Ensemble Machine Learning Approaches for Coastal Planetary Boundary Layer Height Estimation Based on Coherent Doppler Wind Lidar

The planetary boundary layer is instrumental in mediating the exchange of heat and matter between the Earth's surface and the atmosphere. Using continuous high-resolution wind profiling measurements conducted in Qingdao, an eastern coastal city of China, from December 2019 to April 2020, this study constructs and validates an integrated model that integrates the K-means and Random Forest algorithms. The model estimates the planetary boundary layer height across five real-time weather scenarios:clear, rainy, few clouds, moderate clouds, and cloudy, incorporating a suite of meteorological variables sourced from the ERA5 reanalysis dataset. In particular, the training results for <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$C2$</tex> (a few clouds) demonstrate better performance, with a value of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$R^{2}$</tex> of 0.92, indicating a high degree of robustness. This research offers substantial value for meteorological studies and engineering applications in coastal zones.