In the heart of South Africa, a groundbreaking study is revolutionizing the way we understand and cultivate sweet potatoes. Philemon Tsele, a researcher from the Department of Geography, Geoinformatics and Meteorology at the University of Pretoria, has developed a novel approach to phenotyping sweet potato varieties using unmanned aerial vehicles (UAVs) and advanced data analysis techniques. This research, published in the journal ‘Smart Agricultural Technology’ (translated from Afrikaans as ‘Intelligente Landbou Tegnologie’), promises to reshape sweet potato breeding programs and has far-reaching implications for the energy sector.
The leaf area index (LAI) is a crucial metric for assessing plant health, growth, and productivity. By accurately estimating LAI, farmers and breeders can make informed decisions about crop management and selection. Tsele’s study integrates radiative transfer models (RTMs), active learning algorithms, and non-parametric regression methods to retrieve LAI data from UAV multispectral imagery. This hybrid approach offers unprecedented accuracy and efficiency in phenotyping sweet potato varieties.
Tsele and his team tested two machine learning algorithms—boosted regression trees (BRT) and kernel ridge regression (KRR)—to invert the PROSAIL RTM, a widely used model for simulating canopy reflectance. The goal was to retrieve LAI data across 20 different sweet potato varieties during their peak growth stage. The results were impressive: the most accurate LAI retrieval was achieved by combining smaller PROSAIL simulations with random sampling active learning and KRR methods, yielding a coefficient of determination (R2) of 0.52 and a root mean squared error (RMSE) of 0.88 m2.m-2.
However, the BRT algorithm showed even greater potential. “While KRR provided high accuracy, BRT captured more spatial variability of observed LAI,” Tsele explained. “This means BRT can offer better prediction accuracy across diverse sweet potato varieties, which is crucial for large-scale breeding programs.”
The implications of this research are vast. Accurate phenotyping of LAI dynamics can significantly enhance sweet potato breeding programs, leading to the development of new cultivars with improved yield and resilience. This is particularly important in South Africa, where sweet potatoes are a staple crop and a vital source of income for many farmers.
But the benefits extend beyond agriculture. Sweet potatoes are increasingly being recognized as a valuable source of bioenergy. Improved breeding programs can lead to the development of sweet potato varieties with higher biomass yield, making them more suitable for biofuel production. This could help meet the growing demand for renewable energy and reduce dependence on fossil fuels.
Tsele’s research, published in ‘Intelligente Landbou Tegnologie’, represents a significant step forward in the application of UAV technology and advanced data analysis in agriculture. As we look to the future, this hybrid approach to phenotyping could become a standard tool in the toolkit of farmers, breeders, and energy sector professionals alike. By harnessing the power of data and technology, we can cultivate a more sustainable and productive future for sweet potatoes and the industries that depend on them.