Case Study Review: Geological Mapping Using Drone-Based Photogrammetry

Updated: Oct 26

In this series, we will go through many case studies on using drones in mineral exploration. Today we will discuss a case study of Geological Mapping Using Drone-Based Photogrammetry: An Application for Exploration of Vein-Type Cu Mineralization

Honarmand, M.; Shahriari, H. Geological Mapping Using Drone-Based Photogrammetry: An Application for Exploration of Vein-Type Cu Mineralization. Minerals 2021, 11, 585.

This study aimed to conduct a large-scale geological mapping using drone-based photogrammetry to facilitate mineral exploration in the Shahzadeh Abbas Cu deposit.

Shahzadeh Abbas Cu deposit is situated 28 km east of the large Sar Cheshmeh Cu porphyry mine, Kerman province, Iran. Vein-type Cu mineralization occurs in joints and fractures.

The topography of the study area is rough, which causes difficulties for field investigation and geological survey. Thus, drone-based imagery was undertaken to project geological features on a base geology map. In this research, several photogrammetry products are prepared to complete fieldwork data to specify mineralized zones for future studies.

Geology maps are one of the essential requirements of any mineral exploration campaign. Remotely sensed data are generally used for preparing geology maps. In the case of vein-type mineralization, drone imagery can provide high-resolution images suitable for deriving geological and structural information according to mineral exploration objectives.

Five main parts were considered to reach the research target, including

(1) Setting the flight parameters and flying the drone.

Commercial multi-rotor platforms have been successfully used for geological applications. In this research, a DJI Phantom 4 Pro V2.0 was used with its original sensors (Figure 2a). It carries a built-in 1-inch 20 Megapixels CMOS camera on a gimbal. The camera’s lens offers a field of view (FOV) of 84t with a focal length range of 8.8 mm/24 mm (35 mm format equivalent) and aperture f/2.8–f/11, with autofocus.

The flight path was designed in DJIGo software. A minimum of 70% side and 80% front image overlaps were applied to have an accuracy of less than 5 cm. The flight was performed at an altitude of 70 m (flight speed of 72 kph) to meet a spatial resolution of less than 10 cm. The other flight settings were carried out using the Drone Deploy application.

Figure 2. (a) DJI Phantom 4 Pro V2.0, (b) TGCP (marker), (c) SOUTH Galaxy G1+ receiver on a benchmark, and (d) Siemens star target.

(2) Acquiring and processing images

A combination of the built-in GPS of the drone and a grid of temporary ground control points (TGCPs) or markers is necessary to improve the accuracy of the geometrical corrections used.

An accuracy of less than 5 cm was aimed for geological mapping. Accordingly, 220 TGCPs were considered to ascertain the accuracy requirements. TGCPs were marked in the form of red color crosses on the ground surface (Figure 2b). They were accurately positioned using a ground-based real-time kinematic–differential global positioning system (RTK-DGPS) from the SOUTH Galaxy (Figure 2c).

Besides, five checkpoints with known coordinates were considered to estimate the overall accuracy of the orthomosaic image of the whole area. Several types of calibration targets, namely bar target, Slanted Edge Test, and Siemens star are generally performed to assess the spatial resolution of UAVs. The Siemens star enables the spatial resolution to be measured in all directions for the flight path of the UAV

A Siemens star target on a flat surface was used to determine the spatial resolution of the CMOS camera (Figure 2d).

The number of acquired images is a function of the area size and image overlap. After the aerial survey, Agisoft PhotoScan software was applied for photogrammetric processing. Essential data including discrete images and camera locations were loaded from the drone to the software to initialize the photogrammetry procedure.

The World Geodetic System 1984 (WGS84) was selected as the datum.

Camera locations and matching points were used to align images and establish a sparse point cloud model.

The model includes a set of aligned discrete images and must be converted to a georeferenced dense-point cloud model.

After constructing the sparse point cloud model, the georeferencing task was performed using 220 TGCPs. As a result, a dense point-cloud model was built in the next step.

A dense point-cloud model is a single display of all acquired images that are connected based on calculating the depth information for each camera location.

