Unmanned Aerial Geophysical Remote Sensing: A Survey

1. Introduction

Geoscience, which is also referred to as Earth science, includes a wide array of natural sciences focused on studying the Earth [1]. These fields include geophysics, geology, geodesy, geography, and others [2,3,4,5]. Within geoscience, geophysics specifically investigates the physical phenomena and characteristics of Earth and its surrounding spatial environment. It combines the fields of geology, physics, and mathematics in a single approach. It uses quantitative methods to analyze these phenomena. In essence, geophysics offers a means to describe various aspects of the subsurface environment, such as composition and structure thereby providing information on, e.g., objects, geological features, groundwater conditions, or pollution levels, through non-invasive techniques [6]. The term “non-invasive” indicates that these methods do not damage the Earth’s crust due to drilling or excavation. Geophysical methods include, among others, magnetic survey (magnetometry) [7], gravity survey (gravimetry) [8], electromagnetic (EM) survey [9,10], gamma-ray spectrometry (GRS)/radiometry [11], ground penetrating radar (GPR) [12], seismic tomography [13], electrical resistivity tomography (ERT) [14], and others.

Geophysical methods fall into two main categories: traditional close-range methods and modern remote sensing (RS)-based methods [15,16,17]. Close-range methods involve surveys conducted with equipment positioned as close to the Earth’s surface as possible, covering both terrestrial and aquatic environments. On the other hand, RS-based techniques utilize platforms and instruments situated at a considerable distance from the Earth’s surface [16,18]. These RS-based methods can be either spaceborne, employing satellites as platforms, or airborne, utilizing various manned and unmanned aerial platforms.

The increasing popularity of Unmanned Aerial Vehicles (UAVs) in the last decade, along with ongoing efforts by universities and industries to develop UAV-compatible devices and payloads, has profoundly impacted geophysics [19]. UAVs offer several advantages for geophysical surveys, including cost-effectiveness, ability to cover larger areas, lightweight design for easy transport, autonomous flying capability (which eliminates potential risks for onboard pilots in manned aircraft), and low-flying capability, all among its favorable features. These characteristics establish UAV-borne geophysical survey methods as a balanced compromise between traditional airborne and ground-based approaches, effectively combining the strengths of both methods simultaneously.

An extensive examination of literature concerning UAV-based geophysical methods, drawing from both scholarly research and corporate endeavors, reveals that these methods encompass traditional manned airborne techniques and introduce innovative additions such as seismic surveys [20,21]. This evolution has led to the development of UAV-based magnetometry [22], gravimetry [23], EM surveys [10], GPR [24], gamma-ray spectrometry/radiometry [25], as well as seismic surveys [26]. In addition to these basic geophysical methods, UAV photogrammetry [27] and LiDARgrammetry [28], while not traditionally considered geophysical techniques, hold applicability across various geoscientific domains [29,30].

The growing trend of UAV-based geophysical surveys in recent years and the increasing volume of research and papers published annually highlight a gap for a comprehensive review of UAV-based geophysical surveys. Existing research tends to focus on specific aspects of UAV-based geophysical surveys, as exemplified by the works of [22,24], which review UAV-based magnetometry or GPR, respectively. However, a comprehensive review paper, similar to a handbook, covering all feasible UAV-based geophysical survey methods remains needed. To address this gap, we aim to thoroughly review all methods feasibly developed for geophysical (or other geoscientific) surveys using UAVs.

To the best of our knowledge, no prior work has comprehensively compiled and reviewed all the methods mentioned. Thus, the novelty of our work is in collecting all standard UAV-based geophysical methods and providing new insights into methods that may initially not be recognized as geophysical techniques but can be applied effectively in this domain.

2. Research Methodology

A comprehensive search across different databases and sources has been done to address our research objectives. Our search comprises two methods: systematic querying in Scopus and manual search in Google Scholar, ResearchGate, etc. Although it is challenging to determine the definitive superior database for scientometrics, Scopus has been chosen as the foundation for our systematic querying. This decision is based on previous studies that have analyzed bibliometric bases, including [31,32].

Initially, relevant and similar terms were selected in an automated search strategy, and a query was made in Scopus using the filtering possibilities available there. The query for keyword search was as follows: TITLE-ABS-KEY (uav AND geophysics) OR TITLE-ABS-KEY (uav AND magnetic AND survey) OR TITLE-ABS-KEY (uav AND gravity AND survey) OR TITLE-ABS-KEY (uav AND ground AND penetrating AND radar) OR TITLE-ABS-KEY (uav AND gamma AND spectrometry) OR TITLE-ABS-KEY (uav AND gamma AND radiometry) OR TITLE-ABS-KEY (uav AND electromagnetic AND survey). Various word forms were examined and compared to ensure no significant articles were missed in our collection. For instance, in some cases, along with the commonly used term “UAV”, terms like “unmanned aerial vehicle” or “drone” were also used.

In addition to the automated query, a manual searching approach was also employed in the other resources above to address any deficits in studies that might not have been identified in the automated search. These deficits primarily pertain to UAV-borne RS methods such as photogrammetry and LiDARgrammetry, which are not typically considered geophysical methods and may not be easily identified through automated searches1. Relevant studies in manual searches were finalized by reading the papers’ text. Various word forms were examined to ensure that no significant documents were lost. The joint search returned a total of 587 studies, characterized by the word cloud shown in Figure 1. The word cloud offers a visual representation of the interdisciplinary nature of UAV-borne geophysical RS, bridging concepts from both the geoscience discipline and RS technology. The words presented in the cloud provide insight into the research landscape and the various themes explored within this field.

Moreover, an overview of the gathered publications is presented in Figure 2. There is an increase in publications (Figure 2a). This emerging topic is mainly developing through articles in journals and presentations at conferences, with no dedicated books yet available (Figure 2b). The impact of this topic extends across various disciplines, including computer science and environmental science (Figure 2c). Figure 2d illustrates that both developed and developing countries have begun to delve into the domain of UAV-borne geophysical RS, indicating its widespread relevance. Notably, universities and research centers, predominantly from developed countries, are leading in this field (Figure 2e). Furthermore, Figure 2f reveals that research is published in various journals, ranging from geosciences to RS technology, underscoring the interdisciplinary nature of this topic. Following the PRISMA flowchart [33] In Figure 3, the gathered documents were refined (containing three steps: identification, screening, and eligibility). Ultimately, 435 papers were considered eligible for review.

3. UAV-Borne Geophysical Survey Methods

Geophysical methods are classified into two primary categories based on sensor operation: those employing passive sensors and those using active sensors. For detailed information on passive and active sensors and their operation, refer to [34]. Additionally, a third category considers methods that integrate sensors and fuse data. Figure 4 illustrates this categorization. Note that “active/passive methods” indicate whether the method utilizes an active or passive sensor.

3.1. UAV-Borne Geophysical Survey: Passive Methods

3.1.1. Unmanned Aerial Magnetometry

The introduction of UAVs in geophysics, especially in airborne magnetometry, has driven notable global academic and technological progress [35]. UAV-borne magnetometry offers safety, cost-efficiency [22], and prolonged flight endurance, facilitating low-altitude flights and high-resolution magnetic data collection [36,37,38], compared to traditional ground-based or airborne surveys using a low-flying airplane. There is a noticeable trend towards the adoption of UAV-borne magnetometry, with universities and companies actively engaging in pioneering research [39].

Reliable surveying systems on lightweight UAV platforms address the limitations of traditional terrestrial and aerial magnetometry [22,40]. These systems efficiently collect high-quality magnetic data, enhancing spatial resolution at low altitudes [41]. They extend operations to previously inaccessible areas, reducing costs and offering flexibility [42,43]. UAV-based surveys bridge the gap between terrestrial and airborne methods, enhancing detectability [44]. Challenges include ensuring data quality comparable to manned aerial systems and developing lightweight magnetometers for small UAV platforms. Figure 5 depicts a UAV-based magnetometer system, with subsystems detailed in [22].

