jma_logo Welcome to TCC wmo_logo
World Climate
Climate System Monitoring
El Niño Monitoring
NWP Model Prediction
Global Warming
Climate in Japan
Training Module
Press release
HOME > Ensemble Model Prediction > Description of JMA's Seasonal Ensemble Prediction System (JMA/MRI-CPS2)

Description of JMA's Seasonal Ensemble Prediction System (JMA/MRI-CPS2)


The Seasonal Ensemble Prediction System (EPS) is operated at JMA for the purpose of three-month and warm/cold season prediction as well as El Niño monitoring and outlook work. The latest operational Seasonal EPS (JMA/MRI-CPS2: Japan Meteolorogical Agency/Meteorological Research Institute-Coupled Prediction System version 2; Takaya et al. 2018) was implemented into operation in June 2015, replacing the previous system (JMA/MRI-CPS1; Takaya et al. 2017).

Changes made in JMA/MRI-CPS2 include improvements of model resolutions and physics in both atmospheric and oceanic components, and introduction of an interactive sea ice model. In a real-time operational suite, 51-member ensemble integrations are routinely carried out from consecutive initial dates with a five-day interval as in the current configuration of JMA/MRI-CPS1 except for an increased number of ensembles per initial date from 9 to 13. Verification of re-forecasts (hindcasts) shows higher predictive skill of JMA/MRI-CPS2 in three-month, warm/cold season and El Niño prediction, compared to JMA/MRI-CPS1.

Outline of JMA/MRI-CPS2

JMA/MRI-CPS2 adopts an atmosphere-ocean coupled general circulation model (CGCM) as the predecessor (JMA/MRI-CPS1), but with the addition of an interactive sea ice component. A new CGCM (JMA/MRI-CGCM2) has higher resolutions in both atmospheric and oceanic model than the predecessor. The atmospheric model (JMA Global Spectral Model GSM1011C; JMA 2013) has a resolution of TL159 (approximately 110 km grid spacing) with 60 levels (0.1 hPa at the top), and the meridional resolution of the oceanic model is increased from 1 deg. to 0.5 deg. in maximum. The ocean model domain is expanded to the global ocean with the use of a tripolar grid.

The model physics are improved in many aspects, for instance, cumulus convection, clouds, radiation, sea-surface boundary in the atmosphere, and ocean mixed layer, radiation in the ocean. Besides more realistic concentration of Greenhouse Gases (GHGs) is prescribed in model integrations with CMIP5 RCP4.5 scenario, and land conditions are initialized with JRA-55 reanalysis (Kobayashi et al. 2015). An specification of JMA/MRI-CGCM2 is summarized here. Atmospheric initial conditions are also taken from the JRA-55 reanalysis and oceanic initial conditions are produced by an improved version of an ocean assimilation system (MOVE/MRI.COM-G2).

Real-time operation and ensemble technique

The Lagged Average Forecasting (LAF) approach is adopted to produce ensemble initial conditions along with a Breeding of Growing Mode (BGM) method in the similar way as JMA/MRI-CPS1. In addition, a stochastic physics scheme is newly introduced to better represent model uncertainty.

Ensemble configurations of the operational suite are schematically illustrated in Fig. 1. Thirteen-member ensemble integrations are made from consecutive initial dates with a five-day interval. These ensembles consist a total of 51 member ensemble to issue seasonal predictions at JMA.

Schematic figure of ensemble configurations of operational suite
Fig. 1 Schematic figure of ensemble configurations of operational suite.

Performance of JMA/MRI-CPS2

A set of hindcasts (re-forecasts) was carried out in the similar way as the real-time operation except for a small ensemble size (10 members per month). The verification results based on the hindcasts are given from a hindcast verification page. A set of the hindcast data is provided on request from a page for gridded data download (registered users only).


Japan Meteorological Agency, 2013: Outline of the operational numerical weather prediction at the Japan Meteorological Agency. Appendix to WMO Numerical Weather Prediction Progress Report, Tokyo, Japan. [link]

Kobayashi, S., Y. Ota, Y. Harada, A. Ebita, M. Moriya, H. Onoda, K. Onogi, H. Kamahori, C. Kobayashi, H. Endo, K. Miyaoka, and K. Takahashi, 2015: The JRA-55 Reanalysis: General specifications and basic characteristics. J. Meteor. Soc. Japan, 93, 5-48, doi:10.2151/jmsj.2015-001.

Takaya, Y., S. Hirahara, T. Yasuda, S. Matsueda, T. Toyoda, Y. Fujii, H. Sugimoto, C. Matsukawa, I. Ishikawa, H. Mori, R. Nagasawa, Y. Kubo, N. Adachi, G. Yamanaka, T. Kuragano, A. Shimpo, S. Maeda and T. Ose, 2018: Japan Meteorological Agency/Meteorological Research Institute-Coupled Prediction System version 2 (JMA/MRI-CPS2): atmosphere-land-ocean-sea ice coupled prediction system for operational seasonal forecasting. Clim. Dyn., 3-4, 751-765, doi:10.1007/s00382-017-3638-5.

Takaya, Y., T. Yasuda, Y. Fujii, S. Matsumoto, T. Soga, H. Mori, M. Hirai, I. Ishikawa, H. Sato, A. Shimpo, M. Kamachi and T. Ose, 2017: Japan Meteorological Agency/Meteorological Research Institute-Coupled Prediction System version 1 (JMA/MRI-CPS1) for operational seasonal forecasting. Clim. Dyn., 48, 1-2, 313-333, doi:10.1007/s00382-016-3076-9.

Annex: Upgrade history of JMA's coupled prediction system

Implemented date Upgrades & Changes
Aug. 1998 The first JMA's CGCM for ENSO prediction implemented into operation.
Jul. 2003 The second CGCM for ENSO prediction (JMA-CGCM02), upgraded atmospheric component and improved flux adjustment.
Jun. 2005 Frequency of model integration increased (from every 15 days to every 5 days).
Mar. 2008 The Third CGCM for ENSO prediction (JMA/MRI-CGCM1) implemented.
- Atmospheric component changed to lower-resolution version (TL95L40) of the atmospheric global spectral model (GSM) in operation for numerical weather prediction.
- Oceanic component changed to the new ocean general circulation model (MRI.COM) - New ocean data assimilation system (MOVE/MRI.COM-G) implemented with the identical ocean model in the coupled model (JMA/MRI-CGCM1).
Mar. 2009 Oceanic initial perturbations introduced. An ensemble size increased from 12 to 30.
Mar. 2010 Atmospheric perturbations introduced in addition to the oceanic ones. An ensemble size increased from 30 to 51. The CGCM is utilized in seasonal prediction.
Jun. 2015 The forth CGCM for seasonal and ENSO prediction implemented.
- Introduction of a dynamical sea ice model, land initialization, historical GHG concentration.
- Upgrades of atmospheric and oceanic resolutions.
- An ensemble size per initail dates increased from 9 to 13.

page top

Tokyo Climate Center, Climate Prediction Division.
Copyright(C) 2002. Japan Meteorological Agency. - Legal Notice -