IMPLEMENTATION OF CONSTRUCTED ANALOGUE ON SEASONAL RAINFALL FORECAST FROM CFS (CLIMATE FORECAST SYSTEM) OUTPUT CASE STUDY: ISLANDS OF JAVA AND SUMATERA
This study discussed the implementation of a statistical downscaling method called CA (constructed analogue). The method was used to obtain a local scale seasonal (up to 3 months) rainfall forecast from global scale output of CFS-R(Climate Forecast System – Reforecast). As CA is highly sensitive...
Saved in:
| 主要作者: | |
|---|---|
| 格式: | Theses |
| 語言: | Indonesia |
| 在線閱讀: | https://digilib.itb.ac.id/gdl/view/18950 |
| 標簽: |
添加標簽
沒有標簽, 成為第一個標記此記錄!
|
| 總結: | This study discussed the implementation of a statistical downscaling method called CA (constructed analogue). The method was used to obtain a local scale seasonal (up to 3 months) rainfall forecast from global scale output of CFS-R(Climate Forecast System – Reforecast). As CA is highly sensitive to predictor chosen, an analysis to determine the best predictor between three chosen candidates was set as the first step. The next step was intended to overcome CA <br />
<br />
<br />
<br />
<br />
sensitivity of domain, by introducing multi windows approach. Seven windows were chosen as areas which were known to affect local rainfall in Indonesian <br />
<br />
<br />
<br />
<br />
region. The result of this study has shown that the best predictor for local rainfall is stream function at 850 mb. Multi windows approach has also shown to increase the average skill of local rainfall estimation by up to 20% compared to single best analogue etimation. |
|---|