Error Growth Characterization Associated With Lateral Boundary Conditions in Idealized Moist Baroclinic Wave Simulations
Abstract
This study investigates numerical error growth mechanisms associated with lateral boundary conditions (LBCs) and their sensitivity to update frequency within a “Big-Brother” framework. Using an idealized moist WRF baroclinic wave simulation, a high-resolution reference is compared against three nested configurations: traditional one-way 6-hr and 3-hr offline updates, and a two-way online approach. Results characterize error evolution as a hybrid process, where LBC update frequency and nesting configuration dictate the transition between externally forced boundary noise and internal intrinsic dynamics. Crucially, the two-way configuration substantially reduces the rapid initial adjustment evident in one-way offline experiments. In one-way experiments, Difference Total Energy surges immediately during the initial stage, dominated by the large-scale (𝜆 >1000 km) component, reflecting a structural nesting adjustment associated with offline LBC treatment. Conversely, the two-way experiment suppresses this initial growth; its subsequent evolution is driven by small-scale (𝜆 <200 km) instabilities exhibiting an upscale error cascade. Spectral analysis reveals a significant vertical disparity in predictability limits. In the upper troposphere, the two-way configuration delays saturation below the 60% threshold and prevents the total saturation (95% limit) observed in one-way runs by Day 5. However, this advantage is notably reduced in the moist mid-troposphere, where all experiments converge rapidly. Additional sensitivity experiments with a 72-hr spin-up confirm that this predictability hierarchy reflects persistent boundary coupling rather than temporary initialization artifacts. This work provides a robust theoretical basis for optimizing lateral boundary coupling in regional weather prediction.
Minjae Lee
Student in Ph.D. course
