Crop productivity is largely affected by meteorological conditions, and therefore, accurate weather forecasts are very important in crop yield prediction. Especially, the ability to predict long-term weather conditions for range has a significant influence on the estimation of the yields. In South Korea, short-term detailed forecasts are only available up to 10 days, and the forecast beyond that time period is delivered in the form of brief information about how the temperature and precipitation will be compared to the normal year. However, this weather “outlook” cannot give practical information for farmers nor for decision makers and policy makers. With the increasing needs for the subseasonal-to-seasonal (S2S) and seasonal prediction, many research efforts have been recently made to...[ Read more ]

Crop productivity is largely affected by meteorological conditions, and therefore, accurate weather forecasts are very important in crop yield prediction. Especially, the ability to predict long-term weather conditions for range has a significant influence on the estimation of the yields. In South Korea, short-term detailed forecasts are only available up to 10 days, and the forecast beyond that time period is delivered in the form of brief information about how the temperature and precipitation will be compared to the normal year. However, this weather “outlook” cannot give practical information for farmers nor for decision makers and policy makers. With the increasing needs for the subseasonal-to-seasonal (S2S) and seasonal prediction, many research efforts have been recently made to improve its capability through various methods such as operational models and machine learning techniques. Dynamical downscaling is one of the methods to improve the accuracy of seasonal forecasts by allowing very high-resolution climate simulations with much realistic surface boundary conditions. The added value of dynamical downscaling in the simulation of weather events and climatology has been widely researched and well appreciated, but its added value in terms of the application of the downscaled results has not yet been intensively evaluated. In this regard, in this dissertation, the original GCM data and its dynamically downscaled simulations over South Korea are used to develop rice yield prediction models, and their results are compared. It is shown that the downscaled results improve the performance of the prediction models, which demonstrates that dynamical downscaling has the potential to enhance the value of meteorological information by benefitting the end-users in the agricultural field.