Fusion and Visualization of Planetary Boundary Layer Height from Ceilometer and WRF-Chem

Abstract:

We present a novel data fusion mechanism and 2D+T visualization for the Planetary Boundary Layer Height (PBLH) that combines Ceilometer profiles with WRF-Chem model outputs. The Planetary Boundary Layer is the lowest part of the Atmosphere and a region in which all aerosols are concentrated. As determination of the PBLH is an important problem both for the modeling of pollution spread as well as to improve the forecast skill of global circulation. However, at present, model forecasts of PBLH contain large uncertainties often ranging in hundreds or even low thousands of meters. Ceilometers can measure precisely the PBLH due to aerosol backscatter, but the most common form of ceilometers produce only point measurements and cannot model the spatial variability of the PBLH which is highly sensitive to variation in terrain, surface drag, and other quantities. As such, ways to combine model outputs with ceilometer retrievals are an area of scientific emphasis.

Our approach fuses celiometer derived PBLH retrievals with WRF-Chem model outputs using Compressive Sensing. We compare our resulting fusion against a baseline Gaussian Mixture fusion algorithm, and validate the resulting data fusion using withheld measurements of the PBLH from ceilometer profiles. The ceilometer retrievals are distributed over a recently developed prototype ceilometer network spanning the greater Baltimore / Washington D.C. corridor, and the retrieval algorthims combine traditional wavelet techniques with deep learning predictions. The demonstrated visualization is interactive and publicly web accessible, and includes information from the U.S. Census Bureau TIGR database for georectification. We demonstrate that compressive sensing can fuse the ceilometer derived retrievals with the model outputs to produce high quality PBLH animations. As such we consider fusion of ceilometer profiles with WRF-chem model outputs to be a viable approach to create regional and potentially national coverage maps of PBLH at sub-hourly timescales.