Different manufacturers offer a variety of magnetic sensors (magnetometers) compatible with UAVs [41]. Table 1 overviews the most widely available cutting-edge options, with confirmed quality and reliability.

Table 2 presents an examination of diverse cutting-edge UAV-borne magnetometry systems currently accessible worldwide, including both commercial options and those developed through scientific research and development (R&D).

In the domain of UAV magnetometry, one of the critical issues is EM interference. Mounting magnetic sensors on aerial vehicles poses challenges due to magnetic interference from propulsion and flight control systems. This interference originates from both the environment and onboard systems, diminishing sensor sensitivity and detection ranges [39,81,87,94]. Environmental factors include anything in the UAV’s surroundings that impacts the magnetic survey, while onboard factors pertain to various UAV components with magnetic characteristics. Due to these magnetic components, UAVs may compromise the accuracy of total magnetic field measurements [60]. Advances in magnetometer technology enable increased use in small to medium UAVs, but miniature UAVs face challenges due to their compact size and shorter distances between interference sources and sensors [22,43,49,81]. Addressing magnetic interference is a paramount challenge in aerial magnetometry development [39,49,94,95,96]. This section explores solutions to counter magnetic interference, considering both active and passive approaches.

• Active Solution: Post-compensation addresses UAV magnetic interference [22,39,96], utilizing calibration flights to gather high-altitude data and calculate compensation coefficients using a model (Equation (1)) [97]

$$\begin{gathered} H_T=C_{1}Cos\alpha +C_{2}Cos\beta +C_{3}Cos\gamma +H_G\left\{C_4{Cos}^2\alpha +C_{5}Cos\alpha Cos\beta +C_{6}Cos\alpha Cos\gamma +C_7{Cos}^2\beta + \\ {C}_{8}Cos\beta Cos\gamma +C_9{Cos}^2\gamma \right\}+H_G\left\{C_{10}Cos\alpha {\left(Cos\alpha \right)}^{\prime }+C_{11}Cos\alpha {\left(Cos\beta \right)}^{\prime }+C_{12}Cos\alpha {\left(Cos\gamma \right)}^{\prime }+ \\ {C}_{13}Cos\beta {\left(Cos\alpha \right)}^{\prime }+C_{14}Cos\beta {\left(Cos\beta \right)}^{\prime }+C_{15}Cos\beta {\left(Cos\gamma \right)}^{\prime }+C_{16}Cos\gamma {\left(Cos\alpha \right)}^{\prime }+C_{17}Cos\gamma {\left(Cos\beta \right)}^{\prime }+ \\ {C}_{18}Cos\gamma {\left(Cos\gamma \right)}^{\prime }\right\}={\sum }_{i=1}^{18}{C}_i{A}_i \end{gathered}$$

(1)

Where ${\mathrm{H}}_{\mathrm{T}}$ is the total interference field intensity, and ${\mathrm{H}}_{\mathrm{G}}$ represents the geomagnetic field. Cosα, cosβ, and cosγ are the directional cosines of the geomagnetic field vector with respect to the UAV’s axes. Cj (1 ≤ j ≤ 18) are compensation coefficients aimed at mitigating magnetic interference effects.

Compensation coefficients (Cj) are estimated using magnetometer data (α, β, γ) through the least squares method according to Equation (2). These coefficients, along with the model, mitigate aircraft interferences during magnetic surveys.

$$C={\left(A^{T}A\right)}^{-1}{A}^T{H}_{int}$$

(2)

Where ${\mathrm{H}}_{\mathrm{i}\mathrm{n}\mathrm{t}}$ and C are column vectors representing ${\mathrm{H}}_{\mathrm{T}}$ and Cj, respectively, and A is the design matrix [22,39].

To evaluate compensation, the “improvement ratio” and the “fourth-difference” metrics are used [22,98]. Calibration flights correlate UAV maneuvers with magnetic field changes for compensation. Flight of maneuver data should mirror UAV behavior, ideally collected at high altitudes. Post-compensation is not suitable for low-altitude flights, especially for multi-rotor UAVs due to instability. Yet, during high-altitude operations, post-compensation can be applied with magnetometers placed away from the UAV’s platform [22,39].

• Passive Solution: Post-compensation may not suffice for UAV magnetometry, necessitating an alternative approach by placing the magnetic sensor away from the aerial platform. Methods include suspending it beneath the UAV with a semi-rigid tether or affixing it to the UAV frame with a rigid bar. Various sensor attachment configurations are illustrated in Figure 6, accommodating different UAV types [41,99,100,101]. Placing the magnetometer away from the aerial platform can lead to sensor errors and fluctuations due to vibrations. Firmly affixing it to the airframe or using an extended boom may compromise flight stability, especially for fixed-wing UAVs [22,43,60,96,98]. Comparative studies suggest optimal sensor-platform distances of 3 to 5 meters to minimize interference [22,43,49,56,66,81,96,102]. For VTOL fixed-wing systems, mounting sensors at the winglets or nose-tip via a fixed-boom configuration is effective [80].

Figure 6. Different arrangements for mounting magnetometers on UAVs: (a-i to a-iv) fixed-boom design for rotary-wings, fixed-wings, helicopters, and airships, respectively; (b-i to b-iv) towed sensor design for the mentioned UAV types; (c-i to c-iv) towed bird design for the mentioned UAV types; and (d) fixed wing-tip design for fixed-wing UAV.

Figure 6. Different arrangements for mounting magnetometers on UAVs: (a-i to a-iv) fixed-boom design for rotary-wings, fixed-wings, helicopters, and airships, respectively; (b-i to b-iv) towed sensor design for the mentioned UAV types; (c-i to c-iv) towed bird design for the mentioned UAV types; and (d) fixed wing-tip design for fixed-wing UAV.

Table 3 presents a comprehensive compilation of studies conducted in the field of UAV-borne magnetometry, with a specific emphasis on applications.

3.1.2. Unmanned Aerial Gravimetry

Integrating UAV-borne gravity surveys with ground-based methods aids subsurface resource exploration, enhancing gravity field determination for diverse applications [124]. The UAV-based gravimetry system, depicted in Figure 7, includes a ground base station and an airborne gravimetry system. The ground base station comprises components like a ground control/command station (GCS), ground data transmission (GDT) station, and ground support equipment (GSE). The GCS facilitates real-time transmission of gravimeter data via satellite links. The airborne gravimetry system consists of an unmanned vehicle, UAV-compatible gravimeter, data-transferring computer, GNSS signal recorder, and uninterruptible power supply (UPS). The UAV, central to any unmanned aerial system (UAS), can take various forms, including airships, helicopters, fixed-wing, and multirotors, with no usage restrictions [23].

Two main operational modes of UAV-based gravimetry exist: continuous-flight and grasshopper (see Figure 8). Continuous-flight mode entails the drone collecting data while traversing an assigned area, ideal for large surveys such as those in the petrochemical industry. Challenges include distinguishing between platform and gravitational accelerations, often addressed using isolation platforms and gimbals. Grasshopper mode involves the UAV landing at specific points for data collection before taking off again. This mode, suitable for rotary-wing drones, collects data during stationary periods. Challenges in this mode include aligning the gravimeter’s axis and sensitivity to angular errors. Grasshopper mode typically requires more time for gravimetry due to flight and landing [125]. Further details and comparisons are available in the cited reference. The successful integration of UAV-compatible gravimeters onto unmanned platforms for continuous-flight and grasshopper mode operations requires several auxiliary systems. These include isolation (self-leveling) platforms, gimbals, and differential GPS capabilities [125,126].

