Analysis of crop disease and pest occurrences: Insights from Japan’s national surveys

Abstract

Accurate forecasting of crop diseases and pests (CDPs) is crucial for ensuring food security. In Japan, nationwide CDP field surveys have been conducted for over half a century at an annual cost of approximately 300 million JPY, with the primary goals of predicting CDP outbreaks and estimating within-year damages. Despite the magnitude of these efforts, the collected data remain underutilized. Therefore, this study aimed to contribute to the advancement of this field by evaluating the potential of leveraging historical Japanese CDP survey data to forecast CDP occurrences. Through comprehensive analysis and statistical modeling, we found that a simple algorithm—averaging data from the past five years without incorporating seasonal trends or meteorological variables—outperformed more complex models, underscoring the value of historical CDP survey data. However, the prediction error remained substantial, with an RMSE of 6.2 ± 19.6. Notably, as 70.1% of the CDP survey data recorded values of five or less, an error of 6.2 indicates poor predictive accuracy in most cases. Given the challenges of precise forecasting, the high cost of nationwide surveys, and Japan’s lowest self-sufficiency rate, fundamental reforms are needed. Integrating modern technologies, including IoT/ICT and artificial intelligence, could enhance the sustainability of CDP survey, ultimately safeguarding food security.