This study investigates the performance of a dynamical downscaling system based on the Weather Research and Forecasting (WRF) model in simulating the regional climate over the Indonesian Maritime Continent. The domain covers the whole region of the Indonesian Maritime Continent at a 45 km horizontal resolution and the simulation spans three months (June-July-August, JJA) for every year in a 20-year period (1989-2008). The initial and lateral boundary conditions to drive the WRF modeling system are provided using the Community Climate System Model (CCSM3), which APCC has improved the seasonal predictability of, through SST initialization using GODAS (Global Ocean Data Assimilation System) data.

To assess the dynamically downscaled results, the seasonal mean field and interannual variability of APCC/CCSM3 and the WRF results were compared to MERRA reanalysis data. The comparison between the observations and model simulations shows that the WRF can improve the ability of regional climate simulation. The regional details about the temperature produced by WRF are much better than those produced by the APCC/CCSM3. The 20-year mean spatial distribution of daily precipitation from APCC/CCSM3 and WRF showed a dry bias compared to MERRA data. However, the pattern correlation coefficient of dynamically downscaled results are higher than that of the boundary forcing data.

The results from the EOF analysis for the seasonal daily mean precipitation indicate that WRF is able to reproduce the dominant mode of interannual variability when compared to the MERRA reanalysis data. However, the results for temperature do not show the dominant mode of observation from the interannual variability of the temperature anomalies, it appears that dynamically downscaled precipitation depends on the lateral boundary data, namely APCC/CCSM3. However, precipitation anomalies are identified similarly to variability in observations. Statistical validation examined the pattern correlation coefficient for the simulation period for the MERRA, APCC/CCSM3, and WRF results. The seasonal mean temperature and precipitation predictions from the regional climate model, WRF, showed high spatial correlation, better than the lateral boundary conditions, APCC/CCSM3 forecast over the land region. It has been confirmed that the regional climate predictability of a dynamically downscaled model performed better than the forcing data for the Indonesia Maritime Continent.

To investigate the regional impacts of ENSO, the research period was classified into Normal, El Niño and La Niña year, depending on the summer sea surface temperature (SST) anomalies in the Niño 3.4 region. Indonesian climatology is divided into three types, depending on the regional characteristics of the precipitation. The division of analyzed areas is monsoon pattern (Region A), equatorial pattern (Region B), and local pattern (Region C).

The cumulative rainfall for the entire regions is the largest (lowest) during the La Nina (El Nino) period. Region C shows the most significant deviation and in Region B appears the least significant variation, according to the concerned ENSO period. WRF generally can simulate similar features to MERRA data. However, the regional climate model results underestimate accumulated precipitation.

We analyzed the vertical distribution of the wind field and relative humidity and the lower atmosphere divergence fields for the El Nino, La Nina, and Normal years. The easterly winds of the Western Pacific during a La Nina year flowed into Region C. During the El Niño period, that easterly flow weakened.

According to the results, there are systematic biases in the dynamically downscaled model but the seasonal mean climatology and interannual variability of regional climate models are comparable to observations. This may indicate that APCC/CCSM3 can produce optimal seasonal forecasts at high resolution for the Indonesian Maritime Continent. In the future, improvements in the predictive power of APCC/CCSM3 and the application of statistical post-treatments are expected to produce more realistic regional scale predictions.

1. 서론

2. 연구 자료 및 방법

2.1 전구기후모델

2.2 지역기후모델

2.3. 관측자료

2.4. 연구 지역

3. 연구결과 및 분석

3.1 지역기후모델 예측성 평가

3.2 여름철 maritime continent 지역의 ENSO 영향 분석

4. 결론