Project

Assimilating potato crop growth models with drone data

Dutch potato crops will increasingly face drought and nitrogen stress and it is important to understand how different potato varieties respond to these stresses. Crop growth models can be used as a tool to understand these responses but are often inaccurate. Drone data can be used to improve the accuracy of crop growth model predictions but it is unclear which data assimilation method is most suitable for potato crop growth models.

Project description

Due to the effects of climate change and the environmental issues stemming from nitrogen emissions, Dutch potato crops will experience increasing drought and nitrogen stress in the future. Drought and nitrogen stress both limit potato yield but not all potato varieties are equally affected. Some varieties possess traits which make it more resilient against either drought or nitrogen stress, and enabling it to maintain a higher yield than susceptible varieties. The effects of drought and nitrogen stress, individually, have been researched in-depth. Understanding how different potato varieties respond to stress and which crop traits support stress resilience informs potato breeding for resilient varieties, and crop management.

Crop growth models can be a tool to understand plant development and their interaction with the environment. E.g., yield gap analysis, using crop growth model predictions, can be used to determine the potential yield based on the environmental conditions and identify the causes of a lower actual yield. However, the accuracy of crop growth models’ prediction relies on careful calibration of genotype-specific input parameters. Additionally, this calibration often leads to inaccurate predictions in environments which are different from the one the calibration is based on. Recently, drones are becoming increasingly common for measuring crop development (e.g., ground cover or leaf area index) and they are able to capture very quickly, allowing for a high measurement frequency, and
spatial resolution. There are several methods for using drone data to correct
for these calibration errors, and improve biomass and yield prediction, in a
process called data assimilation. The suitability of different data assimilation
approaches have been explored for, e.g., sugarcane. However, potato has a
markedly different development compared to sugarcane and it is unclear how transferable this is to potato crop growth models.

Objectives and methods

Objective

This project will evaluate different approaches of assimilating drone data into potato crop growth models.

Field experiments

We have two large-scale field experiments in Friesland and Zeeland with 20 different potato varieties grown under different combinations of irrigation schedules and nitrogen fertilization.

Drones

Data will be collected on a weekly basis with a multispectral and thermal drone. There is the option to collect your own data by flying the drones (we can help you with getting the necessary drone license and training). The processing of the drone images is outsourced to a specialized company and the in-house data analysis conducted on numeric data.

Data analysis

Appropriate methods are used to ensure the model input data is suitable, determining which variables should be used for assimilation, and to evaluate the data assimilation methods.

Required skills

Basic knowledge of statistics, R/Python programming, fieldwork, and crop physiology is advantageous.

Ideally, you have completed at least one of these courses:

CSA30806 Research Methods in Crop Science
CSA30306 Advanced Crop Physiology
CSA32806 Modelling Functional Diversity in Crop Systems
CSA34806 Advanced Agronomy
Or any course on time series analysis and crop modelling

Types of research/work

This project includes both fieldwork and data analysis.

Period

The field experiments are running from May to October. It is expected that the thesis student will be available to occasionally help with fieldwork during the thesis. The exact starting date is negotiable.