28Jan
On: 28 January 2020 In: Mines & Aggregates

Are drone surveys accurate enough for stockpile inventories in mining, aggregates and construction?

Drones have so many advantages when it comes to surveying stockpiles. But you might be worried about the accuracy of the volume calculations you get from this aerial data. We’ve surveyed a +3000 m³ stockpile using different methods to compare results. Guess what? Drone surveys are even more accurate than traditional ground surveying methods. 

Trusting that the results of stockpile volume calculations are accurate is key for site management and accounting 

If you’re wondering if you can trust drones to perform your stockpile inventory, this article is for you. 

Mining, quarry, and construction companies are using drone technology today to more easily, consistently and effectively digitize their sites, extract and share meaningful insights, and put everyone across their teams on the same map.

These companies perform stockpile inventory on a monthly, quarterly or yearly basis, in order to carry out financial reporting on volume inventories and to monitor production and collaborate with third-party contractors. 

While capturing data more quickly is easier than ever with drones, companies need to understand the level of accuracy of that data as it plays a major role in site management and accounting. 

We at Delair conducted a study at a customer quarry site located near Paris, France. The objective was to compare the accuracy of drone-captured data with that of traditional survey data, both of which would be compared against 3D scanner data. The same stockpile was used for each survey.

Here are the results.

Stockpile volume calculation with traditional surveying methods

Traditionally, site managers conduct surveys on the ground with a GNSS receiver which  is used to determine the exact position of each measured point with centimetric accuracy.

GNSS stands for Global Navigation Satellite System, a constellation of satellites orbiting the earth. Common GNSS constellations include GPS, GLONASS, Galileo, Beidou and other regional systems. A GNSS receiver allows you to measure the precise location of any point anywhere on Earth. 

Step 1: Measure points on the ground with a GNSS receiver

The surveyor is required to collect numerous measurement points around the base of the stockpile, and if it’s safe to do so, will also climb onto the pile to take additional points. A single 1,000 m²  (10,760 ft²) stockpile requires 150 shots (latitude, longitude, and elevation) on average.

Step 2: Calculate the volume from the Triangular Irregular Network surface model

The surveyor then returns to the office to construct a 3D model from which he or she can determine the volume of the stockpile. The coordinates of all points are connected to form a Triangular Irregular Network (TIN) representing the surface. Civil engineering software is then used to build a surface from the TIN and calculate the difference in volume between the TIN and the actual ground elevation.

Fig 1: Point cloud of the stockpile surveyed on the ground, with a GNSS (left), corresponding terrain (right).

Fig 1: Point cloud of the stockpile surveyed on the ground, with a GNSS (left), corresponding terrain (right).

 

While surveyors and site managers recognize the value of physical on-site inventories of stockpile volumes for accounting purposes, most of them would concede that it is still a time consuming and expensive process, often  requiring third-party surveyors to perform the analysis. In addition, there are inherent safety risks with climbing stockpiles.

Drone surveying to measure stockpile volumes

Aerial surveying with drones enables surveyors to measure the stockpile with a dense point cloud using photogrammetry processes.

Step 1: Lay out GCP targets and measure their locations or use PPK

The initial setup of Ground Control Points (or GCPs) is a crucial step to ensure the accuracy and the quality of  drone-acquired data. 

Ground Control Points are points of known geospatial coordinates within an area of interest. They come in large marked targets on the ground,  positioned throughout the area of interest. They reduce positional drift caused by GPS signal interference and increase the absolute accuracy from  within meters to within centimeters or better.

GCPs need to be clearly visible in the pictures taken by the drone. They are placed homogeneously throughout the site, both in elevation (planimetry) and surface (altimetry). A GNSS survey instrument is used to accurately locate each GCP.

High-accuracy GNSS receivers onboard a professional drone and advanced GNSS data processing such as PPK (post-processed kinematic) or RTK (real-time kinematic) can enable highly accurate maps with few or even no GCPs in some cases. PPK or RTK can serve as a supplement, or even an alternative, to using GCPs for a drone survey.

