메뉴 건너뛰기


Home > Research > Climate Prediction


Climate Prediction

This research aims to provide high-cost climate data and information by operating a real-time and well-validated climate prediction system based on a multi-climate model ensemble (MME) technique.

Seasonal forecast operation

The APCC has collected dynamic ensemble seasonal prediction data from NMHs and research institutes (17 operations/institutions form 9 APEC member economies), and produced one-month lead three month mean climate forecasts every month with four deterministic and one probabilistic forecasts and disseminated it to APEC member economies every month. The monthly 3-month MME forecast covers three variables: precipitation (PREC), temperature at the 850hPa (T850) and geopotential height at 500hPa (Z500). Also, in order to assess and improve seasonal predictability, the verification of our forecasts is conducted in compliance with the WMO guideline on verification of long-range forecasts. Before every season, 6-month MME climate forecast including ENSO/IOD forecast based on 1-tier predictions from research institutions in the APEC region has also been performed. MME and ENSO/IOD forecast outlook has been disseminated to APEC member economies and several research institutes.

Multi model ensemble methods

The APCC has and hold the five MME methods including four deterministic and one probabilistic method to climate forecasts.

  • Simple composite method (SCM): a simple averaged MME where the contribution of each single-model is equally weighted.
  • Multiple regression based blend of model ensemble means (MRG): empirically weighted MMEs with coefficients computed using multiple linear regression.
  • Synthetic super ensemble (SSE): MRG with the empirical orthogonal function (EOF)-filtered datasets to minimizing the residual error variance.
  • Step-wise pattern projection (SPM): calibrated MME which is obtained from the adjusted (or corrected) single-model predictions based on a stepwise pattern projection method
  • Probabilistic forecast (PMME) : Probabilistic MME based on position of the forecast PDF in respect to the historical PDF

Statistical downscaling and drought

Many of the developing APEC economies lack the infrastructure to issues to multi-model forecast and to downscale these forecasts to the district level. For the production of more valuable local-level forecasts for users of Asia-Pacific region, APCC is providing multi-model output-based statistical downscaling forecast results. Several methods for statistical downscaling were tested for the forecasts in Taiwan, Philippines, Thailand, South Korea, and Japan, and they were confirmed as the outstanding measures for the district level forecasts.
Moreover, one of these outputs is made full use of predicting the electricity demand in Japan. As well as, the local level forecasts in South Korean and Taiwan are being applied to drought forecast to manage water resource of the country.

climate prediction system