As new AI techniques are developed and various types of big data accumulated, new approaches for pest management are also being attempted. Various spatio-temporal scale big data are being accumulated, and attempts are being made to utilize them to classify target objects and analyze their characteristics. Remote sensing data is widely used across various fields, and is being measured, stored, and shared in diverse formats. Hyperspectral imaging and satellite data are ecologically relevant big data, with distinct formats and potential applications. We will introduce real-world AI examples of utilizing hyperspectral image analysis, as well as estimating pest population density using satellite data.

Climate change has made outbreaks of insect-transmitted plant viruses increasingly unpredictable. Understanding spatio-temporal dynamics of insect vector migration can help forecast virus outbreaks, but the relationship is often poorly characterized. The incidence of Beet curly top virus (BCTV) was examined in 2,196 tomato fields in California from 2013-2022. In addition, we experimentally showed dispersal of the beet leafhopper, the only known vector of BCTV is negatively correlated with plant greenness, and we estimated spring migration timing using a vegetation greenness-based model. Potential environmental factors and spring migration time of beet leafhoppers were associated with BCTV incidence. We found BCTV incidence is strongly associated with spring migration timing rather than environmental factors themselves. In addition, the vegetation greenness-based model was able to accurately predict the severe BCTV outbreaks in 2013 and 2021 in California. The predictive model for spring migration time was implemented into a web-based mapping system, serving as a decision support tool for management purposes.