Estimation of daily primary inoculum of rice blast disease based on weather data

Rice blast is an endemic disease in Korea. Its outbreak depends on the favorableness of the weather conditions for the disease. Recent epidemics had occurred in 1976, 1977, 1980 and 1993. Due to the climate changes, favorable weather conditions are more likely to occur in the future. Therefore the weather conditions for rice blast outbreak should be monitored on a yearly basis. Most rice blast monitoring models incorporate daily spore observation to estimate infection warnings. Daily microscopic observation is, however, a time-consuming process and only applicable to small areas. To extend the effective area of such models, direct observation process need to be substituted by a numerical model, which estimate daily spores based on weather data. Primary inoculum is a cohort of the very first spores of the growing season, which initiates disease cycle of the year. Because not all of the primary inoculum can cause infection, daily existence and amount of the inoculum are very important in estimating daily probability of disease onset. In this study, weather conditions for spore dispersion as well as spore formation were analyzed to estimate daily existence of primary inoculum. Daily observed number of spores and hourly weather data from Hwaseong, Korea during 1994-1999 were used for model development. Daily spores were captured by a rotor type spore sampler during 01:00 – 02:00. Daily inolulum potential (IP) for spore formation was calculated by hourly accumulating the reciprocal of maturation hours for spore-bearing part, which were initially estimated by hourly temperature and relative humidity. Daily IP, air temperature, relative humidity, solar radiation, rainfall, and wind speed were lagged by 1 to 7 days as the independent variables in the following analyses. Discriminant analysis was performed to estimate daily existence of spores. Regression analysis was used to estimate daily number of spores. The stepwise procedure of SAS was used to select significant independent variables for both analyses. The models were validated using a separate dataset from the modeling dataset observed from 10 locations including Hwaseong during 1998-2001. Daily existence of the spores was correctly estimated by 87% and 70% of the times when spores were observed and not observed, respectively. The number of spores was more difficult to estimate: the coefficient of determination for the validation of the regression model was 0.11. These models can be incorporated in initial disease forecasting over wide area based on numerical weather forecasts. Daily number of spores, however, needs more investigation on the usefulness of the model.