Figure 7. UAV-based gravimetry system (the scheme was depicted based on [23,127,128]).

Figure 7. UAV-based gravimetry system (the scheme was depicted based on [23,127,128]).

Figure 8. Operational modes in UAV-borne gravimetry (R: flight rans, P: sampling points).

Figure 8. Operational modes in UAV-borne gravimetry (R: flight rans, P: sampling points).

Various corrections must be applied during gravity data collection, particularly in airborne gravimetry operations, including UAV-borne methods. These corrections include latitude correction, free-air (elevation) correction, Bouguer correction, topography (terrain) correction, Earth tides, Eötvös correction, and low-frequency translation correction, among others [23,129,130,131]. For brevity, readers seeking more detailed insights into these corrections are referred to the cited references.

This section provides a concise overview of gravity sensors compatible with UAVs. Miniature gravity sensors are designed for UAV integration, considering the payload capacity of UAVs. These sensors fall into two main categories: strapdown systems and Micro-Electro-Mechanical-Systems (MEMS) accelerometers. In strapdown configuration, a triad of accelerometers is directly affixed to the airborne vehicle. MEMS accelerometers offer a lightweight solution for UAV-based gravimetry. Triaxial MEMS devices have been successfully deployed in various UAV-based gravimetry operations. Table 4 lists advanced UAV-compatible gravimeters along with additional details. These are among the most widely available options, and their quality has been confirmed.

This section reviews the state-of-the-art UAV-borne gravimetry systems developed to date. Table 5 provides a comprehensive list of these advanced drone-deployed systems, along with relevant information for each.

Table 6 offers a comprehensive examination of the literature on UAV-borne gravimetry, emphasizing its applications. In contrast to magnetometry, the range of applications is currently restricted due to the emerging status of the field, with most studies focused on system R&D.

3.1.3. Unmanned Aerial Gamma-ray Spectrometry and Radiometry

GRS geophysical technologies are challenging to integrate with drones due to their bulkiness and traditional design [161,162]. New UAV-compatible spectrometer systems, designed within drone weight constraints, and improved data processing methods, e.g., [163,164], show promise for enhanced survey efficiency and accessibility. This shift requires innovative equipment design and data analysis methods to maximize potential.

UAV-borne GRS feasibility relies on precise gamma-ray radiation measurement in low-altitude environments with limited air attenuation [165,166]. UAVs use gamma-ray spectrometers to map soil properties, texture, and contamination [167], advancing from proof-of-concept to common practice [168,169]. Radiation mapper UAVs carry payloads suitable for efficiently mapping radionuclide activities [170]. UAV GRS combines ground-based and traditional airborne methods, using UAVs for gamma radiation investigations at lower altitudes, minimizing risks and costs, especially in smaller survey areas [169,171,172]. Radiometric surveys by UAVs are more common than spectrometric surveys due to lightweight UAV constraints, limiting traditional GRS methods’ effectiveness [171]. Ground-based GRS requires stops, while airborne GRS uses large-volume scintillation detectors for high-quality surveys [173]. Semiconductor detectors are expensive and less adopted compared to classical scintillation detectors, preferred for spectrometric measurements during movement [174,175,176]. Unmanned systems utilize heavy detection units with multiple crystals [171,174,175], impacting UAV flight time and range. Increasing UAV size is not cost-effective. Compact scintillation detectors offer economical and reliable data acquisition for geological mapping, ensuring high-quality outcomes in gamma-ray survey [171,175,177].

Let’s examine how a UAV GRS system operates. In a UAV GRS system (Figure 9), the calibrated gamma-ray spectrometer is mounted beneath the UAV to measure radionuclide concentrations [167]. An onboard rover logs GNSS data for georeferencing, with corrections from a ground-based receiver improving accuracy. The spectrometer captures radiation spectra at each position, synchronized with location data and processed onboard [168,178]. Spectrometer data is transmitted to the GCS via RF systems like TETRA network, UMTS, and others [178,179,180,181]. Post-flight processing includes spectrum analysis for radiation heatmap generation [180]. The GCS software integrates radiation levels, coordinates, and altitude for visualization [181]. Measured radionuclide concentrations obtained from the spectrometer are used in application models, correlating them with soil properties or contaminants. Proper sensor calibration is crucial for accurate results, often achieved through laboratory analyses known as ground truthing [167,180]. During field operations, soil samples are collected. These samples undergo air-drying, milling, and sieving for clay and sand content analysis [169,182]. Lab testing includes radionuclide content, grain-size distribution, and clay fraction analysis, essential for application model development. Models can be developed based on as few as 14-20 soil samples, generating soil maps for various applications [167]. Expert analysis can further enrich the final map.

In this section, spectral data processing is explored, specifically the methods for calculating radionuclide concentrations. Fundamental processing steps involve consolidating data from different detectors and filtering out points during UAV stationarity or survey line transitions [183]. Spectrum analysis techniques, such as the Windows method and Full Spectrum Analysis (FSA) procedure, are used to derive radionuclide concentrations from recorded spectra [161,170,184]. The Windows method extracts ⁴⁰K, ²³⁸U, and ²³²Th concentrations from specific energy windows [161,169]. In contrast, the FSA method, introduced in [185,186], utilizes nearly all spectral data, enhancing precision [161,184]. FSA fits “standard spectra” to acquired spectra, reflecting the detector’s response to pure radionuclide sources (⁴⁰K, ²³⁸U, or ²³²Th) [169,184]. Monte Carlo simulations and modeling generate standard spectra, a standardized practice detailed in references such as [163,169,187]. FSA reduces uncertainties by half compared to the Windows method, improving data quality [184]. It enables the use of smaller detectors while maintaining quality and provides richer count and spectrum structure [161].

In UAV-borne GRS, elevation impacts signal reception and footprint size. Higher altitudes reduce detected signal quantity due to increased air attenuation [163]. Elevating the spectrometer broadens the footprint, known as the “table lamp effect,” expanding the effective measured area [169,182]. Gamma-ray spectra conversion to radionuclide concentrations requires altitude correction for signal attenuation and detector field-of-view variations. Traditional corrections for airborne ranges may not suit UAV altitudes. Previous studies proposed corrections, but wide adoption was lacking. Some surveys (e.g., [175,188,189,190]) used International Atomic Energy Agency (IAEA) corrections even at lower altitudes. In [169] the corrections was refined specifically for UAV GRS operational ranges (0-40 m) using experimental and computational methods, as detailed in the cited study. For more detailed insights of the commonly used footprint and height corrections applied in the domain of UAV-borne spectrometry and radiometery and also some innovative methods in this domain type refer to [169,180,191,192].

To provide an overview of UAV-compatible radiation sensors, it is important to note that several ready-to-use gamma-ray spectrometers have been tested and confirmed to be attachable to drones. These spectrometers are listed in Table 7. In Table 8, a thorough list of state-of-the-art UAV-borne GRS systems is provided, offering detailed technical specifications sourced from published papers.

Table 9 presents an overview of the applications of UAV-borne GRS derived from existing research in the field.

3.1.4. Unmanned Aerial Imaging Geophysics

A UAV photogrammetry system comprises both aerial and ground sections, with the aerial section featuring the UAV and various imaging sensors such as visible-light (RGB), multispectral (MS), hyperspectral (HS), or thermal-infrared (TIR) cameras [209,210]. The system operates through fieldwork, including ground operations and aerial surveys, and office tasks involving survey planning and data processing. Key steps involve acquiring overlapped images, identifying key points, and performing Structure-from-Motion (SfM) and Multi-View Stereo (MVS) [211] to generate dense point clouds, orthomosaics, and different other geospatial products. While this overview doesn’t delve deeply into UAV photogrammetry, interested readers can refer to relevant references (e.g., [212]) for more details. UAV photogrammetry extends beyond civil applications [213], offering valuable geometrical, structural, spatial, and spectral data about the natural Earth. Consequently, it’s a versatile UAV RS method applicable to various geoscientific endeavors.

