81 villages in 21 days: Pipeline survey with drones
What we learnt about large scale mapping
— Research by Niloy Jain and Sajil KR
In 2018, a client approached us with a problem of surveying a collection of 81 highly dispersed villages. The drone based survey outputs were to be used by the government to lay pipelines for drinking water. In order to create a robust plan, the authorities needed maps with 5 cm resolution, 25 cm interval contours, and z-level accuracy of 15cm to maintain elevation accuracy required for pipeline planning.
These villages are scattered in an area of 400 sq. km. with the average village area being upto 2 sq. km and the average distance between the villages is around 5 km. Covering them individually with a small drone would have taken a long time. We had to keep the entire project economically and operationally feasible.
To map the whole or in parts that is the question
The first question we encountered was whether to map these 81 villages individually or to map the entire 400 sq. km area containing these villages. The entire operations including laying out the Ground Control Points were to be completed in 20 days. We decided to map the entire 400 sq. km continuous area to avoid myriad small deployments.
What kind of drones to use?
There are the quadcopters, ubiquitous, easy to fly, but very small coverage for operational efficiency. Then there are the fixed wing drones, more complex, needs well trained pilots, faster mapping, which means the crew laying ground control points would have to keep up. However, the latter seemed a less tedious solution even though it needed more technical expertise.
To complete the operations within the timeline of 20 days, we divided the area of interest into multiple plans based on our in-house processing capabilities. We calculated the overlaps, the flying height, the GCP locations based on these plans. The deployment itself was planned to be in four phases, with two fixed wing drones (a flight time of 1 hour each), capturing roughly 4–6 sq. km per flight and 6–8 such flights per day.
The GCPs were planned to enable our tech team to stitch multiple plans together into a single gigantic orthophoto of 400 sq. km. A total of 80 GCPs were placed and DGPS readings of these points taken. Like a patchwork quilt, the GCPs were planted along the edge of the entire swathe and on the overlapping parts of different flights so that they’d be stitched together during processing.
Two operations and 80 drone flights later, we ended up creating the following:
We captured and visualized the data in high resolution maps and geotagged features and assets. Detailed survey grade contour (25 cm interval) and elevation map (15cm accuracy) was used to plan the pipeline work on the ground. All these deliverables were given to the planning and project management teams in the form of a custom application capable of live contours, distance, area, and volume measurements.
Statistics:
- Area covered: 380 sqkm
- Total length of roads digitized: 763.2808 Km
- Total number of elevation points extracted: 96546
- Total buildings extracted: 57100
- Number of wells extracted: 12679
- Hydrological structure: 542
- Waterbodies: 1165
- Number of bridges extracted: 781
CAD Maps:
The extracted features are converted into CAD file formats for pipeline planning with elevation points annotated at each point of interest.
A sample CAD map is shown in the image below:
Web application:
