teams were provided orthomosaics at 0.25 m pixel^–1 and digital elevation models (DEMs) at 1.0 m pixel^–1. Data sources included UAS and airborne survey products (J. R. C. Voigt et al. 2021a; S. P. Scheidt & C. W. Hamilton 2021; Loftmyndir ehf.) that were reprocessed such that the resolution matched that of the DEMs and orthomosaics produced from stereo images captured by the High Resolution Imaging Science Experiment (HiRISE) camera on the Mars Reconnaissance Orbiter (A. S. McEwen et al. 2007; S. S. Sutton et al. 2022). The Science Operations Teams were also provided with two 1:800-scale geomorphological maps illustrating (1) different surface facies of the 2014–2015 Holuhraun lava flow field (J. R. C. Voigt et al. 2021a) and (2) context outside of the Holuhraun lava flow field. Both mapping products are based on differences in color (i.e., albedo at red, blue, and green wavelengths), surface texture, and morphologies as they appear in a 0.2 m pixel^–1 UltraCam-Xp base map (acquired on 2015 September 6; J. R. C. Voigt et al. 2021a).

Using these data sets, the Science Operations Teams determined that the region was dominated by volcanic landforms, including lava that covers the southern half of the exploration area (Figure 2). Sand mantles much of the area. The Science Operations Teams created a premission surface morphology map focusing on descriptive terms and avoiding interpretations of the different units (Figure 2(a)). This map identified two dominant lava morphologies (rougher and smoother), two sediment morphologies (pockmarked terrain and smooth sediment), and areas of interest (blue outlines; Figure 2(a)) for the hydrologic objectives in the STM (Table 1). Further analysis of the provided data sets and discussion among the Science Operations Teams determined that the pockmarked terrain was likely due to the sediment covering older lava (in contrast to the younger lava that was clearly visible), and this surface was generally referred to as the "older lava" during and after the mission. These initial observations and the STM (Table 1) led the UAS Science Operations Team to attempt to answer four key questions during the mission: (1) what is the geologic history of the region? (2) What was the emplacement sequence for the younger lava? (3) How active is eolian transport and what is the source of the sediment? and (4) Is there evidence of past and/or present hydrologic activity?

2.3.1. UAS Landing Site Requirements

2.3.2. Identification of Regions of Interest

2.3.3. Operations Strategies

The UAS Science Operations Team oversaw all planning for the UAS mission and consisted of four members with roles of Lead, UAS Planner, Science Planner, and Documentarian. The Lead conducted planning meetings, made final decisions on UAS plans for each sol, ensured that the mission guidelines were followed, and worked with the UAS Field Implementation Lead to discuss execution of the daily plan and data downlink. The UAS Planner ensured that activities for the UAS in each sol fit within the power and data budgets (Table 2). The Science Planner worked with the UAS planner to translate UAS goals for a sol plan into specific activities and kept track of the STM to ensure that objectives were being met. The Documentarian kept records of all UAS Science Operations Team activities.

The UAS Field Implementation Team was charged with implementing the mission plan created by the UAS Science Operations Team and consisted of two Co-Leads, operators for the LIBS, VISIR spectrometer, and sampling instruments, and a Documentarian. The Co-Leads conducted the plan translation with the Science Operations Team Lead, operation of the UAS for the planned sol, and data downlink.

The UAS Science Operations Team followed a scripted structure for planning each sol's operations based on the format used by the Mars Science Laboratory and Mars 2020 missions (S. M. Milkovich et al. 2022; A. R. Vasavada 2022). Planning began with an assessment of downlinked data from the previous sol and mission progress to date. The bulk of the planning meeting consisted of identifying science targets for the next sol and building a flight and science activity plan that fit within the daily power and data budgets. Plan building typically required multiple iterations as activities were added, modified, or removed until the plan satisfied both the resource constraints and the objectives defined by the team for that sol. Planning concluded with a forward-looking discussion focusing on "N + 1" and "N + n" concepts for the subsequent sols (where "N" represents the current planning sol and "n" represents any number of sols in the future). The planning meeting was restricted to a duration of 2–3 hr to meet logistical constraints for the implementation of the plan at the field site each day.