This section provides an overview of UAV-compatible optical and TIR imaging sensors. In the UAV visible photogrammetry category, two types of visible cameras are commonly used on UAV platforms for imaging and photogrammetry [214]. These include UAV-integrated cameras, exemplified by the 20-MP 4/3-inch image sensor (e.g., used in DJI Mavic 3 drone) and the 20-MP 1-inch image sensor (e.g., used in AUTEL’s EVO II Pro V3 drone), and Digital Single Lens Reflex (DSLR) cameras like the Sony A7RIII. In UAV MS photogrammetry and RS, state-of-the-art sensors include the Parrot Sequoia+, MicaSense Altum-PT, MicaSense RedEdge-MX/P, Sentera 6X, and DJI P4 Multi [214,215]. State-of-the-art TIR sensors include the WIRIS Pro/Pro Sc, Zenmuse XT 2 (dual-light TIR imager), Uncooled FPA 6404, FLIR SC305, and TELEDYNE FLIR VUE Pro. In the realm of UAV HS photogrammetry, cutting-edge sensors include OCI™-UAV-1000/2000, OCI™-F series, and GoldenEye™ (by BaySpec); CHAI S/V-640, S 185 FIREFLEYE SE, S 485 FIREFLEYE XL, and Q 285 FIREFLEYE QE (by Cubert); Nano/Micro-HyperSpec, VNIR-1024, Mjolnir V-1240, and SWIR-384 (by Headwall Photonics HySpex); Rikola, vis-NIR microHSI, Alpha-vis microHSI, SWIR 640 microHSI, and Alpha-SWIR microHSI by MosaicMill NovaSol; MV1-D2048x1088-HS05-96-G2 (by PhotonFocus); Hyperea 660 C1, Pika L, Pika XC2, Pika NIR, and Pika NUV (by Quest Innovations Resonon); VIS-VNIR Snapshot by SENOP; SPECIM FX10/17 (by SPECIM); SOC710-GX (by Surface Optics); and MQ022HG-IM-LS100-NIR/ IM-LS150-VISNIR (by XIMEA) [209]. For the sake of brevity, no further details about the sensors or their specifications are provided here. Readers are referred to the cited references for more information.

In UAV photogrammetry, spatial resolution—denoted by ground sampling distance (GSD)—along with spectral resolution (referring primarily to the number of bands the sensor can capture) and radiometric resolution, are key factors [216,217]. Common corrections in UAV photogrammetry include radiometric and geometric calibration [218,219]. Radiometric calibration aims to derive absolute reflectance measurements from the digital number (DN) [218]. Geometric calibration encompasses band-to-band registration (primarily for multi-lens sensors) and true orthorectification, which allows objects in the orthomosaic to be viewed from a top-down perspective [219]. Additionally, point cloud filtering is another type of correction applied not to images but to point clouds, eliminating non-ground points to generate a bare-land point cloud and digital terrain model (DTM) from the initial point cloud [220].

The geoscientific applications of UAV photogrammetry can be listed as follows:

• Digital Terrain Modeling for Geomorphological Applications: The initial application of UAV photogrammetry in geoscience focuses on ultra-high-resolution digital terrain modeling for geomorphological purposes. A crucial step in this process involves filtering the original point cloud to isolate the natural terrain points [220]. Subsequently, topographical maps are generated by combining the orthomosaic with contour lines. These maps facilitate further geomorphological analysis and applications.
• Landslide Mapping and Monitoring: Table 10 presents a comprehensive review of efforts in landslide mapping and monitoring using UAV photogrammetry.
• Land Subsidence and Ground Failure Mapping:Table 11 provides a thorough list of research conducted in the realm of land subsidence and ground fissure mapping.
• Geothermal Exploration: The application of UAV photogrammetry in geothermal exploration is explored here, with a collection of them provided in Table 12.
• Soil Moisture Mapping: Research conducted in the realm of soil moisture mapping using UAV photogrammetry has been reviewed in Table 13.

• Mineralogy, Mining, and Soil Mapping: Table 14 provides a comprehensive overview of UAV-based imaging and photogrammetry applications in mineralogy, mining, and soil mapping. It is noteworthy that research focusing on mining was gathered and reviewed building on [272].
• Volcanoic Research: Table 15 offeres a collection of researches in the realm of volcanoic mapping using UAV photogrammetry.

3.2. UAV-Borne Geophysical Survey: Active Methods

3.2.1. Unmanned Aerial EM Survey

The EM method, utilizing induced currents to detect conductive underground structures, aids in subsurface geology understanding. It’s divided into time-domain EM (TDEM) and frequency-domain EM (FDEM) survey methods. Aerial EM (AEM) schematics depict induced and measured magnetic fields (see Figure 10). Measurements involve primary EM fields from the transmitter and secondary EM fields correlated with geological features [306,307,308,309]. AEM methods are categorized by excitation modes (AFEM and ATEM), platform types (fixed-wing TEM, fixed-wing FEM, helicopter-borne TDEM or simply HTEM, and helicopter-borne FDEM or simply HFEM), and EM field transmission types (active, passive, and semi-passive/airborne systems) [306,310,311,312,313,314]. Different ATEM and AFEM systems have been developed based on various types of manned aerial platforms [311,313,315,316,317,318,319,320,321,322,323,324,325,326,327]. Despite this variety, UAV-borne EM survey systems are a relatively new topic compared to traditional methods.

AEM has found various geoscientific applications including finding Quick Clay [328,329,330], identifying hazardous substances [331], mapping the fresh-saltwater interface [332], freshwater potential investigation [333], deep groundwater mapping in Antarctica [334,335,336], hydrologic mapping and environmental assessments [337], detection and mapping impermeable aquifer boundaries [310], groundwater and soil investigations [338], exploring the relationship between groundwater and surface water [339], 3D geological modeling of complex buried valleys [340], observations of a collapse-prone volcano [341], mineral exploration [327], gold exploration [342], mapping sub-Phanerozoic basement features [343], UXO detection [63], magnetite ore tonnage estimation [344], and application in the fields of uranium exploration [345,346].

Developing EM-sounding systems for lightweight UAVs faces challenges due to weight and bulkiness issues with conventional techniques, such as helicopter-based systems. Aerial exploration presents difficulties like precise loop positioning and adapting methods to UAVs, requiring bulky generator setups [347]. Integrating EM sensors with UAVs introduces stability, interference, and control challenges, addressed with solutions like sensor suspension, tail fins, noise avoidance, and real-time monitoring [348,349]. These innovations aim to enhance data quality and address engineering challenges in UAV-based EM surveys. The scheme of Figure 10 is also valid for the UAV EM method.

There are two configurations in UAV-borne EM surveys: single-drone and dual-drone configurations. To illustrate these configurations, we focus on the Louhi geophysical EM survey system. Developed by Radai Ltd. under the NEXT project, this system facilitates practical drone-based operations for FDEM methods [309]. It features a single drone equipped with a transmitter, offering flexibility in surveying approaches (see Figure 11). Additionally, a two-drone configuration is utilized to maintain consistent separation distance between the receiver and transmitter drones, allowing for deeper exploration and rapid deployment in challenging terrains [350]. The fixed loop transmitter system enhances source moment and signal-to-noise ratio (SNR), crucial for subsurface exploration. Accurate position and orientation measurements enable conversion to global 3D coordinates, although variations in loop spacing and orientation can introduce noise. Synchronization between the receiver and transmitter units is achieved using GNSS time, allowing for precise data recording and analysis. Payload constraints and synchronization between dual-drone autopilots pose challenges, necessitating ongoing research and development efforts [309].