The mesh and digital elevation (DEM) models are normally built from the dense-point cloud model.

The mesh model which is a display of the surface and/or volume of objects was reconstructed. Both the mesh and DEM models can be used to build the orthophoto image (Orthomosaic).

"Reviewing many publications indicates rapid progress in the use of drones in the mining industry, from mineral exploration to mine exploitation in different places of the world" .

(3) Creating a draft geology map.

Drone-based imagery provides worthy information for the instantaneous discrimination of lithology before undertaking the geological survey. Thus, a draft (first edition) geology map can be prepared based on photogrammetry products to save the time and cost of the fieldwork. If the drone is equipped with a multispectral or hyperspectral sensor, mineral mapping can be carried out in addition to lithological mapping.

The visual interpretation of images can be performed based on the color changes of rock units and can lead to the primary discrimination of lithology in the study area. Therefore, the first edition of the geology map (draft geology map) can be produced. The accuracy of the draft geology map depends on the experience of the image interpreter(s). High-resolution images can also be utilized for extracting lineament features from photogrammetric products. The drift geology map is finalized after conducting fieldwork but it can be applied to plan the subsequent geological survey.

In this research, the orthophoto image was created using the mesh model for providing the draft geology map.

(4) Performing field work.

Field investigation and laboratory studies are essential parts of a mineral exploration program. A geological survey was undertaken considering the draft geology map. Rock samples were gathered from predefined locations to verify the lithology and to check the possible mineralized zones. Thin section studies were performed to specify rock types. Finally, fieldwork data and ground truth were used to finalize the geology map of the study area.

(5) Finalizing the geology map.

5.1. Achievements of Flight Settings Reducing flight altitude is essential to increase the spatial resolution for the accurate determination of geological boundaries and position of faults. However, it increases the time of aerial survey and image processing procedure. The higher image resolution of UAVs compared to satellite images can lead to more accurate structural information being derived. The flight was performed at an altitude of 70 m. Therefore, the spatial resolution of 3.26 cm was calculated using the Siemens star target. This spatial resolution satisfied the required accuracy for outlining the lithology and especially geological lineaments. Flight lines were designed to have 70% lateral and 80% front overlaps. In Figure 3a, the blue color demonstrates the area with excellent image overlap. Therefore, a perfect image overlap with no gap was achieved in an area of 2.02 km2. About 4018 images were acquired to build the dense-point cloud model. Georeferencing the dense-point cloud model was accomplished using 220 TGCPs (about one TGCP per hectare). Figure 3b exhibits the distribution of TGCPs in the study area. A root mean square error (RMSE) of 2.91 cm was measured for the TGCPs. The RMSE of 3.96 cm was obtained for five check-points that satisfied the required accuracy of less than 5 cm for the orthophoto (orthomosaic) image.

Figure 3. (a) Camera locations and image overlap, and (b) distribution of TGCPs in the study area.

5.2. Analysis of Orthophoto Map and Hill-Shade Model

CMOS camera of DJI Phantom4 Pro V2.0 is an RGB sensor. The camera acquires RGB images that are comparable with true color composites (TCCs) of Landsat ETM+/OLI or Sentinel. TCCs of medium resolution spaceborne sensors are used for geological mapping but the higher resolution of the CMOS camera helped to outline geological features more accurately.

The orthophoto image, which was derived from dense point-cloud, is presented in Figure 4a. The image provides a useful insight into the lithology of the study area. Rock units were specified based on their color differences. Geology boundaries were delineated thanks to the high resolution of the image. Primary polygons of rock units were drawn using ArcGIS software. The hill-shade image was utilized to extract lineament structures for detecting vein-type mineralization. The structural layer of the draft geology map was prepared using ArcGIS software. Figure 4b shows the location of structural features in the study area. To finalize the geology map, a geological survey was conducted. Considering the draft geology map, the situation of rock samples was defined based on a time and cost-effective plan to accomplish the fieldwork.

Figure 4. (a) Orthophoto image of the study area, and (b) hill-shade model containing lineament features.