Plan translation, the communication of the sol plan created by the UAS Science Operations Team to the UAS Field Implementation Team, occurred following the science planning meeting via a discussion and exchange of digital files between the team leads. The UAS flight plan was created using the universal Ground Control Software (UgCS)^TM, which allowed the UAS Science Operations Team to automate the precise locations and orientations of UAS takeoff and landing, imaging, and mapping surveys. The flight plan file was copied onto the UAS controller for activation in the field. Other instructions and a summary of data requested to be downlinked were communicated to the implementation leads via a Plan Translation Form. Following the plan translation meeting, no further communication between the operations and implementation teams was permitted.

A DJI Matrice 300 RTK^TM UAS with a DJI Zenmuse P1^TM 35 mm fixed focal length camera was used to implement the flight plan created by the Science Operations Team and capture images. An Apple iPhone 13 Pro^TM was used for the microimager. Images from the P1 camera were downsized from ∼20 to ∼8 MB each using Apple Photos^TM prior to downlink. Images from mapping surveys were processed using Agisoft Metashape Pro^TM v1.8 to create orthomosaics and DEMs. Survey data products were exported at resolutions that were determined by the UAS Science Operations Team to fit within the data budget for downlink. The geotagged image locations provided spatial reference for the maps. Geotagging was enabled using a DJI Differential Real-Time Kinematic D-RTK 2 Mobile Station^TM that was paired to the Matrice 300 UAS. Global navigation satellite system (GNSS) positioning is unrealistic for Mars applications, but it was assumed that a future MSH would have localization capabilities that would enable geotagging of UAS data relative to landmarks in geoencoded HiRISE images.

Two separate DJI Matrice 600^TM UASs were modified for sampling, one equipped with a drill (for hard rock) and another with a variety of scoop attachments (for sediment). Due to the limitations of the UAS drill prototype and the risk of damaging the drill bit during basalt sampling operations, a hand drill was used instead when a lava sample was requested. Storage, transport, and caching of the samples by UAS were not simulated. Instead, after collection by scoop or drill, sample material was placed in labeled bags by the UAS Field Implementation Team.

The LIBS was a SciAps Z300^TM handheld LIBS analyzer with a wavelength range of 200–900 nm. Five separate spectra were collected for each LIBS target designated by the UAS Science Operations Team. Each spectrum resulted in measurements of elemental abundance that were then converted to normalized oxide abundance and downlinked to the UAS Science Operations Team.

The VISIR point spectrometer was a Spectral Evolution OreXpress^TM field spectrometer with a wavelength range of 350–2500 nm and spectral sampling of 2.8 nm channel^–1 between 350 and 1000 nm and 6–8 nm at >1000 nm. VISIR data were collected with 1 s integration time, averaging 10 scans, and corrected to reflectance with internal dark current subtraction and division by a Spectralon white reference collected just before sample collection. Further processing to generate spectral plots and parameters to give to the UAS Science Operations Team were done in python using the HyPyRameter package (Phillips et al. 2024).

The UAS Field Implementation Team Co-Leads initiated the UgCS^TM flight plan created by the UAS Science Operations Team, and the Matrice 300 RTK^TM UAS conducted the flight plan automatically. Matrice 300 RTK^TM landings were conducted manually within a 5 m radius of the landing location that was defined by the UAS Science Operations Team. Flights were only conducted when environmental conditions were optimal for operation of the UAS and the presence of the Field Implementation Team at the site (e.g., winds and temperatures within the specifications of the Matrice 300 RTK^TM and no active precipitation). Given the short duration of the flights (<6 minutes), an acceptable flight window occurred on most days, except for when inclement weather forced the complete cancellation of all field implementation activities. The LIBS, VISIR spectrometer, and microimager were operated by UAS Field Implementation Team members who would walk to the landing site following flight.

The UAS mission began in a large low-hazard area ∼200 m northwest of the margin of the younger lava (Figure 3). This site was ∼600 m east of the mission start location that was selected by the Science Operations Teams prior to deployment (Figures 2 and 3). The decision to change the mission start location was made because the truck delivering the rover to the field site could not transit the final few hundred meters to the chosen start location, and mission parameters dictated that the UAS and rover stand-alone missions begin at the same site. This new location was still within the exploration area that had been initially characterized by the rover and UAS teams.

