Revisiting Intrinsic Predictability of Wave‐Convection Coupled Bands Over Southern China: Variable and Scale‐Dependent Error Growth Characteristics

Abstract This study investigates the variable‐ and scale‐dependent intrinsic predictability of wave‐convection coupled bands over southern China on 30 January 2018, using 1‐km simulations under varying initial moisture conditions. The predictability time is quantified via the Loss Predictability Index (LPI), defined as the ratio of the forecast error power spectrum to the reference (unperturbed) power spectrum at a given scale. Spectral analysis reveals substantial differences in the reference power spectral slopes among variables, while their error growth behaviors consistently exhibit upscale features. The intrinsic predictability limit of the banded convection, measured by the difference total energy (DTE), is approximately 7 hr. Predictability varies with both scale and altitude: smaller scales (i.e., ∼10 km) have shorter limits than larger scales (i.e., ∼40 km), and the middle‐level moist neutral stability layer is less predictable than the low‐level ducting stable layer. In particular, for the moist neutral stability layer, different variables become more correlated under the coupling between gravity waves and moist convection, yielding more coherent predictability characteristics. In the dry experiment, predictability exceeds 12 hr with minimal error growth, regardless of the variable, scale, or altitude. Finally, the decomposition of the horizontal kinetic energy spectrum into divergent and rotational components demonstrates contrasting power spectra, intrinsic predictability limits, and their sensitivity to initial moist content, with the divergent component exhibiting longer predictability in the ducting stable layer at wavelengths <40 km. These findings highlight how vertical flow structure, moisture content, and distinct dynamical components jointly constrain the intrinsic predictability of mesoscale convective systems.