Intercomparison of Four Types of Methods for Estimating Convective Boundary Layer Heights Using Dropsonde Measurements over the Northwest Atlantic

Abstract The planetary boundary layer height (PBLH) is important to characterize because significant atmospheric processes and human activities occur in this lowest layer of the atmosphere. Radiosondes and dropsondes are frequently employed to estimate PBLH, primarily over land, using four different methods: the parcel method, gradient methods using virtual potential temperature or relative humidity, and the bulk Richardson number method. This study comprehensively evaluates eight different algorithms from these four methods using 657 dropsondes for convective planetary boundary layer (PBLs) from the Aerosol Cloud Meteorology Interactions over the Western Atlantic Experiment (ACTIVATE) field campaigns over the western North Atlantic covering winter (30 November–2 April) and summer (May, June, August, and September) seasons from 2020 to 2022. In addition to a dropsonde data smoothing-sensitivity study and a consistency test, we perform additional evaluations using manually estimated PBLHs and High Spectral Resolution Lidar-Generation 2 cloud-top height measurements. The parcel and Richardson number methods are found to be most appropriate, with the parcel method slightly preferred for convective PBLs. Furthermore, the parcel method needs to use both temperature and humidity data, the Richardson number computation needs to include the term related to surface friction velocity, and the gradient method of virtual potential temperature’s threshold value should vary with the amount of smoothing. These methods can be applied to dropsonde data from other field campaigns to consistently estimate convective PBLH over global oceans. Significance Statement The planetary boundary layer (PBL) height is an important parameter to characterize PBL processes that connect the surface and free troposphere. There are four common methods to estimate PBL height: 1) the parcel method, gradient methods using 2) temperature or 3) humidity, and 4) the bulk Richardson number method. In this research, we evaluate these four methods and associated eight algorithms using 657 dropsondes for unstable (i.e., convective) PBLs from the Aerosol Cloud Meteorology Interactions over the Western Atlantic Experiment (ACTIVATE) field campaigns (2020–22). We evaluate these methods using a dropsonde data smoothing-sensitivity study, a consistency test, and manually estimated PBL heights and lidar cloud-top height measurements. We show that the parcel and Richardson number methods are most effective, with the parcel method slightly preferred.