The Madden–Julian Oscillation (MJO) has a reputation: climate models still find it hard to get right. Even with decades of development, its propagation and interaction with land and ocean remain stubbornly tricky. This paper grew out of a simple but nagging question we kept coming back to: how much does the convective adjustment timescale (τ) really matter for the MJO?

What is τ? A thermodynamically unstable atmosphere releases its energy by creating clouds. Deep clouds take time to decay and bring the atmosphere back to equilibrium. The decay time of deep clouds is called convective adjustment timescale or simply adjustment timescale, typically denoted by τ in climate model formulations. In the Community Atmospheric Model (CAM) it is defined as the convective available potential energy (CAPE) consumption time scale and is typically set to 1 hour, everywhere in the globally. In many models τ is a fairly free tuning choice.

Studies have hinted that it might play a pivotal role in shaping intraseasonal variability. Somewhat coincidentally, August 2025 saw three different papers, ours included, digging into τ in CAM, all from slightly different perspectives (Zhou et al., Pathak et al., and Goswami et al.).

Zhou et al. showed that using larger τ values generally improves MJO simulations. Pathak et al. took a more ambitious route, letting τ vary dynamically in space and time, but still ended up with unrealistic MJO propagation. In our study, we wondered whether the missing piece might be something more basic: the contrast between land and ocean.

In our study, Goswami et al., when we prescribed a short τ over land (1 hour) and a longer τ over the ocean (4 hours), CAM produced much more realistic MJO propagation. But applying the same 4-hour τ everywhere, even though it helps over the ocean actually broke the MJO. The take home message from our study is that it’s not just the value of τ that matters, but where it’s applied.

What this points to is the importance of how the boundary layer links surface properties to convection. If a model doesn’t capture land-ocean differences in that coupling, both accurately and efficiently, it’s going to struggle with tropical variability like the MJO.

Stepping back, it was striking to see three studies arrive at related conclusions around the same time. Together, they suggest a growing consensus: improving simulations of organized convection, e.g., the MJO, likely requires moving away from one-size-fits-all convective assumptions and toward approaches that better reflect real-world heterogeneity.

References:

Zhou, Xuan, et al. "Sensitivity of MJO simulation to the convective adjustment timescale in CAM6." Atmospheric and Oceanic Science Letters (2025)

Pathak, Raju, et al. "Formulation of a dynamic convective adjustment time-scale in the CESM1. 2 and its influence on the Indian summer monsoon simulations." Scientific Reports (2025)

Goswami, Bidyut Bikash, Andrea Polesello, and Caroline Muller. "An assessment of representing land‐ocean heterogeneity via CAPE relaxation timescale in the Community Atmospheric Model 6 (CAM6)." Journal of Advances in Modeling Earth Systems (2025)