Mapping 25 Square Miles of the Pearl River
Drones provide better data and no one understands this better than Mississippi State University (MSU). Recently, our team worked alongside MSU to map a fascinating and challenging project at Pearl River; a historical river which runs through the heart of Mississippi and the south-eastern tip of Louisiana, boasting a meander length of 444 miles. Funded by the NOAA Unmanned Aircraft Systems Program, MSU was faced with the monumental task of monitoring 25 square miles of the delta of the Pearl River. To accomplish this task, MSU contacted Altavian about collecting multispectral imagery by using drones.
MSU’s research required aerial mapping of the waterways in order to:
Monitor land and water boundaries
Identify plant and animal species, vegetation type and density
Identify invasive species
Discriminate intertidal water
Detect temporal changes in waterways
To accomplish this, they would need high-resolution digital elevation models (DEM), multispectral imagery as well as water quality assessments. To top it off, they would need all of this data updated every 60 days for the next 2 years, for comparative analysis.
This was a big enough project for a manned aircraft in terms of acres; twenty five square miles equates to about 16,000 acres and our typical cutoff is 10,000 for the Nova and 100 for the Galaxy – but this was not a typical project. Altavian has long worked in the public space (being the only business venue before 333 exemptions allowed for commercial operations) and the downside had always been the lengthy FAA approval process. However, for this project, a different set of rules applied. For these missions we could operate from a moving watercraft and we could operate at higher altitudes than normal, allowing us to easily stretch that 10K acre mark and under the amount of time that we initially thought possible.
The research required a higher degree of detail than was feasible from manned aircraft altitudes and speeds, which would in-turn require a platform that could safely fly slower and closer to the river. Having the capability of a 90-minute flight endurance, the Nova would allow us to cover a large amount of acreage as we followed it around in the boat. The all-environment (AE) version allowed us to land the UAV directly on the water, which would be the only option in the marshy delta. Equipping the Nova with a Fusion payload featuring a large DSLR sensor, we would be able to cover more acreage-per-flight while maintaining a 5cm ground sample distance (GSD). We have a line of Fusion payloads that are modified to add multispectral imagery capabilities for exactly this type of mission and decided to go with a precisely-modified color-infrared sensor to maximize the ultimate usefulness of the data and increase collection efficiency.
To properly map 25 square miles of river, we would have to work directly from boats while navigating the river in real-time. Dividing into to two teams, each with their own boat – the ‘Flight Team’ and the ‘Water Team’, we worked together to gather data simultaneously. As the Flight Team flew overhead, the Water Team would collect water samples, testing for: toxins, nutrients, microscopy, metals, suspended particulate matter, phycocyanin, colored dissolved organic matter, and bacteria counts.
Some of the flights allowed us to follow the main channels of the river, while others required us to be in the much smaller, tighter auxiliary channels. Since it could be difficult to navigate the boats within the narrow channels, we would find an area to drop anchor and plan a flight as large as we could while maintaining visual line of sight (VLOS). The versatility and simplicity of the Nova and Fusion gave us plenty of options to work with during these constraints. The systems are fully-autonomous so we didn’t have to worry about keeping the aircraft flying or collecting data while working the navigation problem on the water, everything preformed beautifully on its own.
Having the ability to hand-launch directly from the bow of the boat and land in the water next to us allowed the crew endless options for anchor locations. However, being anchored in a single spot for up to 90 minutes yielded one unique problem; we tended to draw the attention of several creatures including alligators and all types of insects. The alligators made their presence known and left us alone. The insects however, were relentless.
Working from the main channels was a much different experience. Since there are few limitations to mobility within the wide channels, we found that we could plan much larger mapping flights. These large flight areas required us to have more communication with the Water Team – we would need to inform them of the when and where the Nova would be located throughout each flight so they could navigate to the appropriate positions underneath to collect their water samples.
For the first few trips to Pearl River, we would plan flights containing flight lines that ran parallel to the main channels as long as 5 kilometers. Since the Nova can cruise with speeds ranging from 16 m/s to 22 m/s, we would chase the Nova by racing up and down the river in the boat to maintain VLOS. This was quite exciting but could be very taxing on the crew and the equipment on the boat. It also made it very difficult to communicate with the Water Team as the boat motor was roaring at full throttle.
After a few visits to Pearl River, we found that if we planned the flight lines to run perpendicular to the channel, we could slowly putter up the river as we followed the Nova while it progressed up the channel. This would add more waypoints and flight lines to the flight but did not affect the amount of coverage, battery endurance, functionality of Flare, the autopilot, or payload. We found the flights to be only about 10% longer than before and allowed for a better working environment. We also inadvertently realized that the movement of the boat greatly reduced the amount of wildlife that found us… interesting.
When we returned from our last trip, we created a short video montage of takeoffs and landings from this project for the world to see.
Overall, we were able to cover the entire 25 square miles by dividing the river delta into 19 flights – ranging from 650 acres to 1400 acres per flight. Each time we flew the project, we became more familiar with the river, and working together with MSU we became more efficient. We found that we were able to successfully map the river delta in 3 days’ time, each trip collecting over 100GB of raw multispectral imagery – averaging just over 15,000 images taken.
Throughout the 2 years using the data collected by both the Flight Team and Water Team, MSU had the ability to see small changes in aquatic vegetation over time, with some species increasing and others decreasing. The ability to resolve sediment-rich plumes and high-flow volumes helped resolve a lingering question of where water was being stored, when, and how, as well as the path of water through the marsh during high-water events. And, the imagery provided evidence of the capability of hand-launched UAS to resolve various animal and vegetation wildlife.
Only after seeing the final processed data did it dawn on us the sheer size of this project. Scrolling and zooming around the data, it’s amazing to find yourself on a boat and realize how zoomed in you are:
Here is a small segment of the project, showing a transition fade between the raw CIR data and the processed NDVI map: