Decision support system for precision livestock: prediction of forage mass and nutritional attributes using machine learning algorithms

Abstract

Pressure on the agricultural system has increased with the continuing expansion of the human population. Moreover, new perspectives of agri-technology and precision farming have pushed to intensely use data to enhance agricultural productivity while minimizing environmental impact. Grasslands are globally important because of their extent, their role in regulating the global carbon, and is the major and the cheapest feed source for the livestock industry. Therefore, their regular monitoring is pivotal to ensuring efficient grassland management and the sustainability of pasture-based production systems. Remote-sensing-based techniques have been broadly used to monitor large areas and to capture the spatial variability of grasslands; however, most models were based on linear regressions, which were site-specific and did not have the capability to learn the highly non-linear and complex patterns in the data, which can be improved by data-driven algorithms. Thus, this project aims to assess and develop models for grassland biomass, crude protein, and fiber estimation using satellite spectral reflectance or vegetation indices, based on machine learning algorithms. For this study, a database from tropical and temperate forages will be gathered. Another important issue relies on an accurate assessment of energy expenditure (or heat production) of grass-fed animals that are a major component to estimate the energy that is effectively driven for production, e.g., milk secretion or retained empty body weight gain. Thus, the second aim of this project is to assess and estimate the dynamic of energy expenditure and methane emission of animals fed grass diets using mechanistic models and machine learning algorithms. We expect that this work will develop a reliable and cut-edge technology to assist farmers, rangeland managers, and stakeholders for pasture management decision-making. (AU)