Volumetric Water Content: Keeping your Eye on the Goal
Most scientists agree that it’s productive to attend seminars and conferences in order to talk with peers, share ideas, and learn about what other scientists are doing. However, in both academia and industry, we need to be careful that we are not so easily influenced by other scientists’ opinions that we lose sight of the end goals of our own projects. This happened to us recently at Decagon. Here’s the background: our volumetric water content (VWC) sensors actually measure the dielectric permittivity of the soil and use a transfer function to predict VWC from the measured dielectric value. Most of our sensors receive a “dielectric calibration” during the production process where they are calibrated in five dielectric standards to make sure they all measure dielectric permittivity accurately, thus leading to accurate VWC measurements with our standard transfer function.
We were doing a pretty good job calibrating these sensors in dielectric standards, and our default dielectric-to-VWC transfer function resulted in good VWC accuracy. Then we went to a series of meetings and talked to some of our researcher friends who work on instrumentation. They said, “Look, your water sensors aren’t reading as accurately as they should in dielectric permittivity.” Here’s where the trouble started…
Wanting to make the perfect instrument, we went back and re-evaluated the dielectric calibration standards for these water content sensors and tried to use the book values of dielectric permittivity. This was a bad idea because it fundamentally changed the sensor output. Now, despite the sensors giving a slightly more accurate value for dielectric permittivity, they gave less accurate measurements of VWC. Compounding the problem, we now had a population of sensors that didn’t read the same as earlier sensors of the same type. So when customers started replacing their old sensors they said, “Wait a minute, this sensor reads 4% higher water content than my old water content sensor.” That’s when we realized that we had a real problem.
Our underlying mistake here is that we failed to remember that 99% of the people who buy our VWC sensors don’t even care what dielectric permittivity is. They just want an accurate, repeatable measurement of soil moisture. Essentially, because we were so focused on trying to produce a theoretically perfect sensor for a vocal minority of technically savvy users, we lost sight of the practical matter. Did our sensors produce an accurate water content measurement?
I wonder how often this happens in academia and industry. Scientists are bombarded with input from so many different stakeholders, it’s sometimes difficult to maintain the original focus of their projects. We need to remember to focus on the end goal and filter out things that may distract from that goal.
Get more information on applied environmental research in our