Zoom In Zoom Out Reset image size

The original plan for sol 101, the first day of simulated mission operations, had been to focus on the older lava and sand sheet (Figure 2(a)). However, due to the deviation from the proposed starting location, a new area of scientific interest was now within range of a single UAS flight. This was a depression filled with sediment and boulders to the northeast of the mission start site classified in premission mapping as the "smooth sediment" surface (Figures 2(a), 3, and 4(a)). Despite being of high scientific interest during the premission analysis, this location had not been prioritized in planning because it was too far away from the proposed mission start location (Figure 2). The actual mission start location (Figure 3) provided quick access to this area without impacting the UAS Science Operations Team's ability to address other mission goals. The sediment in the depression lacked the outcroppings of the older lava in the pockmarked terrain unit (Figure 2(a)), suggesting that either the older lava was not present as a substrate or the sediment fill was thicker in this area. The UAS Science Operations Team took advantage of the unexpected proximity of this terrain and planned a mapping survey of the area for sol 101. Additionally on sol 101, the UAS captured oblique airborne photos looking to the southeast toward the younger lava. The UAS Science Operations Team used these images to scout for landing sites for future sols and correlate the landing suitability map to visual observations of the area. After considering the 10 sol mission timeline and the outstanding objectives during "N + n" planning, the team decided not to land and conduct science activities in the smooth sediment area in favor of moving toward the younger lava.

On sol 102, the UAS Science Operations Team resumed its intended plan of investigating the older lava and sand sheet. The UAS conducted a mapping survey of the area between the landing site and the younger lava margin (Figure 3). The UAS landed on a sandy area between the younger lava margin and an outcropping of the older lava. Descent images of the sol 102 landing showed distinct areas of lighter and darker sediments (Figure 4(b)). The surface also showed a series of low-relief ripple or dune forms (approximately centimeter-scale amplitudes and meter-scale wavelengths; Figure 4(b)). A sample was taken on sol 103 using the scoop that included both lighter sediment visible on the surface and darker sediment exposed underneath (Figure 4(c)).

Zoom In Zoom Out Reset image size

The sol 104 plan included oblique airborne images of the younger lava to characterize in detail the smoother and rougher surfaces identified in premission analysis (Figure 2(a)) and identify transitions between these textures that might indicate how the younger lava was emplaced. Images from the UAS revealed that the rougher texture was caused by heavily fractured thin plates of lava flow crust (Figure 4(d)), while the smoother texture was shown to be a more contiguous flow crust with sets of parallel lineations and long cracks (Figure 4(e)). The sol 104 landing site was chosen as the cache location for samples collected during the mission, and the sol 103 sediment sample was deposited here. Photos of outcrops of the older lava captured during sol 104 confirmed that these outcrops were too rough for a UAS landing, an operational constraint the UAS Science Operations Team had suspected based on premission hazard analysis and images from sols 101 and 102. This would prevent the mission from sampling or conducting landed science on the older lava.

Photos from sol 104 were used to identify locations on the younger lava for two separate mapping surveys and a landing site on the lava itself. The mapping surveys targeted a location where the lava surface transitioned between the smoother and rougher textures and the lava landing site (Figures 2(a) and 3). The landing site was chosen because the sol 104 photos confirmed that it was the largest highly suitable area (Figure 2(b)) on the smooth surface of the younger lava that was accessible to the UAS Implementation Team. Both the mapping surveys and lava landing were executed during the sol 105 flight.

Sols 106–108 involved science activities on the younger lava surface. A sample of the lava was taken on sol 106. Descent and landed images from sols 105 and 106 showed that the lineations in the smooth lava surface texture (Figures 4(e) and 5(a)) were overlapping folds in the flow crust with a relief of tens of cm (Figure 5(b)). Microimages of the landing site showed a centimeter-scale roughness with sparse vesicles and slightly elongate, spinelike protrusions on the lava flow surface (Figure 5(c)). The N + 2 plan for sol 107 created during the sol 105 planning meeting called for the UAS to conduct a short flight (∼10 m) and land on a surface with white discoloration to conduct landed science activities. However, upon review of the landed images from sol 105 (Figure 5(b)), the UAS Science Operations Team decided that the selected sol 107 landing site was not viable because the relief on the lava crust surface could cause the UAS to destabilize and crash during landing. Instead, the UAS conducted a short reconnaissance flight on sol 107, returning to land at the same location that it took off from. The surveys flown on sol 105 were also transmitted on sol 107. Using the sol 105 survey and images from sols 104 and 107, a suitable location to land on a white discoloration on the lava's surface on sol 108 was identified (Figure 5(d)). Premission analysis showed that these white discolorations were common in the immediate area (Figures 2(a), 3, and 4(e)) and were prioritized for observation by the UAS Science Operations Team for their potential to represent postemplacement mineral precipitation or alteration processes on the younger lava surface that could inform environmental conditions and the habitability potential of the area. The UAS flew ∼100 m to the chosen location, landed, and conducted landed science activities there on sol 108 (Figure 5(d)).

