HOPPER - AN EXPERT-SYSTEM FOR FORECASTING THE RISK OF WHITE-BACKED PLANTHOPPER ATTACK IN THE 1ST CROP SEASON IN CHINA

This paper describes an expert system - HOPPER - designed to provide f orecast information and recommendations for the white-backed planthopp er (WBPH) Sogatella furcifara (Horvath), a major migratory rice pest i n China. The threat of WBPH attack to the early rice crop varies from year to year, depending on many factors, including immigration level, weather conditions in the winter and during immigration, and rice vari ety, plant vigour, natural enemies and local weather conditions that a ffect the growth rate of the population after immigrants have arrived. The paper describes the three steps involved in constructing the expe rt system: (1) determining the need for the system; (2) specifying the structure of the expert system that will meet this need, and (3) inco rporating the knowledge base and rules into computer software. A preli minary test is described, showing that HOPPER was able to make a reaso nable prediction of white-backed planthopper attack in China in the 19 91 season. The future role of HOPPER for practical problem solving, tr aining, and identifying research and surveillance needs is discussed.