Large-Ensemble Recharge Oscillator Simulation Data

README: RO Parameter Ensemble Simulation Data Version 3 — July 27, 2026 Author: Sooman Han (sooman.han@yale.edu) 1. Description This dataset contains ensemble simulations of a recharge oscillator (RO) model designed to quantify internal variability and assess the detectability of forced changes in the presence of internal variability. The numerical code used to generate these simulations was previously archived as “ENSO Recharge Oscillator: Parameter Sweep Simulation Code.” The code is available at: https://zenodo.org/records/16888262 The dataset includes the following simulation cases: RO_fixed: Fixed parameters based on Han et al. (2026). RO_fixed_perturbed: Same configuration as RO_fixed, but with the linear parameters perturbed within their uncertainty ranges. RO_KA21_B04: Fixed parameters based on Kim and An (2021, Journal of Climate). RO_KA21_B04_perturbed: Same configuration as RO_KA21_B04, but with the linear parameters perturbed within their uncertainty ranges. RO_R_??: A ??% per century increase in (R_0), where ?? = 05, 10, ..., 50. RO_noise_??: A ??% per century increase in (\sigma_T), where ?? = 05, 10, ..., 30. RO_F1_??: A ??% per century increase in (F_1), where ?? = 05, 10, ..., 30. RO_F2_m??: A ??% per century decrease in (F_2), where ?? = 05, 10, ..., 30. RO_epsilon_m??: A ??% per century decrease in (\epsilon), where ?? = 05, 10, ..., 30. Each case contains time series of temperature (T) and thermocline-depth (h) anomalies for 5,000 ensemble members. The simulations were integrated using the Euler–Maruyama scheme with a time step of 0.01 seconds. For this data archive, 2,400 months of each ensemble member were retained, with output saved at a monthly cadence. For further details on the model formulation, see Han et al. (2026, Journal of Climate). 2. File Description Each simulation folder contains the following files. 2.1 Simulated Time Series simul_{simul_num}.txt where: simul_num = 0000, 0001, ..., 4999 Each file contains the simulated time series of T and h. Format: Column 1: T Column 2: h 2.2 Moving-Window Diagnostics and Fitted Parameters std_moving_0040_{simul_num}_multiplicative_LR-F.txt skew_moving_0040_{simul_num}_multiplicative_LR-F.txt lag_moving_0040_{simul_num}_multiplicative_LR-F.txt fitted_parameters_moving_0040_{simul_num}_multiplicative_LR-F.txt These files contain diagnostics calculated over 40-year moving windows, including: the standard deviation of T; the skewness of T; the ENSO period, estimated from the first minimum of the autocorrelation function; and the fitted RO model parameters. 2.3 Ensemble-Mean Slope Distributions {parameter}_ensmean_slope_distribution_L{60 or 100}_0040y_{0100 or 5000}members.txt These files contain Monte Carlo sampling results for trends in the estimated parameters. L60 and L100 indicate analyses based on 100- and 140-year time series, respectively. With a 40-year moving-window length, these time series yield 60 and 100 years over which the moving-window estimates vary (i.e., (100 - 40 = 60) and (140 - 40 = 100)). 0100members indicates distributions obtained from randomly sampled 100-member ensembles for 10,000 times. 5000memberts indicates slope estimates obtained using all 5,000 ensemble members and is used to represent the underlying forced change. 2.4 Observational Data The obs folder contains: nino3_HadISST.txt: HadISST Niño-3 index time series. std_moving_0040_obs.txt: 40-year moving-window standard deviation. lag_moving_0040_obs.txt: 40-year moving-window ENSO period. skew_moving_0040_obs.txt: 40-year moving-window skewness. fitted_parameters_moving_0040_obs_multiplicative_LR-F.txt: RO parameters fitted over 40-year moving windows. 2.5 Plotting Scripts Plotting scripts are provided in the following folders: fig_01 fig_02_S04_S05 fig_03 fig_S01 fig_S02 fig_S03 fig_S06 3. Reference Han, S., A. V. Fedorov, and J. Vialard, 2026: Realistic ENSO dynamics requires a damped nonlinear recharge oscillator. Journal of Climate, 39(1), 77–101. https://doi.org/10.1175/JCLI-D-25-0250.1