The Seasonal Ensemble Prediction System (JMA/MRI-CPS4)

The latest JMA ensemble prediction system for sub-seasonal to seasonal forecasts, JMA/MRI-CPS4 (the Japan Meteorological Agency/Meteorological Research Institute-Coupled Prediction System version 4, or simply "CPS4") was implemented in January 2026 for use in one-month prediction, three-month prediction, warm/cold season outlook and El Niño Outlook. CPS4 specifications and the related operational time schedule are outlined here.

CPS4 is an atmosphere/ocean/land/sea ice-coupled prediction system consisting of an initialization system and a forecast model. Its atmosphere-land surface model is largely based on that of CPS3, which is an enhanced version of JMA-GSM2003 (Yonehara et al. 2020, JMA 2022) for seasonal prediction. CPS4 features refined cloud, stratocumulus and cumulus convection schemes, land-snow and lake models, and other elements. Prognostic ozone with linear ozone photochemistry parametrization is used in the radiation process instead of the monthly climatology derived from MRI-CCM2 reanalysis (Deushi and Shibata 2011) used in CPS3, resulting in enhanced representation of stratospheric circulation. Atmospheric conditions are initialized with JRA-3Q (Kosaka et al. 2024) for re-forecasts, and Global Analysis (GA) for operation, in which atmospheric conditions are updated with a shorter delay. CPS4 has a different land-sea mask from these analyses and a unique lake scheme (Adachi et al. 2025) requiring initialization. To avoid initial shock, offline surface simulation for the period to date is separately run and used for forecasts. Ozone concentration is initialized with ozone analysis generated by MRI-CCM2.1 (Deushi and Shibata 2011, Yukimoto et al. 2019) as with JRA-3Q for use in the initialization of a linear ozone scheme. Ocean and sea ice conditions are initialized with MOVE/MRI.COM-G3 (Fujii et al. 2023) "a global ocean 4DVAR analysis (Usui et al. 2015) downscaled to an eddy-permitting resolution (0.25 by 0.25° in longitude and latitude)". The ensemble prediction system involves a combination of LAF, initial and model perturbation. Perturbed atmospheric conditions are determined for each initial time using SV+LETKF. Ocean perturbations are calculated using 4DVAR minimization history, by which daily analysis error covariance can be approximated (Niwa and Fujii 2020).

System Name		JMA/MRI-CPS3 (Hirahara et al. 2023)	JMA/MRI-CPS4 (Kubo et al. 2025)
Operation Start Date		February 2022	January 2026
Atmospheric model	Model version	GSM2003C	GSM2003C*
	Horizontal Resolution	Global TL319 reduced Gaussian grid(~55km)
	Vertical levels(model top)	100 levels(0.01hPa)	128 levels(0.01hPa)
Ocean model	Model version	MRI.COM v4.6 (Tsujino et al. 2020)	MRI.COM v5.0 (Sakamoto et al. 2023)
	Horizontal Resolution	0.25° (longitude) x 0.25° (latitude)
	Vertical levels	60 levels
Forecast Frequency		5 ensemble members per day	25 ensemble members up to 1-month every Tuesday and Wednesday 5 ensemble members per day
Initial condition	Atmosphere	JRA-3Q (hindcasts) and the Global Analysis (GA; forecasts only)	JRA-3Q (hindcasts), GA (forecasts only) and Ozon Analysis
	Land Surface	Offline model runs forced by JRA-3Q and GA (forecasts only)
	Ocean	MOVE/MRI.COM-G3(Low-res. 4DVAR+High-res. downscaling)
	Sea ice	MOVE/MRI.COM-G3(3DVAR)
Ensemble generation method	Initial Condition perturbation	The breeding of growing mode (BGM) method for the atmosphere Ocean perturbations calculated using 4DVAR minimization history The Lagged Average Forecast (LAF) method	The Singular Vectors (SVs) and Local Ensemble Transform Kalman Filter (LETKF) methods for the atmosphere Ocean perturbations calculated using 4DVAR minimization history The LAF method
	Model Perturbation	Stochastic Perturbation of Physics Tendency (SPPT) for the atmosphere	SPPT and Stochastic Humidity Profile for Convective parametrization (SHPC; Ota 2025) for the atmosphere
Hindcast	Period	Two initial dates per month for 1991-2020
	Ensemble size	5 members / day for 6-month prediction	13 members / day for 1-month prediction 5 members / day for 6-month prediction

