Layer-Dependent Diurnal Moist Static Energy Biases in the Unified Forecast System During the MJO Preconditioning Stage: A DYNAMO Case Study

Accurate simulation of the Madden-Julian Oscillation (MJO) requires realistic diurnal evolution and vertical distribution of moist static energy (MSE). While prior studies documented MSE bias in simulations, vertically dependent physical processes responsible for these biases remain insufficiently understood. This study applies a hierarchical modeling and evaluation approach, combining fully coupled Unified Forecast System (UFS) forecasts with observation-constrained single-column model (SCM) experiments, to enable vertically and diurnally resolved assessment of MSE during two MJO events from the DYNAMO field campaign. Results reveal a diurnal loop of layer-dependent biases: daytime errors precondition nighttime residual energy, which in turn influences MSE and convection on the following day. These biases are associated with deficiencies in the convective parameterization, in which convective latent heating is concentrated in the upper troposphere while the transition layer receives insufficient energy input, reflecting a missing shallow-to-deep convective transition. The convective parameterization also converts cloud condensate, especially ice, too efficiently, reinforcing an overly top-heavy energy structure. This residual energy aloft persists overnight, suppressing the re-establishment of low-level instability and limiting convective development the following day. This hierarchical, vertically resolved analysis provides a process-based pathway for diagnosing issues in tropical energetics on the diurnal timescale during the MJO preconditioning stage, with potential implications to improve MJO prediction in the UFS and related forecast systems.