Date of Award

2021

Document Type

Doctoral Thesis

Degree Name

Doctor of Philosophy

Department

Process, Energy & Transport Engineering

First Advisor

Dr Michael D Murphy

Abstract

Maximising pasture utilisation through optimal grassland management is vital in terms of ensuring the economic sustainability and mitigating the environmental impact of both Irish and global pasture-based livestock production. Frequent and accurate measurement of grass quantity and quality is one of the main methods of maximising grass production and utilisation on pasture-based farms. The most prominent grass measurement tool used on Irish pasture is the rising plate meter (RPM), however, there are a range of issues with this method including operator bias, precision and difficulties in accounting for spatial variation. Currently, there is no definitive protocol for grass measurement which farmers could follow to objectively measure their pastures and to account for the variation of grass growth within paddocks. Furthermore, there is no rapid method of estimating the quality of grass within pasture. This body of work aims to optimise state of the art grass measurement technologies through field experimentation and numerical simulation. The first objective of this thesis was to determine the variation in herbage mass (HM) within grazed swards and evaluate the precision of the RPM for measuring HM. Intensive compressed sward height (CSH) measurements and HM reference cuts were carried out on trial plots and grazed paddocks at Moorepark, Ireland over two grazing seasons. Retrospective analysis simulations were performed in order to calculate the effect of various reduced measurement resolutions on estimated mean CSH error. A repeated measurement analysis was performed on grass samples to determine RPM system error. The value of HM within swards was found to vary on average by 36%. Factors that affected sward heterogeneity were identified as nitrogen fertilisation, clover content, morphology, seasonality and grazing effects. Mean CSH could be estimated to within 5% Abstract xxii error using the RPM by recording 24 measurements ha-1 in a random stratified manner. Recording ≥ 40 RPM measurements ha-1 resulted in a diminishing rate of returns in terms of reducing mean CSH estimation error. Measurement system error for the RPM, in terms of the standard deviation of measurement repeatability, was calculated to be 4.34 mm. The second objective was to optimise the HM prediction accuracy of the RPM by investigating the utilisation of additional grassland management and meteorological data using multiple linear regression in combination with backward sequential variable selection, all subsets regression and k-fold validation. Monthly regression models were created from a selection of 17 variables with data collected over two grazing seasons. Reductions in HM prediction error of 20 – 33% (root mean squared error) were achieved in comparison to state of the art models used on Irish pastures. The inclusion of meteorological variables slightly improved HM prediction performance, however, this improvement was not sufficient to warrant the investment in on-farm meteorological sensors. The optimum model utilised variables CSH, N fertilisation rate and grazing rotation number and is suitable for integration with grassland decision support software currently used on Irish farms. The third objective was to develop a near infrared spectroscopy (NIRS) calibration to more rapidly predict the quality of fresh grass in terms of dry matter (DM) and crude protein (CP) content. Perennial ryegrass samples (n = 1,812) were collected over three grazing seasons and scanned using a FOSS 6500 spectrometer to develop NIRS calibration and validation datasets. Reference wet chemistry analysis was carried out for DM and CP and the resultant data were calibrated against spectral data by means of modified partial least squares regression. The ratio of percentage deviation (RPD) was used to rank the developed NIRS calibrations, which were sufficiently precise to replace laboratory oven drying methods for DM (RPD =2.63) and were capable of categorising the quality of pasture in terms of CP (RPD = 2.37). These calibrations have the potential to reduce laboratory costs, streamline herbage Abstract xxiii quality analysis and may be used for benchmarking future grass sensing technologies. The fourth objective was to develop a grass measurement optimisation tool (GMOT) to accurately and efficiently measure grazed pastures. The developed prototype was a Visual Basic Application for MS Excel decision support tool that generated interactive paddock measurement guide maps. Farmers could use the GMOT to create optimised protocols for measuring their pastures in a random stratified manner based on GPS co-ordinates, resulting in accurate non-biased estimations of sward parameters. Measurement routes were optimised using a genetic algorithm based on a traveling salesman problem. Actual survey error was estimated using Monte Carlo simulations that combined measurement and calibration error distributions generated from data collected as part of objectives 2 and 3. Actual error was estimated to 28.1% (relative prediction error) for the RPM and the optimum measurement rate that minimised both cost and error was 8 measurements ha-1. The GMOT was developed to generate objective spatially balanced and geo-tagged grass measurement protocols that could be used for a range of pasture measurement methods, including quality analysis. The main research outputs from this study were: 1) state of the art measurement technologies for fresh grass quantity and quality have been optimised in terms of accuracy 2) the GMOT has been developed to generate accurate, efficient and objective grass measurement protocols that account for within pasture variations 3) grass measurement benchmarks have been set in terms of both accuracy and methodology. Outputs from this thesis will benefit grassland farmers by increasing the precision and efficiency of both pasture quantity and quality measurement, which will subsequently increase grass utilisation and reduce whole farm feed and fertilisation inputs. This body Abstract xxiv of work will further facilitate the development of future precision agricultural technologies for pasture-based livestock production.

Creative Commons License

Creative Commons Attribution 4.0 International License
This work is licensed under a Creative Commons Attribution 4.0 International License.

Access Level

info:eu-repo/semantics/openAccess

Project Identifier

info:eu-repo/grantAgreement/EU/H2020/15ICTAGRI_1/IE/ICT technologies for sustainable food systems/ERA-NET, ICT—AGRI

Download

Share

COinS