Pasture API: A digital platform to support grazing management for southern Australia

Abstract

Extensive ruminant livestock production is complex, and it is difficult and time-consuming to obtain quantitative information for decision making (e.g. biomass and quality of forage). Increasingly data from historic records, seasonal forecasts, or near real-time data from remote or on-farm sensors. However, there is limited capacity to compile, integrate and analyse these data. In some cases, there are decision support tools available with such analytical capability, but the time required to learn and apply these to gain the benefits is significant, and a major disincentive to their widespread adoption. Based on the need for simple, entry level, decision support for the livestock industry we have developed the Pasture API (Application Programming Interface) platform. The aim of the Pasture API was to build a seasonal pasture forecast system that can provide a forecast to specific locations up to 6 months into the future for anywhere in Australia and made available to a variety of client software packages. The platform is able to forecast a wide range of simulated data outputs including pasture biomass, ground cover, supplementary feeding and livestock growth. The Pasture API application required the integration of a number of both new and existing analytical capabilities, which are summarised below and described in more detail in this paper. 1)Ruminant grazing simulation engine: The GrazPlan biophysical pasture and ruminant nutritionmodel (adapted from GrassGroTM software) was repurposed for use as the modelling engine in a back-end service infrastructure. 2)Flexible tactical grazing scenarios: The GrazPlan models were incorporated into a new softwareapplication for batch processing of tactical grazing scenarios. This application was called GGTactical. 3)Dynamic platform for connecting the pasture simulation engine with data streams: We usedCSIRO's Senaps platform to connect the GGTactical application with a range of spatiotemporal datastreams so that simulation scenario workflows could be implemented. 4)Demonstration interface: A demonstration website (Pasture Tracker) was built to interact with thePasture API application and implement workflow's based on location and livestock enterprise details.Currently the software is hosted internally by CSIRO, with the intention that a version becomepublicly available in the near future. We demonstrated that Pasture API is able to replicate simulation of a livestock grazing scenario, as can be done with more complex modelling software, such as GrassGroTM. Key production metrics such as net primary productivity (NPP) of pasture, supplementary feeding, ground cover and liveweight of stock were charted. These were reported both as historic percentile values across a season, the now-cast (current) value, and a probabilistic forecast for a predefined period of time (e.g. 3 or 6 months). To compare the effects on forecasts for various input data streams that were available, a sensitivity analyses was conducted. This provided information about the suitability of more generic data streams (e.g. national soils database) for forecasting, comparing forecast outcomes for those where local data were available. The Pasture API project demonstrates the ability to create easy to use, yet powerful, decision support systems for the livestock industry. This is a novel integrating technology that we expect to continue to develop to make use of the many sources of sensor and archive data that are collected within livestock businesses. This information is expected to increase the precision across a range of interventions, including; stocking density, timing of paddock rotations, and supplementary feeding. In the future we expect to increase the use of local data streams and refine the data delivery processes to produce the site-specific information sought by the industry for decision making.