The Forecasting Methods and Systems for Rice Blast Disease

Forecasting is a major component of decision making in the rice blast disease management system. For rice blast disease management various methods for forecasting disease occurrence have been proposed, including methods based on the knowledge of type of blast occurrence , of behavior of blast fungus,of growth phase and chemical components of rice plant , of analyses of weather patterns , and of relation between weather conditions and outbreak of disease etc. All these forecasting methods proposed for blast disease of rice can be categorized into three forms : empirical rules and holistic (statistical) models and systems analytic (simulation) models. Empirical rules is usually based on the assumption of coincidence between the meteorological conditions and disease occurrence. Statistical models for forecasting blast disease of rice can be divided into two types, e.g. simple regression models and multiple regression models. Simple regression models provide estimates of disease intensity with any climatic factor, such as temperature, rainfall, duration of sunshine etc. at a given time period. Multiple regression modele estimate disease intensity with many climatic factors at different time periods. In recent years, the efforts to develop simulation models for rice blast disease were made by a group of Japanese researchers, such as a model smulating the rice leaf blast called BLASTL. In Taiwan, two main types of forecasting models, i.e., the long-range and short-range forecasting models, have been developed. The long-range forecasting models, relating the disease prevalence to winter climate, are primarily developed to estimate the probable maximum intensity of the epidemics. The short-range forecasting models, relating the initial occurrence and elements of disease progress curve to the weather conditions from time of transplanting until active tillering stage, are used to estimate the amount and time of disease occurrence. The short-range forecasting models are served as tactical models, they can be used for negative forecast and made available for the on-site forecasting system (electronic rice blast alarm). Models for forecasting disease prevalence can be used for generating the long-range forecast as an aid in planning of management strategy. Both types of model are integral parts of the forecasting system of rice blast disease, although these methods are different, they should not necessarily be regarded as mutually exclusive, they should be regarded as mutually complementary. In the past decade, the use of computers has led to develop various types of computer-based systems in Japan, to assist in data filing and processing, and in modelling. Most of which consited three independent subsystems, i.e. the system for processing data from fixed plots, the system for processing data from patrolling fields, and the system for processing data from meteorological observations at Agricultural Experiment Station. Moreover, a nation-wide new on-line system for forecast informations has been established and made avalable for disease forecasting and warning service