Zoom In Zoom Out Reset image size

At this point in the mission, the UAS had collected images of targets on all four surface types identified premission and conducted landed science on two (Figures 2(a) and 3), achieving many of the mission science objectives (Table 1). The UAS Science Operations Team decided to prioritize two objectives during the remaining mission sols. The first would use airborne images to provide additional details on lava and sand sheet features cursorily identified in previous sols. The second would identify a location for the third sample and landed science that would help meet hydrology objectives, potentially by traversing further to the west or by returning to the smooth sediment plain surveyed on sol 101.

The UAS returned to the sample cache on sol 109 to deposit the sol 106 lava sample. Images taken on sol 109 aimed to further characterize the surface textures and topography between the younger lava margin and interior. These images and those captured on previous sols showed that the smooth surface was similar in appearance throughout the younger lava. However, the smoother surface in the interior was more expansive and located on a plateau-like topographic high (Figure 6(a)) compared to along the margin where the smoother texture was found on separate meter-scale flow lobes and broken by areas of rougher texture (Figures 6(a) and (b)). On sol 110, the UAS imaged older lava outcrops on the sand sheet to constrain observations from previous sols showing preferential accumulation of sediment on the east side of rock outcrops (Figure 6(c)). The UAS landed on a small marginal flow lobe on sol 110 (Figure 6(b)) to conduct landed science at a second location on the younger lava (landed science on sediment surfaces had already been conducted at three locations).

Zoom In Zoom Out Reset image size

The UAS Science Operations Team had decided to end the mission by returning to the smooth sediment plain visited on sol 101 and collecting a sample to achieve the hydrology goal. Returning to this location was preferred over the western location because of the additional contextual information the team had already collected, the significant resources required to fly further west, and the smooth sediment plain's better proximity to the sample cache. However, due to incoming inclement weather at the field site, this plan was not possible for the UAS Field Implementation Team to execute in the time remaining. Instead, the UAS Science Operations Team opted for a more easily implemented plan where the third sample required by the mission goals was collected on the younger lava on sol 111, where the UAS had already landed on sol 110.

Sol 112 acquired oblique airborne images of a plateau in the younger lava interior where the surface was observed in the premission data to transition between the rougher and smoother surfaces (Figures 2(a) and 3). This target was important because more detailed images of the contact between these surfaces taken by the UAS could provide key observations to inform the emplacement history of the younger lava (Figure 6(d)). The UAS then flew back to the east and landed at the sample cache to deposit the sol 111 sample. The simulated mission ended at this time after 12 sols.

In total, the UAS returned 86 images, nine LIBS measurements, nine VISIR measurements, and orthomosaics and DEMs from four surveys to the UAS Science Operations Team (Table 3). The UAS collected three samples, one of sediment and two of the younger lava (Figure 3). The images included 51 oblique in-flight images (e.g., Figure 6), 19 nadir images taken during ascent or descent (e.g., Figures 5(a) and (d)), nine landed horizontal images (e.g., Figure 5(b)), and seven images from the microimager (e.g., Figure 5(c)). Surveys covered a total of 70,000 m² and returned DEMs at 30 cm pixel^–1 resolution and orthomosaics at 6.0–9.6 cm pixel^–1 to fit within the UAS data budget. The resolution of the survey data represents an improvement by a factor of ∼3–4 for both the DEMs and orthomosaics over the resolution of the HiRISE-equivalent data provided to the Science Operations Teams prior to the mission (Figure 7).

