Using low-cost Arduino based GNSS boards to collect centimeter accurate topography from drone orthophotography

A low-cost, cm accurate, multi-frequency GNSS survey system was built and used to georeference photogrammetric Digital Elevation Models (DEMs) made from drone orthophotography. The GNSS system uses a pair of Arduino based GNSS boards that are equipped with the U-blox ZED-F9P GNSS module. One of the boards is configured as a base station that sends RTK corrections to the second board that serves as the rover. A Raspberry Pi collects and stores raw GNSS data received during operation of the base station. Using these raw GNSS data, a solution for a precise base-station position can be obtained from NOAA NGS OPUS or the NRCAN-PPP service. The rover unit is equipped with Bluetooth, and its position can be logged with a Bluetooth enabled cell phone.

Two test surveys were conducted using the GNSS equipment that were coupled with drone collected aerial imagery in East Brunswick, New Jersey. The designated GNSS base station unit was placed over a survey mark, twice, and collected 4 hours of raw data on each occasion. The raw data from the base unit was sent to the NRCAN-PPP service for an estimate of the "base" survey mark's position. Non-toxic spray chalk was used to marked 64, 2 ft. x 2 ft. "X" marks on site that were subsequently measured with the RTK rover, twice, separated by a period of an hour. The drone imagery was processed using Structure from Motion photogrammetry software to create a DEM, and ground control points (GCPs) were used for georeferencing. The twice-measured survey marks served as: 1) measurements for testing the accuracy of the RTK system; 2) GCPs for the orthophotos; and 3) measurements for testing the accuracy of the drone derived DEM.

The difference in position between the corrected solutions for the base-station location was 1.3 cm horizontally and 3.1 cm vertically. Testing the accuracy of the RTK data revealed a 2.9 cm horizontal RMSE and a 4.4 cm vertical RMSE. Photogrammetric DEMs derived from the aerial imagery tied to the RTK-based GCPs achieved sub-decimeter level accuracy. This technology facilitates the use of drone photogrammetry for studies that require precise measurements of stratigraphic sections and coastal geomorphological change. The accuracy and flexibility of multi-frequency base-rover GNSS at the cost of this "hacked" system supports the conclusion that it can be considered for use in further studies.