In the realm of UAV-based EM RS, the concept of the “Semi-UAV-borne EM (SUEM) System” is emerging as a promising innovation. The Semi-Airborne EM (SAEM) setup presents a promising avenue for UAV-based EM surveys [351]. In this configuration, the transmitter remains stationary on the ground while the receiver is deployed on a UAV (similar to Figure 11b), offering unique advantages over conventional AEM [352,353]. However, the payload limitations of current UAVs restrict the maximum speed of the SAEM system. Addressing challenges related to turbulence-induced shaking and motion-related noise in the dataset necessitates innovative solutions [311]. Advancements in drone technology have enabled the adaptation of SAEM systems to UAVs, reducing application and maintenance costs and facilitating multi-component surveys. Notably, MGT’s SAEM system, developed in collaboration with the geophysical industry and research organizations in Germany, exemplifies the successful integration of an EM system onto an unmanned multicopter, enabling data collection with enhanced depth penetration and accuracy. These developments mark significant progress in the evolution of SUEM systems, holding considerable potential for future UAV EM RS applications [311,352,353,354].

Very-low Frequency (VLF) EM methods are frequently mentioned in the literature, especially in the context of UAV EM applications. UAV integration with the VLF method evolved since Kipfinger’s lightweight system in 1996-97 [353,355]. Recent developments led to VLF systems on rotary-wing UAVs with a 12 kg payload capacity [356]. VLF induces primary horizontal magnetic fields and captures secondary fields in conductive subsurface formations with a receiver equipped with two coils [357,358,359]. Equation (3) calculates the vertical magnetic field component using transfer functions or “tippers” [360,361].

$$H_z\left(\omega \right)=A\left(\omega \right).H_x\left(\omega \right)+B\left(\omega \right).H_y\left(\omega \right)$$

(3)

Modern VLF instruments capture time series data for all three magnetic field components, enabling analysis in both time and frequency domains. Automated spectral analysis, like the method proposed in [362], identifies VLF transmitters. These advances highlight the potential of UAV-based VLF EM systems for large-scale geophysical surveys [356,363].

Let’s explore the data collection modes in UAV-borne EM surveys. UAV-borne EM surveys use fixed-point and continuous data collection modes [364]. In fixed-point mode, the UAV hovers at each measurement point, ensuring high-quality data but limiting data points due to energy-intensive hovering. Continuous mode collects data seamlessly during low-level flight, offering substantial data volumes similar to grasshopper mode [125]. Post-processing ensures data quality comparable to fixed-point mode [365].

This section discusses three key aspects of UAV-based EM surveys: compatible EM sensors for UAVs, UAV platforms used in EM surveys, and the development of UAV-based EM systems.

• EM Sensors: Various EM instruments have been utilized in UAV-borne surveys, including GEM-3D, MPV, MPV-II, Pedemis, High-Frequency EMI, Dualem-1S, EM38, Profiler 400-EMP, CMD MiniExplorer, US Army’s drone-mounted EM induction sensor, CAS & Jilin University single-component sensor and others [311,349,366,367]. Among these, the GEM-2UAV is prominent, weighing 3 kg and operating at ten frequencies (25 Hz to 96 kHz). It requires a GNSS antenna, WinGEM software, and consumes 20 W during surveys, offering configurable operational modes and data logging initiated through a control unit [348,349,368].
• UAV Platforms: UAV platforms for carrying EM instruments are categorized into four types: multi-rotor [347,349,352], fixed-wing [73,369], helicopters [356], and airships/balloons [370]. The choice of platform depends on factors like survey objectives and the size of the EM instrument.
• UAV-borne EM Survey Systems: Building on the information provided in the preceding two bullets, Table 16 offers a thorough overview of the UAV-borne EM systems that have been developed.

Table 16. Review of state-of-the-art UAV-borne EM survey systems.

Sys.	Platform Type	UAV Name/Model	EM Instruments	References
1	Multi-rotor	MTOW octocopter	Miniaturized induction coil triple	[352,353]
2	Multi-rotor	X825 octocopter	Metronix SHFT-02e induction coil triple	[351]
3	Multi-rotor	SibGIS hexacopter	A measuring system with an inductive sensor	[201,347,371,372]
4	Multi-rotor	SibGIS hexacopter	A grounded transmitter line spanning 2.2 km serves as the origin of the current pulses, coupled with an airborne PDI-50 receiver loop on the UAV.	[373]
5	Helicopter	Aeroscout Scout B1-100	The Super High-Frequency Induction Coil Triple sensor, in conjunction with the ADU07 data logging module, both developed by MGT.	[356]
6	Multi-rotor	DJI Matrice 600 Pro	GEM-2UAV CSEM sensor	[349]
7	Fixed-wing	VTOL Mother-Goose	Louhi portable EM transmitter and three-component receiver	[309]
8	Unmanned VTOL airship	Quaddirigible (filled with helium)	A VLF-type EM survey system	[370]
9	Multi-rotor	ZION CH940 and LAB6106 multicopters	GEM-2	[348]
10	Multi-rotor	Hexacopter	A coil wound with enameled copper wire, comprising 25 turns and a diameter of 25 cm, designed for the generation of a magnetic field.	[374]
11	Hexacopter and fixed-wing	SGU’s fixed-wing VLF	Three orthogonally mounted induction coil sensors and a data acquisition system with up to 1 MHz continuous data sampling of the EM components.	[369]
12	Multi-rotor	Hexacopter	The drone-borne TEM system utilized a central loop device.	[365]
13	Not specified	Not specified	D-GREATEM system	[375]
14	Fixed-wing	Silver Fox UAV	A sensing coil towed behind the UAV	[73]
15	Multi-rotor	Hexacopter	A measurement setup using an inductive sensor (receiving loop) is tethered by a UAV, while a galvanically grounded power transmitter is positioned on the ground and linked to a pulse generator.	[376]
16	Unmanned helicopter	Tianxiang V-750	A SAEM system, designed by the CAS, encompasses a single-component sensor, transmitter, and receiver, all equipped with vibration isolation.	[311]
17	Multi-rotor	Hexacopter	A SAEM system, engineered by Jilin University, consists of a robust ground-based transmitter generating high-power signals and a single-component sensor.	[311]
18	Multi-rotor	DJI Matrice 600	The Geophex multi-coil, CMD MiniExplorer EM instruments, and GEM-2	[366]
19	Multi-rotor	DJI Wind4 quadcopter	Geophex GEM-2UAV	[377]
20	Fixed-wing	A long-range drone	UAV-borne gravity and EM sensors	[158,160]
21	Octocopter	System name: MGT-GEO Radio EM	The sensor system weighs 6.5 grams, operates in a frequency range of 1-524 kHz, encompasses channels for Hx, Hy, and Hz, boasts a sample rate of up to 524 kHz, achieves synchronization through GPS, and utilizes compact flash disk storage media.	[366]
22	Rotary-wing	DroneSAM	A low-frequency hybrid geophysical system integrating a ground active source transmitter system with a drone for slow-flying, low-level data acquisition of TEM and magnetometric resistivity data.	[378]
23	Rotary-wing	Multi-rotor (DronEM)	Drone for EM fields Measurements (DronEM) is outfitted with a Selective Electric Triaxial Probe and is capable of scanning the EM spectrum ranging from 10 MHz to 3 GHz at altitudes up to 200 m.	[379]
24	Rotary-wing	Not specified	UAV VLF EM System: two VLF UAV sensor coils with cables accompanied by other instruments.	[380]
25	Rotary-wing	Octo/helicopter	3-component EM sensor (induction coil DEEP) and fluxgate magnetometer.	[381]
26	Rotary-wing	Hexacopter	Time-domain EM system suspended beneath the UAV	[10]

Table 16. Review of state-of-the-art UAV-borne EM survey systems.

Sys.	Platform Type	UAV Name/Model	EM Instruments	References
1	Multi-rotor	MTOW octocopter	Miniaturized induction coil triple	[352,353]
2	Multi-rotor	X825 octocopter	Metronix SHFT-02e induction coil triple	[351]
3	Multi-rotor	SibGIS hexacopter	A measuring system with an inductive sensor	[201,347,371,372]
4	Multi-rotor	SibGIS hexacopter	A grounded transmitter line spanning 2.2 km serves as the origin of the current pulses, coupled with an airborne PDI-50 receiver loop on the UAV.	[373]
5	Helicopter	Aeroscout Scout B1-100	The Super High-Frequency Induction Coil Triple sensor, in conjunction with the ADU07 data logging module, both developed by MGT.	[356]
6	Multi-rotor	DJI Matrice 600 Pro	GEM-2UAV CSEM sensor	[349]
7	Fixed-wing	VTOL Mother-Goose	Louhi portable EM transmitter and three-component receiver	[309]
8	Unmanned VTOL airship	Quaddirigible (filled with helium)	A VLF-type EM survey system	[370]
9	Multi-rotor	ZION CH940 and LAB6106 multicopters	GEM-2	[348]
10	Multi-rotor	Hexacopter	A coil wound with enameled copper wire, comprising 25 turns and a diameter of 25 cm, designed for the generation of a magnetic field.	[374]
11	Hexacopter and fixed-wing	SGU’s fixed-wing VLF	Three orthogonally mounted induction coil sensors and a data acquisition system with up to 1 MHz continuous data sampling of the EM components.	[369]
12	Multi-rotor	Hexacopter	The drone-borne TEM system utilized a central loop device.	[365]
13	Not specified	Not specified	D-GREATEM system	[375]
14	Fixed-wing	Silver Fox UAV	A sensing coil towed behind the UAV	[73]
15	Multi-rotor	Hexacopter	A measurement setup using an inductive sensor (receiving loop) is tethered by a UAV, while a galvanically grounded power transmitter is positioned on the ground and linked to a pulse generator.	[376]
16	Unmanned helicopter	Tianxiang V-750	A SAEM system, designed by the CAS, encompasses a single-component sensor, transmitter, and receiver, all equipped with vibration isolation.	[311]
17	Multi-rotor	Hexacopter	A SAEM system, engineered by Jilin University, consists of a robust ground-based transmitter generating high-power signals and a single-component sensor.	[311]
18	Multi-rotor	DJI Matrice 600	The Geophex multi-coil, CMD MiniExplorer EM instruments, and GEM-2	[366]
19	Multi-rotor	DJI Wind4 quadcopter	Geophex GEM-2UAV	[377]
20	Fixed-wing	A long-range drone	UAV-borne gravity and EM sensors	[158,160]
21	Octocopter	System name: MGT-GEO Radio EM	The sensor system weighs 6.5 grams, operates in a frequency range of 1-524 kHz, encompasses channels for Hx, Hy, and Hz, boasts a sample rate of up to 524 kHz, achieves synchronization through GPS, and utilizes compact flash disk storage media.	[366]
22	Rotary-wing	DroneSAM	A low-frequency hybrid geophysical system integrating a ground active source transmitter system with a drone for slow-flying, low-level data acquisition of TEM and magnetometric resistivity data.	[378]
23	Rotary-wing	Multi-rotor (DronEM)	Drone for EM fields Measurements (DronEM) is outfitted with a Selective Electric Triaxial Probe and is capable of scanning the EM spectrum ranging from 10 MHz to 3 GHz at altitudes up to 200 m.	[379]
24	Rotary-wing	Not specified	UAV VLF EM System: two VLF UAV sensor coils with cables accompanied by other instruments.	[380]
25	Rotary-wing	Octo/helicopter	3-component EM sensor (induction coil DEEP) and fluxgate magnetometer.	[381]
26	Rotary-wing	Hexacopter	Time-domain EM system suspended beneath the UAV	[10]

In this part an overview is provided on the data integration, processing and inversion principles, and methods applied. The UAV-based EM data processing, following methodologies by [366,382], involves several key steps (Figure 12). Initially, raw datasets from various sources, including drone, EM instrument data, and additional sources (e.g., LiDAR, if available) are integrated, synchronized, and preprocessed for interpretation [383]. Noise filtering enhances data quality by removing high-frequency noise using a low-pass filter. Data segmentation categorizes raw EM sensor data into non-informative, grid/profile, and vertical sounding segments, facilitating anomaly analysis [366]. Inversion iteratively updates resistivity models by comparing field data with synthetic data, employing methods like Layered Constrained Inversion and Spatially Constrained Inversion [384,385,386]. Validation involves addressing noise sources and correlating EM data with geological maps and boreholes to discern soil characteristics and validate results. Additional geophysical data and hand drilling may be utilized for validation when alternative datasets are unavailable. A comprehensive review of UAV-borne EM survey geoscientific applications is summarized in Table 17.

3.2.2. Unmanned Aerial GPR

Over the past decade, UAV-based radar imaging has seen significant advancement, with various radar technologies, unmanned platforms, and payload configurations showcased [390,391,392,393,394,395]. Early research by [396,397] laid the foundation, leading to tests with high-frequency radars at P, X, and C bands [398,399,400], despite limited penetration capabilities. In [401] multi-frequency GPR for rotary-wing UAVs was explored, sparking a surge in contactless GPR research with UAV platforms. The convergence of UAVs and radar technology offers all-weather data recording and buried object detection, driving interest across scientific and industrial sectors [402,403,404,405]. Applications range from landmine detection to environmental monitoring, highlighting UAV-based GPR’s promising future in geophysical surveys and underground exploration [406,407,408]. UAV-based GPR systems are categorized as prototypes explicitly designed for UAVs or conventional GPR devices adapted for UAVs, with systems further classified based on whether antennas contact the ground, leading to ground-coupled and air-launched GPR systems [409,410,411].

Two approaches exist for assembling payloads in the UAV-GPR systems: independent and integrated designs (see Figure 13). In the independent approach, the UAV and payload are separate subsystems, while in the integrated method, they are developed collaboratively. The independent setup allows for compatibility with various UAV platforms but requires a separate interface for integration. Conversely, the integrated architecture simplifies subsystem synchronization and enables high-accuracy geo-referencing for navigation without redundant sensors. This approach is preferred for UAV systems with advanced sensors or specialized flight modes [24]. Visual examples include [412], for independent architecture, and [392], for integrated architecture.

In this part, the observation modes in UAV-borne GPR are reviewed. The scanning strategies employed in UAV-borne GPR are of significant importance. These systems are categorized into three main types based on observation modes and antenna orientation relative to the Earth’s surface (see Figure 14) [392,413].

Figure 13. Payload assembly architectures in UAV-GPR systems: (a) Independent-payload and (b) Integrated-payload architectures. The principles were borrowed from [24,414], with the flowcharts being reconfigured.

Figure 13. Payload assembly architectures in UAV-GPR systems: (a) Independent-payload and (b) Integrated-payload architectures. The principles were borrowed from [24,414], with the flowcharts being reconfigured.

Figure 14. Observation modes in UAV-borne GPR survey (reconfigured based on [24]).

Figure 14. Observation modes in UAV-borne GPR survey (reconfigured based on [24]).

