2026-04-21 東京大学
バレイショの収量を予測するパイプライン
＜関連情報＞
- https://www.a.u-tokyo.ac.jp/topics/topics_20260421-2.html
- https://acsess.onlinelibrary.wiley.com/doi/10.1002/ppj2.70078
無人航空機を用いたジャガイモの収量予測のための時空間表現型解析と成長モデリング Unmanned aerial vehicle–based spatiotemporal phenotyping and growth modeling for forecasting potato yield
Yuto Imachi, Kunihiro Tanaka, Minato Miyauchi, Kyosuke Miyamoto, Masahiro Okada, Pieter M. Blok, Wei Guo, Hiroyoshi Iwata
The Plant Phenome Journal Published: 20 April 2026
DOI:https://doi.org/10.1002/ppj2.70078
Abstract
Monitoring spatial variations in plant growth and forecasting yield before harvest provides valuable insights for optimizing agronomic decision-making in potato (Solanum tuberosum L.) cultivation. Although unmanned aerial vehicle (UAV)-based remote sensing has recently enabled the development of tuber fresh weight (TW) estimation models, their integration into practical yield-forecasting systems remains limited. In this study, we developed machine learning models to estimate tuber weights at multiple preharvest time points using RGB and multispectral UAV imagery. Image-derived features were extracted from the orthomosaic and digital surface model images for each plot, and a random forest regression model was trained for TW estimation. The estimated values were subsequently used to fit the Gompertz growth curves, which were then used to forecast the yield at the expected harvest time. The correlation between the estimated and observed values was strong in the UAV-based TW estimation, with correlation coefficients exceeding 0.8 and coefficients of determination (R2) above 0.6 at all time points. Yield forecasts based on fitted growth curves achieved a correlation of 0.78 and an R2 of −0.17 in 2023 and 0.70 and an R2 of 0.47 in 2024. These results demonstrate that UAV-based sampling combined with machine learning is a feasible approach for monitoring spatiotemporal variations in tuber growth and forecasting potato yield at the plot level prior to harvest.
Plain Language Summary
Potato yield is difficult to predict before harvest because tubers grow underground. In this study, we developed a two-step approach to forecast yield. First, we used drone images in visible and multispectral light to monitor plant growth and estimate tuber weight during the growing season. Machine learning models linked the image data to measured tuber weights. Second, we fitted S-shaped growth curves to the estimated values to predict the final yield. The results showed that drone-based estimates closely matched field measurements, allowing reliable yield prediction in different years. This approach combines drone remote sensing, machine learning, and growth curve modeling into a practical tool for precision agriculture. It can help farmers make better decisions about harvest timing and crop management to improve potato production efficiency.
