Using Eco-drone Sensors To Map Water Quality

The recent Lake Merritt Bioblitz organized by Nerds For Nature was a huge success from several angles. Besides the amazing biodiversity that was observed and documented there, and the scores of kids and families digging in the mud flats and otherwise discovering the nature around them, a few of us Nerds For Nature demonstrated a brand-new method to measure water quality.

Sean Headrick of Aerotestra is ready to put the lid on his “Hugo” quadcopter before the test run — photo by Ken McGary

Using a floating quadcopter made by local start-up Aerotestra, we proved that it is feasible for an autonomous flying robot to map various parameters at multiple GPS waypoints without human intervention. Lake Merritt is a particularly interesting location for this sort of environmental monitoring, as there are many different water flows into this tidal lagoon, which can dramatically affect the wildlife that can survive and even thrive here, as this research paper explains.

 

Arduino Micro board forms the heart of the DroneLogger — photo by Ken McGary

The story is summarized nicely in this article in Scientific American. Besides coming up with the idea and organizing the project overall, my technical contribution was a custom Arduino-based datalogger that can take readings from temperature, pH, salinity, and other sensing probes at regular intervals (once per second in this case) and record the data in CSV format on an xD memory card. You can see a lot more bioblitz photos in this Nerds For Nature Flickr set.

 

Eco-drone in flight between waypoints above Oakland’s Lake Merritt — photo by Ken McGary

The collected data shows probes in the water at start/end and four intermediate waypoints. You can also play with the dataset graphs yourself on the ManyLabs Data Hub. No formal calibration protocol was used but it’s in the ballpark and good enough for our proof of concept. Next up, accuracy!

 Logged data from temperature (top) and pH (bottom) probes from multiple waypoints — chart image by Ken McGary

Clearly the temperature sensor needed more time at each waypoint to equilibrate. Next time we’ll have a much better idea of how to optimize the testing sequence eliminate these errors. At any rate, we proved the concept wonderfully, thanks Sean! I only wish we had it all on video…

 

Comments are closed.