Down-looking GPR (DLGPR) configurations position antennas perpendicular to the surface, offering advantages in detecting deeper targets while potentially obscuring shallow ones due to reflections at the air-soil interface [24,409,415,416,417]. Forward-looking GPR (FLGPR) orientations minimize clutter by reducing reflections from the air-ground interface through oblique antenna incidence [24,416,418,419]. Side-looking GPR (SLGPR) configurations utilize tilted antennas to mitigate specular reflection from the air-soil interface, often moving laterally or following circular paths for data capture [392,420,421,422,423].

The scanning architecture choice depends on the target depth and scenario. FLGPR or SLGPR configurations suit shallow targets, minimizing clutter and enhancing detection. DLGPR systems excel for deeper targets, despite increased clutter, offering extended dynamic range. DLGPR and FLGPR imaging capabilities are compared, with FLGPR optimizing transverse magnetic wave penetration, while DLGPR provides enhanced resolution but increased clutter [413,415].

UAV-based GPR methods are divided into fully-airborne and semi-airborne setups. In fully airborne configurations, both antennas are on one UAV or split between two UAVs for transmission and reception. Semi-airborne setups involve a ground-based vehicle with the transmission antenna and a UAV with the reception antenna, operating in down-looking mode (see Figure 15).

Data processing in UAV-borne GPR, inspired by air-launched systems (e.g., [416]), involves two categories: Standard methods and Advanced algorithms (see Figure 16). Standard methods filter noise and correct motion errors, while advanced algorithms aim for high-resolution imaging [395].

In standard processing, data undergoes positioning management and preprocessing, including background removal [394], dewow [425], and clutter elimination. Techniques such as time-gating [427], average subtraction [428,429], and SVD-based filtering [430] are employed for clutter removal. Ground profile retrieval and height correction are also essential for data accuracy [426].

Focusing methods aim to enhance image resolution and interpretability by transforming diffraction hyperbolas into distinct bright spots [431,432]. Figure 17 depicts a UAV-mounted DL-GPR surveying a region of interest, capturing backscattered radar signals along its flight trajectory (Γ) across the angular frequency range $\mathsf{\Omega }=[{\omega }_{min},{\omega }_{max}]$. Each measured point (m) in the volumetric subsurface domain (D) is defined by the position vector ${\overrightarrow{r}}_m=x_m\widehat{i}+y_m\widehat{j}+z_m\widehat{k}$, representing various buried targets. The radar imaging process employs a simplified linear scattering model [433]. Equation (4) defines the scattered field at each point, with ${\mathrm{E}}_{\mathrm{s}}$ representing the scattered field, ${\mathrm{E}}_{\mathrm{i}}$ denoting the incident field within domain D, χ(r) characterizing the unknown contrast function at any point $\mathrm{r}$ in D, G corresponding to Green’s function, and k is the propagation constant.

$$E_s\left(r_m,\omega \right)=k^2{\iiint }_{D}G\left(r_m,r,\omega \right)E_i(r,\omega )\chi (r)dr$$

(4)

Various algorithms were proposed to address the UAV-GPR imaging challenge, discussed as follows:

• Migration Techniques: Migration techniques like Kirchhoff’s wave-equation and phase-shift migration (PSM) algorithms, commonly used in GPR, are applied in UAV-based GPR data processing for efficient analysis [435,436,437,438,439]. However, PSM requires data interpolation to a regular grid, which may pose challenges in irregular survey trajectories. Newer approaches like Piecewise SAR (P-SAR) address these limitations by considering the reflection and transmission coefficients of EM waves through diverse subsurface layers [440].
• Back-projection (BP) or Delayand-and-Sum (DAS) Method: UAV-based GPR often employs beam-forming or SAR-like Back-projection (BP) or Delay-and-Sum (DAS) methods due to non-rectilinear measurement trajectories [414]. They integrate radar echoes from the flight path to generate a reflectivity map using Equation (5).

$$\chi \left(r\right)={\int }_{\mathsf{\Gamma }}{\int }_{\mathsf{\Omega }}{E}_s\left(r_m,\omega \right)\left(G\left(r_m,r,\omega \right)E_i(r,\omega )\right)\mathrm{*}d{r}_{m}d\omega$$

(5)

Where * represents the conjugation operator.

DAS involves summation of measurements at focal points across the target area [441]. Reflectivity $\mathsf{\rho }\left(\mathrm{r}\right)$ is determined using scattered field data from N acquisition points and M discrete frequencies (Equation (6)), considering phase shifts from wave propagation (Equation (7)).

$$\mathsf{\rho }\left(\mathrm{r}\right)={\sum }_{n=1}^N{\sum }_{m=1}^M{E}_s(r_n,f_m)e^{+j2({\varphi }_0+{\varphi }_1)}$$

(6)

$${\varphi }_0+{\varphi }_1=k_{0,m}{‖r^{\prime }-r_n‖}_2$$

(7)

Where $r_n$ represents the position where the n-th measurement was acquired, $f_m$ stands for the m-th discrete frequency (it is related to angular frequency), and ${\varphi }_0$ and ${\varphi }_1$ correspond to the phase shifts resulting from wave propagation. $k_{0,m}$ represents the free-space wavenumber for the m-th discrete frequency, while $r_{i,n}$ signifies the refraction point on the air-ground interface. The refraction point position ($r_{i,n}$) is determine using Snell’s law [441].

DAS, suitable for irregular flight trajectories, are comparable to PSM techniques in performance for UAV-based GPR processing. They find application in various UAV-based GPR studies [392,394,397,406,414,428,429,442].

• Microwave Tomography (MT): MT focusing algorithms, using inverse filtering techniques, aim to solve the EM inverse scattering problem [432,443]. Unlike SAR-like methods, MT directly inverts the linear integral equation [24,395], addressing the imaging problem through a solution to the linear inverse problem (Equation (8)).

$$E_s=L_{\chi }$$

(8)

$\mathrm{L}:{\mathcal{L}}^2\left(\mathrm{D}\right)\to {\mathcal{L}}^2(\mathsf{\Gamma }\times \mathsf{\Omega })$ maps unknown to data space, both being square-integrable function spaces. Regularization, often through truncated SVD, stabilizes the ill-posed inverse problem [444]. MT’s resilience to noise surpasses SAR-like methods, promoting its widespread adoption in UAV-based GPR imaging across various data collection scenarios and environments [24,395,425].

• Full-waveform Inversion (FWI) or Integral Equation (IE)-based Methods: FWI or IE GPR processing links radar backscattered fields with soil EM properties [404]. It estimates soil conductivity and permittivity by minimizing a cost function comparing the model and observed data [24]. In UAV-based GPR, FWI estimates soil permittivity, facilitating SWC mapping [395,406,434].

The mentioned methods reconstruct 2D images or 3D models of the subsurface and potential buried targets. These results can be refined using automatic target detection and recognition techniques, from traditional CFAR detectors [445] to advanced DL-based methods [446,447].

In this part the state-of-the-art UAV-compatible GPR antennas are reviewed. UAV-borne GPR systems rely on antennas to convert guided waves [424,448]. Modern UAV-compatible antennas include Vivaldi-like antennas, Archimedean spiral antennas, and helix antennas, as reported in [24,393,413,414,422], detailed in Table 18. Antennas in UAV-based GPR systems prioritize weight, dimensions, and radiation performance, often favoring horn-like or planar designs. These antennas are commonly featured in UAV-based GPR systems developed by [402,406,413,422,426,449,450]. Table 19 provides a comprehensive inspection of cutting-edge UAV-based radar and GPR systems, extracting detailed technical specifications from published works.

