Digital tools for sustainable ranching on Southwestern US rangelands

TitleDigital tools for sustainable ranching on Southwestern US rangelands
Publication TypeConference Proceedings
Year of Publication2022
AuthorsUtsumi S.A, Nyamuryekung'e S., McIntosh MM, Cibils AF, Estell RE, Spiegal S., Duff G., Cao H, Boucheron LE, Chen H., Le T., Winkler Z., Rahman Z., Gong Q, Cox A, Gifford C, Ragosta JP, Krohn M., Gouvea V., Brandani C, Waterhouse T, Holland JP, Elias EH, Aney S, Bestelmeyer BT, Steiner J
Conference Name2022 Society for Range Management Annual Meeting
Date Published2/8/2022
ARIS Log Number392986
Keywordsdigital tools, rangelands, southwestern US, sustainable ranching

The Sustainable Southwest Beef Project is partnering with ranchers and stakeholders to develop a Digital Ranching Platform aimed at informing decision making of ranch-level management tasks and climate -resilient livestock systems. This aspirational management approach fuses traditional statistical and advanced data science with visualization dashboards to inform key indicators of animal welfare and ranch resources. The system collects large streams of real-time data, which is logged and transmitted through a network of high throughput sensors, gateways and cloud computing services. Internet of Things infrastructure includes field sensors and high throughput GPS sensors, and accelerometers mounted on animals, operating on a Long Range Wide Area Network (LoRaWAN) solar or grid powered and Ethernet, WiFi backhaul, or GSM communication. Software engineering and IT project components are focused on unifying a web-based dashboard and server application for visualization and retrieval of computed data and configuration of field devices and sensors. Utilities include improvements of operational efficiencies through near- to real-time tracking and scouting of livestock, rapid animal welfare assessments, remote monitoring of rain gauge tipping buckets and tracking of water level in cattle drinking troughs. Procedures seek to facilitate the harmonization (i.e. common feature representation) and curation of varying streams of data (i.e. erroneous sensor data and missing data points) prior to implementing machine learning for advance classification and prediction objectives. Future analytics will aim to merge patterns of animal data with insights on animal activity budgets, early warnings of breeding status, and faulty animal health or grazing performance. Unobtrusive scoring of cattle body condition is being collected using machine learning classifiers via high throughput video imagery of infrared depth cameras. Pilot case studies suggested important utilities of the system and have reveled several areas for improvement of existing sensor and network infrastructure and applications, which will be discussed in this symposium.