Zoom In Zoom Out Reset image size

The LIBS, VISIR spectrometer, and microimager collected data at each new landing site during the mission (sols 101, 102, 104, 105, 108, and 110). After each sample was collected (sols 103, 106, and 111), a second set of landed science measurements were taken. Landed science measurements were taken at four different geologic units or features: (1) the younger lava (sols 105, 106, 110, and 111), (2) a white discoloration on the younger lava (sol 108), (3) lighter sediment (sols 101, 102, and 104), and (4) darker sediment (sol 103). The darker sediment measurement was from the postsampling landed science on sol 103, where the sampling scoop exposed darker sediment beneath the lighter sediment on the surface (Figure 4(e)). VISIR spectra show general similarities between the four target categories and an absorption band at ∼1000 nm for relatively iron-rich silicate minerals common in basaltic lavas such as olivine and pyroxene (Figure 8). The differences in the width of this absorption between surface types may be due to differing abundances of olivine relative to pyroxene, variation in olivine and pyroxene composition, or the presence of glass (J. B. Adams et al. 1974; E. A. Cloutis & M. J. Gaffey 1991; R. G. Burns 1993; R. L. Klima et al. 2011). The clearest absorption present in the VISIR data is that of H₂O (1950 nm and the overtone band at 1450 nm) in the darker sediment from sol 103 and the white discoloration of the lava on sol 108 (Figure 8). The LIBS oxide data showed a basaltic-andesite composition with low alkali content and ∼53% SiO₂ for all nine targets. However, we consider the LIBS measurements collected by the UAS Field Implementation Team to be unreliable, given that measurement error strongly correlated with measurement quantity, and there was no difference in measurements between the lava, sediment, and white discoloration on sol 108 (see the Appendix).

Zoom In Zoom Out Reset image size

The UAS flew a total of 10 km over nine flight sols (Table 3; the other three sols were used for sample collection). The longest flight was 1857 m on sol 104, during which the UAS imaged a variety of different lava textures (Figure 3). The shortest flight was 93 m on sol 108, when the UAS flew from the sol 106 lava sampling location to the white discoloration on the younger lava (Figure 3). Total flight time was 28 minutes, with the longest single duration flight of 5 minutes on sol 105, during which the UAS conducted two separate mapping surveys (Table 3; Figure 3). The shortest-duration flight of 39 s occurred on sol 108.

The UAS used >90% of its 333.3 Wh power budget on eight sols (101, 102, 103, 104, 105, 106, 111, and 112) and transmitted >90% of the 140 MB data budget on five sols (101, 102, 104, 107, and 112) (Table 3; Figure 9). The UAS plan was energy-limited—defined here as no further activities could be included in the plan because there was insufficient power to do them—on five sols (101, 103, 105, 106, and 111), which included the three sols when a sample was taken. The UAS plan was data-limited—defined here as no further activities could be included in the plan because there were insufficient data remaining to transmit the results—on three sols (104, 107, and 112). On sol 102, the UAS plan was both data- and energy-limited (Table 3; Figure 9). On sols 108, 109, and 110, the planned activities were neither data- nor energy-limited. On one occasion, the UAS Science Operations Team circumvented the data limit by caching processed mapping survey data on the UAS at the end of one sol (105) and transmitting it on a later sol (107) that had minimal data acquisition and low power consumption.

Zoom In Zoom Out Reset image size

Power was divided into "science" (operation of the instruments), "flight" (including takeoff and landing), "mapping surveys" (including flight, camera use, and processing time), "sampling" (use of the sampling apparatus), and "engineering" (data transmission, instrument start-up, maintaining vehicle temperature when not in use) categories (Table 2; Figure 9). Relatively long flights and mapping surveys used the most power (20%–64% and 38%–60% of daily cap; Figure 9(a)). Operation of science instrumentation (aside from sampling) used comparatively little energy (0.1%–2.4% of daily cap). Data volume was dominated by airborne images (11%–80% of daily cap; Figure 9(b)). On sols involving mapping surveys, substantial power (38%–60% of daily cap) and data volume (48%–72% of daily cap) were used on surveying, restricting any additional activities for that sol. Sampling also involved extensive power usage (84% of daily cap) and restricted additional activities on those sols.

Key science interpretations made by the UAS Science Operations Team using data collected during the UAS mission include the following.

• 1.  
  The terrain with smooth sediment and dispersed light-toned boulders surveyed in sol 101 (Figure 4(a)) is a flood deposit, potentially a glacial outburst flood, or jökulhlaup, given the location's proximity to the nearby Vatnajökull ice cap (Figure 1).
• 2.  
  The morphology of the younger lava was identified as "spiny" (or "toothpaste") lava (S. K. Rowland & G. P. L. Walker 1987), a transitional lava type indicative of emplacement of a relatively degassed lava at a slow rate. The plateaus and steplike topography of the younger lava suggests emplacement by a process of inflation and flow margin breakout of lava lobes.
• 3.  
  The white discolorations on the younger lava are hydrated mineral precipitates from fumaroles. These features represent the best location identified by the UAS Science Operations Team during the mission to preserve biosignatures.

