The WGEN, SIMMETEO, and LARs-WC stochastic weather generators are widely used in a range of studies, including agricultural crop simulations to determine the potential impact on crop growth, development, and yield. To assess the potential and limitations of weather generators, their climate simulation performance should be evaluated when applied over a new region, or linked with crop models. The Asia-Pacific Economic Cooperation Climate Center (APCC) recently developed a 6-month lead multi-model ensemble (MME) prediction system. Weather generators could link APCC seasonal climate predictions to crop models to generate daily weather data and provide real-time simulation of crop growth and assessment of crop productivity. The objectives of this study are to evaluate the three generators for 63 weather station sites over south Korea, and determine the potential for translating APCC’s seasonal climate outlook into predictions of rice yield using the CEREs-Rice model.

The study concludes that the performances of the generators are accurate in simulating maximum and minimum mean temperatures, and solar radiation; however, they have difficulty in reproducing precipitation. The performances vary from site to site. The LARs-WG matches the observed data closer than the WGEN and sIMMETEo. In general, rice yields simulated using synthetic weather data generated with and without bias correction are similar to the yields stimulated using observed weather date. simulated rice yields of different cultivar responses to weather data are different. Unkwang is less sensitive to variation of weather factors than Nampyeong and saegyewha. The APCC climate outlook has the potential to forecast rice yields over Korea. however, the accuracy and reliability of yield forecasting may vary considerably from year to year and site to site.

1. INTRODUCTION

1.1 Paddy Rice in Korea

1.2 Rice modeling and decision making

1.3 Weather and crop modeling

1.4 APCC climate outlook

1.5 Downscaling

1.5.1 statistical downscaling

1.5.2 Weather generators

1.6 Bias corrections

1.7 Applications of weather generators

1.8 objectives

2. MATERIALS AND METHODS

2.1 Weather Generators

2.1.1 Markov chain model

2.1.2 Alternating renewal process

2.1.3 Daily precipitation amount

2.1.4 Daily maximum and minimum temperature

2.2 Weather generators

2.2.1 WGEN

2.2.2 sIMMETEo

2.2.3 LARs-WG

2.4 CEREs-Rice Model

2.4.1 Growth stages

2.4.2 Genetic Coefficients

2.4.3 Rice Yield

2.5 Input Data

2.5.1 spatial downscaled weather data

2.5.2 observed weather data

2.5.3 Parameters estimations in LsRs-WG

2.5.4 The soil data

2.5.5 Cultivar genetic coefficients

2.5.6 Management practices

2.6 Generation of synthetic weather data

3. RESULTS

3.1 Rice yield simulated using observed weather data

3.2 Quality of unconditional weather series

3.2.1 Precipitation amount

3.2.2 Minimum temperature

3.2.3 Maximum temperature

3.2.4 Daily maximum temperature

3.2.5 Radiation

3.3 Rice yield