This project explores the effect of parameter uncertainty in models of crop production when crops are under simultaneous water and heat stress. Bayes Nets will be used to investigate linkages between parameters that might hide poor model performance.
Department: Agri-Environment
Supervised by: Lindsay Todman
Crop models are an important tools for predicting the likely effect of changes in climate and management practices on future food production yet there are uncertainties in model parameters and structures. This uncertainty may be particularly important when crops are experiencing multiple stresses simultaneously, as for practical reasons models are often developed and tested by considering the effect of each of these stresses separately. In this project the student will investigate the simulated crop response to combined water and heat stresses using the model AQUACROP. They will then consider the sensitivity of this simulated response to chosen parameter values, some of which are particularly uncertain. To explore this uncertainty in model parameters, the student will develop a Bayes net and use this statistical approach to understand how parameter values for different parts of the model (e.g. those for heat and drought stress) interact. This project will be linked with another UROP (if successful) which will investigate the interaction of water and heat stresses experimentally. Thus, at the end of the project, there will be an opportunity to compare the results modelled in this project with new experimental observations.
The student will: • Calibrate AQUACROP using data from a UK wheat field to identify multiple plausible parameter sets • Run the AQUACROP model to simulate heat and drought stresses • Develop a Bayes Net that represents the key processes in AQUACROP • Use the results from AQUACROP simulations to populate the probabilities in the Bayes Net and explore interlinkages between parameters and the effect of these parameters on the water and heat stress simulated in the model.
The student should either have (i) experience of programming and be keen to learn how this can be applied to understand crop production, or (ii) knowledge of agronomy and be keen to learn programming.
• programming (in MATLAB or another language if the student has a preference) • Bayesian methods • crop science • team work and collaboration • multidisciplinarity
School of Agriculture, Policy and Development (SAPD, Agriculture Building), Whiteknights Campus
210
Monday 09 July 2018 - Friday 17 August 2018
The post will be advertised centrally on the UROP website until Monday 30th April. Students should submit a CV and 1 page cover letter explaining your interest in the project to Dr Lindsay Todman (l.todman@reading.ac.uk). Applicants will subsequently be shortlisted for interview after the closing date.