4.4.1. Sediments

4.4.2. Lava

4.4.3. Geologic History

The UAS Science Operations Team developed a geological history of the mission area using observations from the UAS mission and the analog orbiter data. The oldest unit observed during the UAS mission is the smooth sediment plain with dispersed light-toned boulders interpreted to be part of a widespread glacial outwash flood deposit (Figure 10(a)). As the boulders are not observed anywhere else, the most recent flood to affect the area likely occurred prior to emplacement of the lava. Next, the older lava was emplaced on top of the flood deposit. Analysis of the analog orbiter orthomosaic suggests that this lava is a flow sourced from the northwest, whereas the younger lava is part of a flow field sourced from the southwest (Figures 1 and 10(a)), suggesting that these lavas come from different volcanic systems. However, the UAS was not able to collect any geochemical data or samples from the older lava that would have enabled this interpretation to be tested more thoroughly. Eolian transport of sediment has been active at least since emplacement of the older lava (and likely much longer), as the older lava flow had significantly more sediment mantling its surface (to the point where interpretation of its morphology is not possible) than the younger lava.

Zoom In Zoom Out Reset image size

Emplacement of the younger lava flow field is interpreted to have begun with its central pathway, located ∼1 km south of the mission area (Figure 10(a)). The lava observed by the UAS during the mission, referred to now as the northern flow lobe, was emplaced later from a source to the southwest near the vent, perhaps a levee failure or change in vent location or geometry that led lava to stop flowing down the central pathway and be redirected into the northern flow lobe. These interpretations are supported by observations in the analog orbiter orthomosaic and DEM of a trough between the northern flow lobe and the central pathway that includes gaps showing the preeruptive sediment surface between these flows (Figures 10(a) and (b)). A smaller trough is located between the northern lobe and the older lava (Figure 10(b)), suggesting that the northern lobe was emplaced in a topographic low between the older lava flow margin and the central pathway of the younger lava flow field. Furthermore, the two smaller northward-extending flow lobes that were the focus of the UAS mission have lower elevations than the northern lobe (Figure 10(c)). This topographic break trends east–west and appears to follow the boundary of the older lava flow and a fluvial area (orange line; Figure 10(a)). It is likely that these northernmost lava lobes formed only after the northern lobe rose above the older lava, allowing lava to break out from the inflation plateau and inundate the area. Thus, the younger lava surface investigated during the UAS mission may represent one of the final stages of emplacement during the eruption that formed the younger lava flow field.

The fumarolic deposits on the flow surface are concentrated on smoother surfaces of the younger lava flow-field plateau and are not found in the northernmost lobes. Much of the younger lava flow field, including the plateau, was potentially emplaced above a fluvial system that followed the south margin of the older lava flow, visible to the west of the younger flow in the analog orbiter orthomosaic (Figure 10(a)). The presence of lava above water may explain the concentration of fumaroles in this area and the lack of fumaroles in the northernmost lobes that were emplaced above the older lava flow, which presumably held less moisture.

Knowing the lava type and the width and thickness of the northern lava lobe (Figure 10(b)), the UAS Science Operations Team estimated the emplacement parameters for the younger lava flow field. S. K. Rowland & G. P. L. Walker (1987) show that spiny lava is emplaced at a flow rate of 10–100 minutes m^–1. The cross-sectional area of the northern lobe at the cross section location in Figure 10(b) is 2700 m². This results in a rough estimate of the eruption rate of ∼3 m³ s⁻¹ during emplacement of the northern lobe. The entire younger lava flow field has a surface area of ∼80 km². Extrapolating from the elevation of the ground next to the flow margins suggests that the lava may have an average thickness on the order of ∼20 m; thus, the flow field may have a bulk volume of ∼1.6 km³. Applying our 3 m³ s⁻¹ effusion rate estimate to the entire eruption—a major assumption—gives an eruption duration on the order of ∼15 yr. Given that eruption rates often peak in the initial phases of an eruption at orders of magnitude greater than rates observed in the waning phase of an eruption (G. Wadge 1981), and this extrapolation ignores the possibility of multiple flow lobes being emplaced simultaneously, this duration estimate was considered to represent a maximum value. We note here that the main phase of the 2014–2015 Holuhraun eruption actually lasted for only 180 days (i.e., from 2014 August 31 until 2015 February 27; E. Bonny et al. 2018).

