PBLHsat: Machine learning-based improvement of planetary boundary layer height from atmospheric model simulations using CALIOP satellite and ACTRIS Earlinet ground-based lidar observations
PIANO NAZIONALE DI RIPRESA E RESILIENZA (PNRR)
Missione 4 “Istruzione e Ricerca” - Componente C2
Investimento 1.1, “Fondo per il Programma Nazionale di Ricerca e Progetti di Rilevante Interesse Nazionale (PRIN)”
Codice progetto: P20224AT3W
CUP: E53D23021840001
Contributo MUR per Ricerca: 221.595
Contributo UnivAQ: 106.890
Partners
- Università degli Studi dell’AQUILA, Responsabile: Gabriele Curci
- Consiglio Nazionale delle Ricerche (Coordinatore), Responsabile: Simone Lolli
Brief description and main objectives
In this project, we aim at improving the accuracy of estimating Planetary Boundary Layer Height (PBLH) simulated by atmospheric models using a machine learning approach. The basic underlying idea is to train different machine learning algorithms using data from the Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) on board the Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation mission satellite and integrated with the ground-based EUMETNET Profiling Programme automatic lidars and ceilometers (ALC) network (E-profile) observations.
The CALIOP data will provide us with PBLH estimates at a global scale but with a low temporal resolution, while on the contrary, the E-profile data will provide ground-based lidar and ceilometer PBLH retrievals at a very high temporal resolution but for a single location. ALC E-profile permanent observation sites are predominantly located in Europe. These observations are used during the training phase of the ML method for deriving a bias correction of PBLHs simulated by the fifth generation ECMWF reanalysis model (ERA5).
The results are daily maps of bias-corrected PBLHs with full horizontal coverage over Europe. We will test four types of ML algorithms: multiple linear regression (MLR), random forests (RF), gradient boosting (XGB), and multiple-layer perceptron networks (NN).
We will perform intercomparisons with independent PBLH ground-based lidar observations from the pan-European Aerosol, Clouds, and Trace Gases Research Infrastructure (ACTRIS) research infrastructure. The final stage of the project will demonstrate the benefit of an improved estimate of PBLH using simulation at higher spatial resolution at selected locations using the Weather Research and Forecasting model (WRF), correcting the PBLH calculated by the model with different PBL schemes and evaluating the benefit in terms of comparison with independent observations of meteorological quantities near the surface.
Main results and publications
To be completed.
