Considering the Variability of Capacitance Sensors Due to Placement in a Greenhouse Production Area

Substrate volumetric water content (VWC) measurements can provide automated irrigation or irrigation decision support, based on capacitance sensor readings. However, proper sensor placement is critical to ensure reliable VWC measurement over large production areas. We investigated the spatial variability of sensors in a commercial snapdragon production greenhouse (Bauers Greenhouse, Jarrettsville, MD) that uses a closed-system hydroponic / perlite substrate production system. Each production bed was 28 m in length with a 3% slope; each bed contained 6 rows of 15 perlite bags, each row irrigated by two dripper tubes controlled by a solenoid. The top half of each row in the production area was irrigated from one solenoid; the lower half of each row irrigated from another solenoid at the base of the row. This was done to ensure equal water pressure to each area. Twenty capacitance sensors (EC-5; Decagon Devices, Inc., Pullman, WA) were all horizontally inserted at the same depth into five perlite bags at four different locations in the production area (top and bottom of each irrigation lateral). Substrate VWC (m³∙m^-3) was continuously measured every 15 minutes over a 14-week production period. When seedlings were transplanted into each bag, the VWC was maintained at an average of 0.31 m³∙m^-3 and there were no significant differences between the locations. However, as plants increased in size and water uptake, the top of the production bed showed a significant decrease in VWC to 0.27 m³∙m^-3, whereas the substrate VWC at the bottom of the production area increased to 0.32 m³∙m^-3. There were no significant differences between sensors in the five different bags at each location, and the sensor-to-sensor variability had an average standard error of 0.01 m³∙m^-3.  EC-5 sensors proved reliable in monitoring VWC in this perlite hydroponic system; however, uniform sensor placement both in terms of depth in the bag and within the production area should be carefully considered to reduce variability, especially when those readings are used for control-based irrigation decisions.