COS 69-1 - Using Unmanned Aerial Vehicles (UAVs) to sample and remotely sense aquatic ecosystems

Tuesday, August 8, 2017: 1:30 PM
B114, Oregon Convention Center
Amy J. Burgin1, Keunyea Song2, Carrick Detweiler3, Ajay Shankar4, Austin Song2 and Shawyan Ahmadian2, (1)Kansas Biological Survey, University of Kansas, Lawrence, KS, (2)University of Kansas, Lawrence, KS, (3)Computer Science & Engineering, University of Nebraska Lincoln, Lincoln, NE, (4)University of Nebraska Lincoln, Lincoln, NE

The environmental applications of unmanned aerial vehicles (UAVs) for research and management has taken off over the last decade. However, a water-sampling UAV has not yet been developed and tested. The application of a water sampling UAV could replace conventional water sampling, which are labor-intensive and limited by accessibility to sites and boat deployment. Additionally, current methods are limited by sampling one or few points grab sampling and or placing a limited number of sensors in static locations for in-situ readings, both of which often ignore the spatial and temporal variations within a lake. In this study, we ask: Is a water-sampling UAV effective in terms of cost, time resource, and labor and does it provide comparable data to traditional methods? We examined our UAV’s capability to collect water samples and physicochemical readings in a pond and controlled experimental mesocosms. We then validated the data quality (spatiotemporal resolution and credibility) and resources (sampling time, preparation time, and cost) by comparing with those with traditional methods, manual grab sampling, readings and sensors. We compared three methods: 1) manual devices (e.g., YSI), 2) sensors (e.g., HOBO), and 3) our UAV, each measuring temperature and conductivity in controlled mesocosms and in a reservoir.


We successfully forced mesocosm stratification, allowing us to compare temperature profiles in three (10,000L) tanks with a 6°C temperature gradient to three control tanks (~1.5°C temperature change). Temperature profiles were similar for all three methods (YSI, HOBO and UAV), however, the UAV and YSI were most similar with only a fraction of a degree’s difference, whereas the HOBO sensors were up to 1°C higher at the top of the tank. We also compared the three methods by collecting a temperature profile from a reservoir at different times of day. Temperature differences were not as large as in our mesocosms, and thus the comparison of the three methods were very similar in terms of temporal and spatial variation. We also manipulated conductivity in tanks, forcing a salt gradient in three tanks to compare to three control tanks. Conductivity comparisons among the three methods did not agree as well as the temperature comparisons. Again, the UAV and YSI methods were comparable, but the HOBO sensors often were 500-1000 µS greater than the YSI or UAV. Variation between replicate measurements was smallest with the UAV and largest among the HOBO sensors. We found that UAV complemented manual and sensor readings and sampling.