Estimates of the spatially complete, observational-data-driven planetary boundary layer height over the contiguous United States

This study aims to generate a spatially complete planetary boundary layer height (PBLH) product over the contiguous United States (CONUS). An eXtreme Gradient Boosting (XGB) regression model was developed using selected meteorological and geographical data fields as explanatory variables to fit the PBLH values derived from Aircraft Meteorological DAta Relay (AMDAR) reports hourly profiles at 13:00-14:00 LST (local solar time) during 2005-2019. A preprocessing step was implemented to exclude AMDAR data points that were unexplainable by the predictors, mostly under stable conditions. The PBLH prediction by this work as well as PBLHs from three reanalysis datasets (the fifth-generation European Centre for Medium-Range Weather Forecasts atmospheric reanalysis of the global climate - ERA5; the Modern-Era Retrospective analysis for Research andApplications, Version 2 - MERRA-2; and the North American Regional Reanalysis - NARR) were compared to reference PBLH observations from spaceborne lidar (Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations, CALIPSO), airborne lidar (High Spectral Resolution Lidar, HSRL), and in situ research aircraft profiles from the Deriving Information on Surface Conditions from Column and Vertically Resolved Observations Relevant to Air Quality (DISCOVER-AQ) campaigns. Compared with PBLHs from reanalysis products, the PBLH prediction from this work shows closer agreement with the reference observations, with the caveat that different PBLH products and estimates have different ways of identifying the PBLH; thus, their comparisons should be interpreted with caution. The reanalysis products show significant high biases in the western CONUS relative to the reference observations. One direct application of the dataset generated by this work is that it enables sampling of the PBLH at the sounding locations and times of sensors aboard satellites with an overpass time in the early afternoon, e.g., the Afternoon Train (A-train), the Suomi National Polar-orbiting Partnership (Suomi NPP), the Joint Polar Satellite System (JPSS), and the Sentinel-5 Precursor (Sentinel-5P) satellite sensors. As both AMDAR and ERA5 are continuous at hourly resolution, the observational-data-driven PBLHs may be extended to other daytime hours.