Drone data assimilation in potato crop growth models

Introduction

As the fourth most important staple food crop in the world potato, aside from
being tasty, plays an important role in global food security. However, climate
change and stricter environmental regulations lead to increasing exposure of
potato crops to drought and nitrogen stress, causing yield losses. As the world’s
biggest exporter of seed potatoes, the Dutch potato breeding sector has an
important role in solving these challenges. The effects of drought and nitrogen
stress on potato, individually, have been researched in-depth and different varieties show varying stress tolerance. However, the effect of potato being exposed to both stress factors at the same time remains under-explored, and it is not clear whether the same stress tolerance mechanisms are effective against
combined drought and nitrogen stress. A deeper understanding of stress
tolerance mechanisms is essential to inform potato breeding for resilient
varieties and improve crop management.

Project description

The CropXR (extra resilient) institute is a is a public-private partnership dedicated to improve crop breeding and management. As part of CropXR, we collaborate with other universities and leading potato breeding, and processing companies to improve our understanding of drought and nitrogen stress tolerance mechanisms in potato.

Crop growth models are one tool used for understanding the complex interactions between the crop, environment, and crop management. E.g., by independently estimating the crops’ potential, water- and nutrient-limited yields or by separating the immediate stress response from its subsequent consequences. However, most crop growth models combine mechanistic and empirical approaches and require variety- and environment-specific calibration to accurately estimate development, growth, and yield. The labour-intensive destructive measurements needed for calibration limit the number of varieties for which calibration can be done.

Remote sensing, e.g., using drone-mounted cameras, can track plant development at high temporal resolution for a large number of plots and assimilating this data into crop growth models has been shown to improve yield prediction accuracy for various crops. However, to date, there is no comparison of data assimilation approaches for potato. Additionally, while remote sensing is a high-throughput data collection approach, any data collection requires resources and it remains unclear how the number of assimilated observations affect the models’ yield prediction accuracy.

The aim of this project is to: 1. Compare different data assimilation approaches for the WOFOST potato crop growth model; 2. Investigate the effect of temporal resolution and flight planning on WOFOST’s yield prediction accuracy; 3. Calibrate WOFOST for a large population of potato varieties; and 4. Identify potato crop traits which support drought and nitrogen stress resilience using the calibrated WOFOST models.