Scaling Sensor Networks for Scheduling Irrigations in a Commercial Pot-in-Pot Nursery - Robotics Institute Carnegie Mellon University

Scaling Sensor Networks for Scheduling Irrigations in a Commercial Pot-in-Pot Nursery

Bruk E. Belayneh, David Kohanbash, and John D. Lea-Cox
Conference Paper, Proceedings of 111th American Society for Horticultural Science Annual Conference (ASHS '14), July, 2014

Abstract

Wireless sensor networks (WSNs) can be used as monitoring systems in nurseries and provide direct feedback on tree water needs and environmental conditions, thereby adding value to existing irrigation systems. More advanced WSNs can make irrigation control decisions based on real-time root zone-soil moisture conditions, and have been shown to significantly reduce water and nutrient leaching from nursery tree pot-in-pot production. However, deployment of WSNs in large operations can be costly if done without considering scaling-up strategies for both monitoring and control purposes. In 2013, we implemented a scaling strategy at a large pot-in-pot nursery using six indicator species representing different functional water use groups: low (dogwood and crepe myrtle), medium (hornbeam and red oak) and high (river birch and red maple). Two rows of each species (n=10 per row) were installed in a small block on the nursery. The volumetric water content (VWC) of five trees in each row was sensed using 10HS soil moisture sensors (Decagon Devices, Inc.). One row of trees from each species were irrigated using grower-scheduled cyclic irrigations; the second rows of trees were independently irrigated using VWC set-point strategy. Badger flow meters (Badger Meters, Inc.) were installed at the beginning of each row to measure irrigation amounts. Runoff volume and leachate electrical conductivity for each row were measured, respectively, by ECRN-100 rain gauge and ES-2 sensor (Decagon Devices, Inc.). All sensor readings were logged by a combination of Em50R (monitoring) and nR5-DC (control) data loggers on 15-minute basis and relayed to a base-station in the farm office. Sensorweb software (Carnegie Mellon University), with nR5 nodes, was utilized to schedule micro-pulse irrigation events to the sensor-controlled rows based on VWC set-points and running averages of five sensors in each row. For the study period from March to November 2013, total irrigation water applications were computed and compared. For all species, the sensor-controlled irrigation strategy used less water compared to the grower-scheduled irrigation. Irrigation water savings were 56.6% (dogwood), 70.8% (crepe myrtle), 49.1% (hornbeam), 45.0% (red oak), 16.9% (river birch) and 10.3% (red maple). Water savings were highest for the low water use species and lowest for the high water use species. For all tree species, there were no significant differences in tree growth rates (based on tree caliper measurements taken at 15 cm height) between the two irrigation strategies.

BibTeX

@conference{Belayneh-2014-122510,
author = {Bruk E. Belayneh and David Kohanbash and John D. Lea-Cox},
title = {Scaling Sensor Networks for Scheduling Irrigations in a Commercial Pot-in-Pot Nursery},
booktitle = {Proceedings of 111th American Society for Horticultural Science Annual Conference (ASHS '14)},
year = {2014},
month = {July},
}