The positioning and surveying of GCPs on the ground before the drone flight.
The positioning and surveying of GCPs on the ground before the drone flight.
The positioning and surveying of GCPs on the ground before the drone flight.

Fig 2: The positioning and surveying of GCPs on the ground before the drone flight.

Step 2: Fly the mapping drone

Once the GCPs are positioned and surveyed, the operator flies the drone and maps the entire site automatically by taking multiple pictures.

In order to capture the best and most accurate data, Delair recommends to:

  • Perform the flights at  120 meters (400 feet) maximum, with maximum longitudinal overlap and lateral overlaps (60% – 80%). The overlap corresponds to the amount by which one photograph includes the same area covered by another.
  • Ensure good picture quality and flight altitude to get the right GSD (resolution), which should be 3 cm/pixel (1.8 inch/pixel) maximum. Avoid blur, over/under exposition or high ISO.

Fig 3: Flight plan overlap illustration

Fig 3: Flight plan overlap illustration

 

Step 3: Photogrammetric processing of drone imagery and associated metadata

The drone data is uploaded to the Delair Aerial Intelligence platform (delair.ai) for automatic processing, to generate 2D and 3D models.

The processed data consists of millions of points allowing 3D reconstruction of the site and enabling volume calculation.

The photogrammetry process matches the georeferenced points that were surveyed with the GCPs with the overlapping pictures. Each point is georeferenced in XYZ as with a traditional survey.

Point cloud of the stockpile surveyed with a fixed wing drone

Fig 4: Point cloud of the stockpile surveyed with an eBee Plus at 120 m – 400 feet (left), corresponding terrain (right).

Side-by-side comparison of ground-based GNSS vs aerial survey methods for stockpile volume calculation 

We surveyed the same stockpile with a laser scanner, traditional GNSS survey methods and two different drones in order to compare the number of data points. For this test we used the Leica ScanStation P20, on multiple fixed locations. The volume obtained with this method: 3,192.7m3 (112,749 ft3).

Fig 5. Point cloud of the stockpile surveyed with a terrestrial LiDAR (scanner) on the left, corresponding terrain (right).

Point density
(dense point cloud)
Point accuracy Volume % Difference  with scanner
Traditional surveying 0.13 pt/m²

(0.01 pt/ft²)

2 cm XY, 3 cm Z

(0.8 inch XY, 0.8 inch Z)

3,263.7 m³

(115,230 ft³)

2.22%

Drone-based surveying:

Fixed-wing drone at 120 m

442 pts/m²

(41 pts/ft²)

3 cm XY, 2,6 cm Z

(1.2 inch XY, 1 inch Z)

3,179.7 m³

(112,265 ft³)

0.31%

Drone-based surveying:

DJI Phantom 4 Pro at 120 m

462 pts/m²

(43 pts/ft²

3 cm XY, 4 cm Z

(1.2inch Z, 1.6 inch Z)

3,182.8 m³

(112,371 ft³)

0.41%

 

Performing an aerial survey on this stockpile yielded a variation of only 0.31% to 0.41% from laser scanning, compared to >2% from traditional surveying. Drone data is ideal for stockpile volume measurements, as there is only a 10 m3 (0.3%) difference from laser scanning.

 

Fig 7. Stockpile Output Cross-Sectional Views

Fig 6. Stockpile Output Cross-Sectional Views

 

Drone surveys deliver accurate results for stockpile volume calculation

As a conclusion, based on this analysis, the drone-based stockpile volume obtained with delair.ai cloud platform is within 0.31% and 0.41% of laser scanning and produces more accurate results than GNSS rover based surveys. With this level of precision, customers have found that drones and cloud-based data processing with delair.ai provides the accuracy necessary to comply with accounting requirements.

Aerial surveying allows for fast and accurate stockpile volume calculation.

Additional benefits of drone surveying include speed and safety. Typically, a drone survey can be completed much more quickly than a GNSS survey and does not require that a surveyor climb on stockpiles which can be a tremendous safety risk.