4.4.4. Astrobiological Implications

Analysis of the orthoimages (0.25 m pixel^–1) and DEMs (1.0 m pixel^–1) of the mission area prior to implementation was crucial to mission success. These data were essential for mapping UAS landing suitability (Figure 2(b)), identifying the main surface units and their characteristics (Figure 2(a)), and creating initial hypotheses to test during the mission. The change in the mission start location from a site further west to the implemented location further east impacted the efficiency of the initial sols of the UAS mission. During premission planning, the UAS Science Operations Team had prioritized targets to the west and created tentative plans for the initial mission sols based on the assumption of this starting location (Figure 2(a)). When the start location changed (Figure 3), the team was less familiar with that area and had to perform additional work to identify potential targets concurrently with planning the UAS activities. A preferred and commonly implemented strategy for missions (but not performed by the UAS Science Operations Team here) is to map contacts between geological units and facies to identify regions of interest over a larger area rather than only identifying specific targets.

5.2.1. Images

5.2.2. Photogrammetric Surveys

5.2.3. Landed Science

The UAS Science Operations Team developed a "reconnaissance-then-science" strategy, where they planned a long flight to capture as many images (∼15 sol^–1 given the image resolution and data limit) of as many sites of interest as possible over a large area. The images from this flight were used to identify the optimal locations to target on subsequent sols for a mapping survey, landed science, and/or landing. This strategy is shown in practice on sols 104–108 of the UAS mission (Figure 3). The long sol 104 flight allowed the UAS Science Operations Team to identify areas to survey and land on sol 105, sample on sol 106, and conduct landed science on sol 108. This strategy combines the positive aspects of both covering a large area with sparse observations for context and conducting activities at select targets for in-depth investigations. Reconnaissance flights improve mission safety by confirming landing suitability and increase scientific return by helping identify ideal locations for surveying and landed science. This strategy is conceptually similar to the "walkabout" strategy tested by R. A. Yingst et al. (2020) for rover operations and preferred over a "linear" approach for similar reasons discussed here; the "walkabout" approach provided more regional context and greater confidence in interpretations. Applying a "walkabout" strategy is potentially even more advantageous for UAS operations (or joint rover–UAS operations), as the UAS can acquire the contextual information faster compared to a rover, likely using a single flight at relatively high altitude to capture airborne images of a large region of interest.

The "reconnaissance-then-science" strategy also worked well with the mission constraint that the UAS return to a cache location after collecting each sample. The "tethering" of the UAS to a central location allowed for retargeting of features from different perspectives or detailed targeting of new features of interest identified in downlinked images and contributed to a greater understanding of the geology of the mission area. The requirement to cache samples was among the rationale for the UAS Science Operations Team's decision to favor focused investigations at high-priority locations, as opposed to a "linear" approach to maximize UAS flight distance, which was discussed during premission planning. For a longer mission, a cache requirement would impose a more significant constraint on the ability of the UAS to cover terrain, and the Science Operations Team would have to prioritize between collecting more samples in the immediate area (if the sampler design allowed for more samples) and exploring new terrain further away.

Flexibility in altering intended sol plans was essential for achieving mission success. The revised planning during sols 107–108 highlights a case where our landing suitability mapping (Figure 2(b)) and oblique image scouting were insufficient to identify potential hazards. The steplike topographic breaks created by the folds in the spiny lava texture (Figure 5(b)) were averaged out across the 1.0 m pixel size of the DEM, and the airborne images from sol 104 only viewed the landing site from one direction, where the relief was difficult to discern (Figure 4(e)). This example demonstrates that subpixel-sized relief can impact UAS operations even when the pixel size is of similar or smaller dimensions relative to the vehicle itself. Mitigation of the risk posed to UAS landings by smaller-scale but sharp topographic relief can involve three components. First, scouting of landing sites during previous sols should include viewing potential landing zones from multiple directions and altitudes. Second, prior characterization of the surface unit where a landing is proposed will help identify on which surface types hazardous small features are more likely to exist. Third, the UAS should be equipped with both an autonomous landing site selection system (e.g., R. Brockers et al. 2021) and robust landing gear capable of adapting to any surface features too small to be detected at the resolution of the data available to the Operations Team.

