Contents Shortcut
Main Menu Shortcut

Methodology

APCC Seasonal Forecasts

The APCC seasonal forecast is based on multi-model ensemble (MME) prediction system and disseminated to APEC member economics around 20th of every month. Currently, 14 operational centers and research institutes from 10 countries around the world participate in the APCC MME operational prediction system by routinely providing their predictions in the form of ensembles of global forecast fields (More information on participating models).

The APCC’s real-time operational forecasts are issued in both deterministic (based on ensemble mean) and probabilistic (based on full set of ensemble members) forms and more detailed description of the methods is as follows.

Deterministic MME Forecast

The deterministic forecast is based on a simply average of bias-corrected ensemble means from each model with equal weight to create a multi-model forecast. The ensemble mean anomaly forecasts for each individual model is calculated by their own climatology from the hindcasts.

Probabilistic MME Forecast

The probabilistic forecast is based on an uncalibrated MME with model weights being proportional to the square root of ensemble size, and a Gaussian fitting method for the estimation of the tercile-based categorical probabilities, that is, the probability of below-normal (BN), near-normal (NN), and above-normal (AN) categories with respect to climatology (Min et al. 2009). The procedure for the probabilistic forecast consists of following two steps.

1.Estimation of individual model probabilities: The upper and lower terciles are determined separately for each model using their mean and standard deviation of hindcasts. Then, the forecast probability for each category is estimated as a portion of the cumulative probability of their forecast sample associated with the category.

2.Multi-model combination: The forecast probabilities for each model are averaged together with model weights being inversely proportional to the random errors in the forecast probability associated with the standard error of the ensemble mean (i.e., proportional to the square root of ensemble size) to create a probabilistic multi-model ensemble forecast. For each grid point, probability forecasts in the dominant category being statistically significant at 5% level based on Pearson’s chi-square test are displayed in colors.

NOTE:

When using the data, please cite Min et al. (2017) and in the acknowledgements please note that "The authors acknowledge that the APCC Multi Model Ensemble (MME) Producing Centers for making their hindcast/forecast data available for analysis and the APEC Climate Center for collecting and archiving them and for organizing APCC MME prediction."

The MME and individual model data can be found in the APCC Data Service.

Reference:

Min, Y.-M., V. N. Kryjov, S. M. Oh, H.-J. Lee (2017): Skill of real-time operational forecasts with the APCC multi-model ensemble prediction system during the period 2008-2015. Climate Dynamics, 49:4141-4156. doi:10.1007/s00382-017-3576-2.

Min, Y.-M., V.N. Kryjov, C.-K. Park, 2009: Probabilistic Multimodel Ensemble Approach to Seasonal Prediction. Weather and Forecasting, 24, 812-828.