The figure below shows how CPS4 runs operationally. The 25-member ensemble is generated every Tuesday and Wednesday from 00UTC analysis for one-month prediction. JMA uses 25-members for 2 LAF dates to make 50-member model statistics for one-month prediction products once a week (Fig. 1). The five-member ensemble is generated once a day from 00UTC analysis for seasonal prediction. JMA uses five members for 17 LAF dates to create 85-member model statistics for three-month prediction, warm/cold season and El Niño Outlook (Fig. 2). Registered users can download all members.

Fig. 1 Schematic figure of "one-month prediction" ensemble configurations of operational suite.

Fig. 2 Schematic figure of "three-month prediction", "warm/cold season", and "El Niño Outlook" ensemble configurations of operational suite.

To assess forecast performance, a set of hindcasts (re-forecasts) was carried out as in real-time operation (other than for a smaller ensemble size). Scores are available on the hindcast verification page, and registered users can access hindcast data via the download page.

• Adachi, Y., K. Ochi, S. Hirahara, Y. Kubo, R. Sekiguchi, 2025: A simplified lake model for seasonal prediction. WGNE Res. Activ. Earth Sys. Model., 55, 4-47.
• Deushi, M. and K. Shibata, 2011: Development of a Meteorological Research Institute Chemistry-Climate Model version 2 for the study of tropospheric and stratospheric chemistry. Pap. Meteorol. Geophys., 62, 1–46.
• Fujii, Y., T. Yoshida, H. Sugimoto, I. Ishikawa, and S. Urakawa, 2023: Evaluation of a global ocean reanalysis generated by a global ocean data assimilation system based on a Four-Dimensional Variational (4DVAR) method. Front Clim, 4 1-20, doi:10.3389/fclim.2022.1019673.
• Hirahara, S., Y. Kubo, T. Yoshida, T. Komori, J. Chiba, T. Takakura, T. Kanehama, R. Sekiguchi, K. Ochi, H. Sugimoto, Y. Adachi, I. Ishikawa, and Y. Fujii, 2023: Japan Meteorological Agency/Meteorological Research Institute Coupled Prediction System version 3 (JMA/MRI-CPS3). J. Meteor. Soc. Japan, 101, 149-169, doi:10.2151/jmsj.2023-009.
• Japan Meteorological Agency, 2022: Outline of the operational numerical weather prediction at the Japan Meteorological Agency. Appendix to WMO Numerical Weather Prediction Progress Report, Tokyo, Japan. [link]
• Kosaka, Y., S. Kobayashi, Y. Harada, C. Kobayashi, H. Naoe, K. Yoshimoto, M. Harada, N. Goto, J. Chiba, K. Miyaoka, R. Sekiguchi, M. Deushi, H. Kamahori, T. Nakaegawa, T. Y. Tanaka, T. Tokuhiro, Y. Sato, Y. Matsushita, and K. Onogi, 2024: The JRA-3Q reanalysis. J. Meteor. Soc. Japan, 102, 49-109, doi:10.2151/jmsj.2024-004.
• Kubo, Y., K. Ochi, J. Chiba, T. Yoshida, T. Takakura, R. Sekiguchi, Y. Adachi, M. Deushi, and S. Hirahara, 2025: Upgrade of the JMA Sub-Seasonal and Seasonal Ensemble Prediction System (JMA/MRI-CPS4). WGNE Res. Activ. Earth Sys. Model., 55, 1-14.
• Niwa, Y. and Y. Fujii, 2020: A conjugate BFGS method for accurate estimation of a posterior error covariance matrix in a linear inverse problem. Quart. J. Roy. Meteor. Soc., 146, 3118-3143.
• Ota, Y., 2025: Introduction of Stochastic Humidity Profile for Convective parametrization (SHPC) method in JMA’s Global Ensemble Prediction System. WGNE Res. Activ. Earth Sys. Model., 55, 3-34.
• Sakamoto, K., H. Nakano, S. Urakawa, T. Toyoda, Y. Kawakami, H. Tsujino, and G. Yamanaka, 2023: Reference manual for the Meteorological Research Institute Community Ocean Model version 5 (MRI.COMv5). Tech. Rep. MRI, 87.
• Tsujino, H., M. Hirabara, H. Nakano, T. Yasuda, T. Motoi and G. Yamanaka, 2020: Simulating present climate of the global ocean-ice system using the Meteorological Research Institute Community Ocean Model (MRI.COM): simulation characteristics and variability in the Pacific sector. Journal of Oceanography, 67, 449-479.
• Usui, N., Y. Fujii, K. Sakamoto and M. Kamachi, 2015: Development of a Four-Dimensional Variational Assimilation System for Coastal Data Assimilation around Japan. Monthly Weather Review, 143, 10, 3874-3892.
• Yonehara, H., C. Matsukawa, T. Nabetani, T. Kanehama, T. Tokuhiro, K. Yamada, R. Nagasawa, Y. Adachi, and R. Sekiguchi, 2020: Upgrade of JMA's Operational Global Model. WGNE. Res. Activ. Earth. Sys. Modell., 50, 6-19.
• Yukimoto, S., H. Kawai, T. Koshiro, N. Oshima, K. Yoshida, S. Urakawa, H. Tsujino, M. Deushi, T. Tanaka, M. Hosaka, S. Yabu, H. Yoshimura, E. Shindo, R. Mizuta, A. Obata, Y. Adachi, and M. Ishii, 2019: The Meteorological Research Institute Earth System Model, Version 2.0, MRI-ESM2.0: Description and basic evaluation of the physical component. J. Meteor. Soc. Japan, 97, 931–965.

