WiField, an IEEE 802.11-based agricultural sensor data gathering and logging platform
A new agricultural sensor data logging platform (WiField) is described, based on IEEE 802.11 WiFi technology. It is low-cost, low-power, and achieves long (>2km) range communication to on-farm WiFi access points. WiFi is an attractive choice for this application because of the wide range of other devices that increasingly need internet access in farming systems. The WiField devices include interfaces for many sensor types; weather, infrastructure (tank and irrigation water levels), and soil status sensing. The interfaces and example corresponding sensors include SDI-12 (capacitive soil moisture probes), soil tension (matric potential), analog voltage and current, UART (water depth sensing using ultrasonic transducers with a digital interface), RS-422 (integrated weather stations), one-wire (DS18B20 temperature sensors) and pulse (flow meters, wind and rain sensors). It integrates solar charging of rechargeable batteries, or can be run off disposable batteries for at least an entire growing season due to design choices that minimize power consumption. It is designed to upload data to cloud services in real-time. The data is then processed in the cloud and interactive graphs are produced, so multiple users can access up-to-date information in order to make optimized, timely farming decisions. The use of the WiField devices in a cotton farming operation is described, for scheduling irrigations and determining crop water use through the soil profile.
Soil moisture forecasting for irrigation recommendation
This study integrates measured soil moisture sensor data, a remotely sensed crop vegetation index, and weather data to train models, in order to predict future soil moisture. The study was carried out on a cotton farm, with wireless soil moisture monitoring equipment deployed across five plots. Lasso, Decision Tree, Random Forest and Support Vector Machine modeling methods were trialled. Random Forest models gave consistently good results (mean 7-day prediction error from 8.0 to 16.9 kPA except in one plot with malfunctioning sensors). Linear regression with two of the most important predictor variables was not as accurate, but allowed extraction of an interpretable model. The system was implemented in Google Cloud Platform and a model was trained continuously through the season. An online irrigation dashboard was created showing previous and forecast soil moisture conditions, along with weather and normalized difference vegetation index (NDVI). This was used to guide operators in advance of irrigation water needs. The methodology developed in this study could be used as part of a closed-loop sensing and irrigation automation system.