Investigation of the Atmospheric Boundary Layer Height Using Radio Occultation: A Case Study during Twelve Super Typhoons over the Northwest Pacific

This study investigated the relationship between variations in the atmospheric boundary layer height (ABLH) and typhoons over the Northwest Pacific using global navigation satellite system (GNSS) radio occultation (RO) data during the local summer typhoon season (July–October in the Northern Hemisphere) from 2007 to 2020. The minimum gradient of refractivity derived from COSMIC and COSMIC-2 was used to determine the ABLH. The RO profiles were co-located with the position of a typhoon track base within a 600 km space window and different time windows. ABLH climatology with a 2.5° × 2.5° horizontal resolution was developed, which can be used to obtain the interpolated mean ABLH at any target position. The mean ABLH at the central typhoon position in a specific year was compared with the results interpolated from the climatology of the same location (excluding the year in which the investigated typhoon occurred). In this paper, the results indicate that the ABLH is lower in the vicinity of typhoons relative to the undisturbed atmosphere by a significant amount, and that the reduction in ABLH ranges from 0.13 km to 0.39 km. It was also found that the ABLH was negatively correlated with wind speed, and that the mean correlation coefficient was −0.607. Moreover, similar results can be obtained via the RO water vapor partial pressure profile compared to the refractivity results.

Deriving boundary layer height from aerosol lidar using machine learning: KABL and ADABL algorithms

The atmospheric boundary layer height (BLH) is a key parameter for many meteorological applications, including air quality forecasts. Several algorithms have been proposed to automatically estimate BLH from lidar backscatter profiles. However recent advances in computing have enabled new approaches using machine learning that are seemingly well suited to this problem. Machine learning can handle complex classification problems and can be trained by a human expert. This paper describes and compares two machine-learning methods, the K-means unsupervised algorithm and the AdaBoost supervised algorithm, to derive BLH from lidar backscatter profiles. The K-means for Atmospheric Boundary Layer (KABL) and AdaBoost for Atmospheric Boundary Layer (ADABL) algorithm codes used in this study are free and open source. Both methods were compared to reference BLHs derived from colocated radiosonde data over a 2-year period (2017–2018) at two Météo-France operational network sites (Trappes and Brest). A large discrepancy between the root-mean-square error (RMSE) and correlation with radiosondes was observed between the two sites. At the Trappes site, KABL and ADABL outperformed the manufacturer's algorithm, while the performance was clearly reversed at the Brest site. We conclude that ADABL is a promising algorithm (RMSE of 550 m at Trappes, 800 m for manufacturer) but has training issues that need to be resolved; KABL has a lower performance (RMSE of 800 m at Trappes) than ADABL but is much more versatile.

Assessment of Trends and Uncertainties in the Atmospheric Boundary Layer Height Estimated Using Radiosounding Observations over Europe

Trends in atmospheric boundary layer height may represent an indication of climate changes. The related modified interaction between the surface and free atmosphere affects both thermodynamics variables and dilution of chemical constituents. Boundary layer is also a major player in various feedback mechanisms of interest for climate models. This paper investigates trends in the nocturnal and convective boundary layer height at mid-latitudes in Europe using radiosounding profiles from the Integrated Global Radiosounding Archive (IGRA). Atmospheric data from the European Centre for Medium-Range Weather Forecasts (ECMWF) ReAnalysis v5 (ERA5) and from the GCOS Reference Upper-Air Network (GRUAN) Lindenberg station are used as intercomparison datasets for the study of structural and parametric uncertainties in the trend analysis. Trends are calculated after the removal of the lag-1 autocorrelation term for each time series. The study confirms the large differences reported in literature between the boundary layer height estimates obtained with the two different algorithms used for IGRA and ERA5 data: ERA5 shows a density distribution with median values of 350 m and 1150 m for the night and the daytime data, respectively, while the corresponding IGRA median values are of 1150 m and 1750 m. An overall good agreement between the estimated trends is found for nighttime data, while daytime ERA5 boundary layer height estimates over Europe are characterized by a lower spatial homogeneity than IGRA. Parametric uncertainties due to missing data in both the time and space domain are also investigated: the former is not exceeding 1.5 m, while the latter are within 10 m during night and 17 m during the day. Recommendations on dataset filtering based on time series completeness are provided. Finally, the comparison between the Lindenberg data as processed at high-resolution by GRUAN and as provided to IGRA at a lower resolution, shows the significant impact of using high-resolution data in the determination of the boundary layer height, with differences from about 200 m to 450 m for both night and day, as well as a large deviation in the estimated trend.