Estimating Planetary Boundary Layer Height Using CALIPSO Lidar Data: A Machine Learning Approach

The planetary boundary layer height (PBLH) is a critical atmospheric parameter influencing air quality, pollutant dispersion, and weather forecasting. Traditional methods for PBLH retrieval rely on radiosondes and ground-based sensors, but their spatial and temporal coverage is limited. In this study, we present a novel application of Random Forest (RF) machine learning to estimate PBLH using lidar measurements from the CALIPSO satellite's Level 1 data spanning a decade. Our RF model is trained with an extensive dataset of radiosonde-derived PBLH values coinciding with CALIPSO overpasses. This approach leverages CALIOP's lidar backscatter profiles to achieve robust performance (R² = 0.6, RMSE = 333.59 m) across a range of atmospheric conditions, including cloudy and dust-laden scenarios, without requiring atmospheric typing or ancillary data. The results surpass state-of-the-art methods in global applicability and accuracy, offering improved spatial and temporal resolution of PBLH estimates. We also discuss the model's performance variations between day- and nighttime scenarios and highlight challenges, such as data bias and surface reflection contamination, which inform future model refinements. This study underscores the potential of integrating machine learning and lidar remote sensing for advancing atmospheric science [1-2].

 

REFERENCES

[1] S. Lolli, W. Y. Khor, M. M. Z. Matjafri, and H. S. Lim, "Monsoon season quantitative assessment of biomass burning clear-sky aerosol radiative effect at surface by ground-based lidar observations in Pulau Pinang, Malaysia in 2014," Remote Sensing, vol. 11, no. 22, 2019.

[2] C. Sivaraman, S. McFarlane, E. Chapman, M. Jensen, Toto, S. Liu, and M. Fischer, Planetary Boundary Layer Height (PBL) Value Added Product (VAP): Radiosonde Retrievals, Tech. Rep., DOE Office of Science Atmospheric Radiation Measurement (ARM) Program, United States, Aug. 2013.

ACKNOWLEDGEMENTS
This research is part of the project PID2021-126436OB-C21 funded by Ministerio de Ciencia e Investigación (MCIN)/Agencia Estatal de Investigación (AEI)/ 10.13039/501100011033 y FEDER “Una manera de hacer Europa” and part of the PRIN 2022 PNRR, Project P20224AT3W funded by Ministero dell’Universit`a e della Ricerca. The European Commission collaborated under projects H2020 ATMO-ACCESS (GA-101008004) and H2020 ACTRIS-IMP (GA-871115).