Process-based understanding of improved MJO propagation across the Maritime Continent in GloSea6

Given the critical role of the Madden–Julian Oscillation (MJO) in modulating global climate variability and subseasonal-to-seasonal (S2S) predictability, this study evaluates its simulation in the Korea Meteorological Administration’s Global Seasonal Forecasting System version 6 (GloSea6) and compares it with version 5 (GloSea5), focusing on prediction skill and key physical processes over the Maritime Continent (MC). Both models exhibit systematic biases, including weaker amplitudes and a tendency for the MJO to stall over the MC. Nevertheless, GloSea6 shows enhanced propagation across the MC, consistent with improved thermodynamic processes. The eastward-to-westward spectral power ratio increases from 1.52 in GloSea5 to 1.93 in GloSea6, closer to the observed 2.79, reflecting a more realistic dominance of eastward propagation. Process-based diagnostics reveal region-dependent improvements: more pronounced over the MC but limited over the Indian Ocean (IO). MC improvements are linked to better simulation of lower-level moisture convergence, equivalent potential temperature, and available potential energy, supported by reduced SST biases and a steeper meridional moisture gradient. These background-state changes strengthen moistening processes that precondition convection and sustain eastward propagation over the MC. These findings highlight that thermodynamic and mean-state improvements in GloSea6 are process- and region-dependent, and play a key role in shaping MJO-driven variability relevant to subtropical climate, emphasizing the importance of reducing systematic biases for improving S2S prediction system. However, improvements in spatial pattern similarity did not always translate into propagation skill gains, particularly over the IO, underscoring the complexity of dynaical responses.