Building an early prediction method of US crop yields based on machine learning algorithm

2020 Fiscal Year Final Research Report

• Project/Area Number: 17K08037
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Agricultural environmental engineering/Agricultural information engineering
• Research Institution: National Agriculture and Food Research Organization
• Principal Investigator: Sakamoto Toshihiro 国立研究開発法人農業・食品産業技術総合研究機構, 農業環境変動研究センター, 上級研究員 (20354053)
• Project Period (FY): 2017-04-01 – 2021-03-31
• Keywords: 機械学習 / 食料安全保障 / トウモロコシ / 大豆 / 小麦 / フェノロジー / 作柄予測 / 作付分類

Outline of Final Research Achievements

This study aimed to establish early crop yield prediction technology for U.S. crops. Firstly, the versatility of crop phenology detection method was improved to be applicable to 36 growth stages of 8 crops. Secondly, the early crop yield prediction method was improved by considering weather and environmental conditions. Then, the prediction accuracy of corn and soybean yield was improved. Finally, the early crop classification method was modified to enable estimation of crop coverage ratio within a MODIS pixel. Then, the crop classification accuracy was improved. Consequently, a new crop yield prediction method was developed in terms of using machine learning algorithm based on the combined use of high-frequency observation satellite data (MODIS) and meteorological environmental data.

Free Research Field

リモートセンシング

Academic Significance and Societal Importance of the Research Achievements

日本は、輸入トウモロコシ・大豆の約7割、輸入小麦の約5割を米国からの輸入に依存している。また、世界的な食料需給情勢の不安定化を背景に、国際的な政策協調として、世界の農業・食料市場に関する正確かつ透明な情報を取得するための衛星リモートセンシング技術を用いた監視ネットワークの構築が推進されている。本研究成果は、作柄早期予測を確立するための基盤的な知見を提供するとともに、国内外の食料安全保障に資する技術としても活用が期待される。