Nicole suggested that I post about the little research project that I'm doing in Boise. First, the non-technical details. My dad started a technology irrigation company about 10 years ago...it has huge upside and over the years has attracted a lot of attention from several prominent irrigation companies and venture capital groups. The heart of the business is a soil moisture sensor that, as you might have guessed, can accurately determine the moisture content in the soil where it's buried. The sensor is used to control when to irrigate and how much water to apply. Because of the drastic water savings this offers, some installations have paid for themselves in a single watering season. My dad's company, Acclima, pride's itself in having, by far, the best soil moisture sensor in the industry. Most other sensors fail when the soil is slightly conductive (salty). For my little research project, I was tasked to stretch the capabilities of the sensor even farther.The techy stuff for my nerdy friends: the sensor in my right hand (bottom) is a low-cost time-domain reflectometer (TDR), and the sensor in my left hand (top) is a low-cost time-domain transmissometer (TDT). In both cases, a step function propagates down the transmission rods. After it travels through the medium (soil), it can be sampled at a resolution of 5 picoseconds (200 GHz)! Features of the sampled waveform are extracted to deduce bulk permittivity and conductivity. Using a dielectric mixing model, you can back out the volumetric soil moisture content. My six weeks focused on creating experiments and simulations to determine the limits of the sensor (e.g.,. how precise can we expect measurements to be?), and implementing microcontroller code to achieve it. What did I find? If it didn't kill your grass, you could water your lawn from the Great Salt Lake, and our sensor could still track the moisture.
