Estimation of planetary boundary layer height (PBLH) from profile data at Bosung site

The planetary boundary layer (PBL) is the lowest part of the atmosphere and it is directly influenced by its contact with a planetary surface and the PBL height (PBLH) varies with time, season and topography. In this study, PBLH is estimated from atmospheric profiles utilizing the parcel method, the bulk Richardson method and the temperature inversion method using the potential temperature, turbulence and temperature gradients, respectively. Additionally, the theta gradient profile and the mixing ratio gradient profile are used to estimate PBL discontinuity in the previous methods. The ERA5 reanalysis is used as reference data in order to develop the PBLH estimation algorithm. During the day, the PBLH is first estimated by the parcel method. If the estimated PBLH is less than 3000 m, it is final PBLH. If not, secondly, calculate the bulk Richardson number to find the height below criteria (0.25) as the final PBLH. When these conditions are not satisfied, finally, the height of the maximum theta gradient subtracts the difference between the height obtained from the bulk Richardson number and the height of the maximum theta gradient. These differences are hourly averaged over the duration of the study. During the nighttime (from evening to morning), PBLH is estimated by the temperature inversion method, and if it greater than 1500 m or not calculated, PBLH is calculated using the theta gradient method and the mixing ratio gradient method. The algorithm is applied to Ground-based microwave radiometer (MWR) profiles, the one-dimensional variational (1D-VAR) retrieval profiles based on the optimal estimation by combining MWR brightness temperatures and a priori information from Local Data Assimilation and Prediction System model. The PBLH difference between MWR and ERA5 shows a large correlation coefficient (0.59) than between 1D-VAR and ERA5. On contrary, the statistical error of 1D-VAR is lower than MWR. Compared to MWR, The RMSE is reduced to 12 % and bias is reduced to 82.5 %. Detailed methods and results in this study will be presented at the conference.