Precision agriculture is an emerging area at the intersection of engineering and agriculture, with the goal of intelligently managing crops at a microscale to maximize yield while minimizing necessary resource. Achieving these goals requires sensors and systems with predictive models to constantly monitor crop and environment status. Large datasets from various sensors are critical in developing predictive models which can optimally manage necessary resources. Initial experiments at University of St. Thomas (UST) greenhouse have demonstrated the feasibility of sensor system for use in Precision Agriculture. Hundreds of high-value vegetables such as lettuce, beans, peas, onions, spinach and bok-choy were planted in different types of soils and nutrients. Soil moisture and pH level of each plant were monitored and collected by soil sensors. A weather station was installed to collect the air temperature, air moisture, light, and wind information in the greenhouse. A Photosynthetically Active Radiation (PAR) sensor was also installed to monitor how the plant grows responding to different wavelengths. A multi-spectral camera was also used to observe the Near-Infrared Reflection (NIR) from each plant. This information reveals the amount of photosynthesis occurring in each plant, providing an important indicator of plant’s health. All information collected was time stamped such that different sensor information could be correlated. This information was then fed back to a controller to release water and nutrition to different plant groups at different times in order to meet different growth patterns.

During the 50-day growth period, over one billion data points were generated from 6 different plant types. In order to handle and process such big data. Cloudera Hadoop Cluster and software modules to process such data is being developed. Using the information collected from our infrastructure, more sophisticated models can be developed enabling more sustainable farming.