Data collected by UAS were essential to developing an interpretation of the area's geology at the level of detail described in Section 4.4 The analog orbiter orthomosaics and DEMs provided prior to the mission simulation provided an initial characterization of the terrain and guided the development of hypotheses to test during the mission but could not resolve the details or provide data types that led to the key science conclusions of the mission (Section 4.4). For example, it was not possible to determine the spiny lava type of the younger lava from the analog orbiter data. However, the higher-resolution images and surveys from the UAS (Figures 6 and 7) were able to observe the diagnostic features (Section 4.4.2; Figure 6) that led to this interpretation. The emplacement conditions of spiny lava are significantly different from those for 'a'ā and pāhoehoe lava types (S. K. Rowland & G. P. L. Walker 1987); thus, this interpretation made possible by UAS data had significant implications for the UAS Science Operation Team's understanding of the formation of the younger lava flow field.

Data from the UAS mission helped confirm, elaborate on, or explain observations and hypotheses from the analog orbiter data. The possible interpretation of the white discolorations as mineral precipitates was identified during premission planning, but UAS images were needed to confirm that the color was a coating on the lava surface and the UAS VISIR spectrometer provided observations of a composition different than that of the lava with evidence for hydration (Figure 8). The identification of these features as fumaroles on the lava surface was the most significant mission finding related to astrobiological studies and implications for habitability. Similarly, the improved resolution of the UAS surveys and images (Figure 7) allowed the UAS Science Operations Team to measure boulders in the flood deposit and wind ripples on the sand sheet. This analysis was not possible given the resolution of the analog orbiter data and helped confirm premission hypotheses for these terrains and add detail to the interpretation of their formation. The analog orbiter orthomosaics and DEMs were the main sources for interpreting the region's geologic history (Section 4.4.3; Figure 10), but details from UAS observations provided supporting information that improved the UAS Science Operations Team's confidence in their conclusions.

The UAS successfully completed a 12 sol mission and collected three samples, as required in the mission guidelines (Section 2.2). The UAS was able to achieve the science objectives outlined in the STM (Table 1) while following the operational and implementation limitations (Section 2). The detailed interpretation of data collected during the mission (Section 4.4) addressed at least in part all four of the key questions identified by the UAS Science Operations Team prior to the mission (Section 2.3) and constructed a geologic history of the area (Section 4.4.3). Every major surface texture in the area identified premission by the Science Operations Teams was captured in oblique airborne images and mapping surveys (Table 4). However, only the pockmarked terrain and smoother surface of the younger lava were targeted for landed science, as the other units were too rough for the UAS to land on. Three of four sample targets were achieved. Samples of the sediment (on sol 103) and younger lava (on sols 106 and 111) were collected, and, as a hydration signal was detected at the location of the sediment sample, this sample also accomplishes the hydrology objective (Table 1; Figure 8). No sample of the older lava was collected.

The greatest hindrance to further mission achievements were "testisms" imposed by the brief duration of the simulated mission and inclement weather encountered in the field, as well as limitations in personnel. Time constraints during mission implementation motivated the UAS Science Operations Team to continually move on to new areas to achieve all mission objectives and increased the consequences of small inefficiencies in the mission plan. For example, the team chose not to conduct landed science activities in the smooth sediment plain on sols 101 and 102 in favor of moving toward the younger lava, and the mission simulation ended before the UAS could return to this location to collect a sample. Although the size of the UAS Science Operations (four people) and Implementation (six people) Teams was appropriate for the tasks of creating and executing the UAS plan for each sol (Section 3.1), the UAS Science Operations Team would have benefited from a greater breadth of scientific expertise. The team lacked expert knowledge related to sedimentology and glaciology, which would have helped to inform the selection of science targets on the sand sheet and interpretations of the mission data. Additional Science Operations Team support would have enabled more data analysis during the mission (e.g., calculation of image ground sampling distances) and more precise mission plans (e.g., ensuring that the resolution of surveys would resolve key details at science targets).

UASs offer numerous advantages for exploring terrain in the Holuhraun region of Iceland. The broad homogeneity of the terrain, with contacts between different surface units spaced hundreds of meters apart (Figure 2(a)), and the variable roughness of these units (Figure 2(b)) makes traversing the area time-consuming and hazardous on the ground. The UAS was able to travel relatively large distances in short amounts of time (e.g., on sol 104, the UAS covered 1857 m during a flight lasting 3 minutes and 41 s; Figure 3 and Table 3) and bypass rough terrain by flying over it. This capability enabled the UAS Science Operations Team to characterize multiple surface types (Table 4) in a region of ∼1 km² during a short 12 sol mission.