Appendix: Upgrade history of JMA's coupled prediction system

Implemented date	Upgrades and Changes
Aug. 1998	The first JMA's CGCM for ENSO prediction implemented to operation.
Jul. 2003	The second CGCM for ENSO prediction (JMA-CGCM02) with an improved atmospheric model and sea-surface flux adjustment.
Jun. 2005	Frequent forecast update (from every 15 days to every 5 days).
Mar. 2008	The Third CGCM for ENSO prediction (JMA/MRI-CGCM1). Oceanic component changed to the new ocean general circulation model "MRI.COM" and data assimilation system "MOVE/MRI.COM-G".
Mar. 2009	Introduced an ocean initial perturbation method. The ensemble size increased from 12 to 30.
Mar. 2010	The first integrated forecast system "JMA/MRI-CPS1" that covers seasonal forecasts in addition to the "El Niño outlook". Introduced an atmospheric perturbation method. The ensemble size increased from 30 to 51.
Jun. 2015	JMA/MRI-CPS2: a dynamical sea ice model, land initialization, historical GHG concentration. The ensemble size per initial date increased from 9 to 13.
Feb. 2022	JMA/MRI-CPS3: 4DVAR ocean and 3DVAR sea ice initialization, an eddy-permitting ocean resolution, improved atmospheric physics, daily forecast update (5 ensemble members per day).
Jan. 2026	JMA/MRI-CPS4: Prognostic ozone with linear ozone photochemistry parametrization, SV+LETKF+SHPC as ensemble generation method, improved atmospheric physics, ensemble size update, use in "One-month prediction".