Derivation of Optimal Cropping System for Dairy Farming: Constructing a Yield Map Using Machine Learning

2023 Fiscal Year Final Research Report

• Project/Area Number: 20K15607
• Research Category: Grant-in-Aid for Early-Career Scientists
• Allocation Type: Multi-year Fund
• Review Section: Basic Section 41010:Agricultural and food economics-related
• Research Institution: The University of Tokyo
• Principal Investigator: Sato Takeshi 東京大学, 大学院農学生命科学研究科(農学部), 准教授 (30756599)
• Project Period (FY): 2020-04-01 – 2024-03-31
• Keywords: 機械学習 / ハイパースペクトル / GIS / 計量経済分析 / 酪農 / 牧草地

Outline of Final Research Achievements

With the cooperation of a local agricultural support organization, a professional staff member acquired accurate vegetation survey data of the pasture, and detailed spectral reflectance characteristics of each grass species and season were captured using a hyperspectral camera. The results of machine learning analysis on the collected data indicated that the vegetation survey could achieve high classification accuracy of up to 97% in certain vegetation conditions. Furthermore, a simulation of machine work efficiency optimization demonstrated the potential for reducing work time by approximately 20%, and numerically illustrated that work time savings could be even more substantial through plot consolidation.Concurrently, econometric analysis was conducted on household-based managerial economic data of dairy farmers to examine labor and productivity.

Free Research Field

農業経済学

Academic Significance and Societal Importance of the Research Achievements

酪農業において安定的な自給飼料生産は極めて重要である．その中で，草地の管理は大きな課題となるが，植生調査や植生改善の効果の検証と経営データとの接合は容易ではない．本研究ではハイパースペクトルカメラやドローンを用い，機械学習を用いて植生の識別を行った．また，酪農の経営データと結びつけ，計量経済分析を行った．得られた成果を普及させることで，より簡便な圃場の植生の識別，圃場管理の方法が示される．それらの自給飼料生産・管理が酪農経営に与える影響を定量的に分析したことで，草地管理が酪農経営にとっても重要であることが示唆された．