In this part, applications of the UAV-borne GPR survey are reviewed. Table 20 provides an outline of the key application domains in which the previously discussed UAV-borne GPR prototypes find utility.

3.2.3. Unmanned Aerial LiDARgrammetry

A UAV-based LiDAR system consists of an aerial section with a UAV platform and sensors, including compact LiDAR, GNSS receiver, and IMU, while the ground section comprises a control terminal, flight control system, and ground control points [469,470,471]. LiDAR emits laser pulses to the Earth’s surface, measuring distances based on pulse travel time. Integration of timing, LiDAR orientation, and location data determines point cloud accuracy [469]. For further details, interested readers are referred to [472,473].

When discussing UAV-compatible LiDAR sensors, it is important to distinguish between their mechanisms: Mechanical Scanning-based Sensors, Solid-state Technique-based Sensors, and Solid-state Hybrid Sensors [214,470,474,475]. The cutting-edge survey-grade UAV LiDAR sensors include Riegl VUX-1 UAV, Riegl mini VUX-1 UAV, Riegl mini VUX-1DL, Riegl VUX-240, Velodyne Puck LITE VLP-16, Quanergy M8, and Livox Mid-40. For more information, readers are referred to the relevant references (e.g., [470]).

To provide an overview of data processing and corrections in UAV-borne LiDARgrammetry, a comprehensive workflow comprising three main parts was proposed in [470]: point cloud production, point cloud refinement, and DEM generation. The initial phase involves calibration [476] and a combination of laser data and trajectory for initial point cloud generation [477]. Point cloud refinement includes removal of LiDAR point noises and outliers [478], as well as ground point classification [479]. Finally, DEM generation encompasses ground point filtering, followed by visualization and analysis. For further details, interested readers are referred to the provided reference.

Alongside the applications of UAV-borne LiDARgrammetry in urban mapping [28], this RS method are increasingly used in geoscientific studies for mapping inaccessible terrain and generating 3D models of geological features like cliffs, coasts, and volcanoes. This section explores UAV-borne LiDARgrammetry applications in geoscience, emphasizing its distinct utility from airborne LiDARgrammetry despite shared feasibility (Table 21). The applications mainly focus on capturing the surface geometry of the Earth’s exterior.

3.3. UAV-Borne Geophysical Survey: Integrated Approach

Sensor integration and data fusion are hot topics in traditional geophysical methods, encompassing ground-based methods [513,514,515,516,517,518,519,520,521,522,523,524], as well as spaceborne and manned airborne methods [525,526,527,528,529,530]. Even fusion scenarios involving ground-based geophysical data and spaceborne imageries are explored [531]. Extending into UAV-based geophysical surveys, sensor integration and data fusion continue to be of interest. Table 22 offers a comprehensive review of studies in this domain.

4. Discussions and Conclusions

This study presents a comprehensive review of cutting-edge geophysical survey techniques applicable via UAVs. To the best our knowledge, this is the first review to systematically compile and evaluate these methods. The reviewed methods encompass traditional geophysical approaches such as magnetometry, gravimetry, EM survey, GPR, gamma spectrometry, and radiometry, as well as non-geophysical methods like photogrammetry and LiDARgrammetry. The collected papers were categorized based on sensor type (active or passive) and another category explores sensor integration and data fusion concepts in UAV-based geophysical surveys.

While our study initially focused on geophysical survey methods, we also observed a significant rise in the use of UAV-based RS methods for various geoscientific applications, including geological mapping. The number of research conducted in these areas has increased dramatically, leading to the emergence of numerous UAV-based geophysical RS systems worldwide, with their numbers continuing to grow. This growing interest can be attributed to the cost-efficiency and effectiveness of UAVs, sensors, and related devices, which strike a balance between traditional RS-based (satellite-based and manned aerial) geophysical methods and ground-based methods. Additionally, the unmanned nature of UAVs reduces risks and challenges associated with traditional surveying methods, making them a groundbreaking tool in geoscience and related disciplines, which usually deal with rough terrains.

Upon reviewing the literature, it became evident that UAV-based magnetometry and GPR survey methods are the most commonly utilized among standard geophysical survey techniques. In contrast, UAV-based gravimetry receives less attention due to the challenges associated with deploying gravimetric instruments on lightweight drones. Our observations also revealed a variety of options for UAV platform types suitable for unmanned aerial geophysical RS. Rotary-wing multi-rotor drones, known for their high maneuverability, are the most commonly used platform type, with unmanned helicopters showing similar applicability. Fixed-wing UAVs are better suited for surveying larger areas, and the addition of VTOL capability enhances their applicability in scenarios where traditional runways are unavailable. However, unmanned airships are not as favored as the other three drone types. Regarding sensors, magnetometry benefits from a wide variety of sensor types among standard geophysical methods, while photogrammetry, as a non-standard method, exhibits the most variability in this regard.

Discussions on applications revealed that mineral exploration, detection of near-surface ferrous objects (primarily in magnetometry and GPR), soil contamination mapping (mainly using gamma survey), and landslide/subsidence mapping (mainly in photogrammetry and LiDARgrammetry) were the most prevalent applications. While some methods such as photogrammetry and LiDARgrammetry were primarily used for spatial and geometric analysis of the Earth’s crust (e.g., deformation mapping), others like MS imaging and HS spectroscopy found various applications related to soil mapping and analysis. Interestingly, certain applications such as mine or UXO detection were shared between different geophysical methods, illustrating the versatility of UAV-based geophysical techniques.

Although UAV-based RS is unmanned, it does not eliminate the need for fieldwork entirely. While certain survey methods like GPR require less fieldwork, others such as spectrometry/radiometry (for collecting ground-truth samples for laboratory analysis) and photogrammetry (for establishing ground control points) necessitate more extensive fieldwork. Thus, despite advancements, fieldwork remains essential in many UAV-borne survey methods.

The reviewed literature highlights significant attention to sensor integration in UAV-based geophysical RS. This integration involves deploying multiple geophysical sensors on a single or multiple UAV platforms. The benefits are evident, as it reduces costs, time, and human resources by enabling the collection of diverse data modalities in a single sortie. Notably, while individual sensors offer limited insights, combining data from multiple sensors can uncover novel perspectives not achievable with a single sensor. For instance, while an optical camera captures surface information, magnetometers delve into subsurface details, enhancing our understanding of the study area. However, sensor integration and data fusion pose several challenges, including differences in data modalities, acquisition time, misregistration, varying viewpoints, and spatial resolutions. These challenges become more pronounced when integrating UAV-based data with data from different sources such as space-borne, airborne, and ground-based platforms, as observed in our review. Consequently, careful consideration is essential when fusing different types of geophysical data to ensure accurate and meaningful results.

In conclusion, the reviewed studies unanimously affirm that UAV-borne geophysical RS methods offer comparable results to traditional ground-based and aerial methods. The cost-effectiveness and unmanned nature of UAVs, sensors, and related devices have revolutionized geoscience and related disciplines, bridging the gap between satellite-based, aerial, and ground-based geophysical methods. Despite significant progress, ongoing efforts are essential for further advancement in sensors, platforms, and methods. In terms of sensors, there is a need for more options capable of deployment on lightweight UAVs, similar to the variety available for traditional methods. Platforms are also evolving to become lighter and more endurance-focused, benefiting not only UAV-borne geophysical RS but also all domains of unmanned aerial RS. Methodologies require tailored development to suit UAV-specific requirements, such as customized processing methods.

The contribution of this review study extends to researchers across geoscience disciplines, providing a comprehensive and systematic overview of the possibilities offered by UAV RS in geophysical surveying. For instance, it aids researchers in selecting suitable methods, sensors, and UAV platforms for their desired applications. Overall, this review acts as a valuable resource, much like a mini-handbook, for the geoscience community.