CEILOMETER DERIVED 2D+TIME PLANETARY BOUNDARY LAYER HEIGHT VIA COMPRESSED SENSING

Abstract

We employ the Compressed Sensing (CS) technique for the fused 2D+time estimate of Planetary Boundary Layer (PBL) height as derived from Ceilometer backscatter profiles in tandem with model outputs from the Weather Research and Forecasting model coupled with Chemistry (WRF-Chem). The PBL is the lowest part of the atmosphere, and the only region of the atmosphere for which aerosols are present in abundance. Weather forecasting models including WRF-Chem often have large uncertainties in PBL height, and errors in estimation can greatly affect forecast skill as well as introduce derivative errors into pollution spread models. Ceilometers measure the aerosol profile through Lidar backscatter, which can be used to retrieve the height of the PBL with great accuracy. Data fusion is the process of combining the data from different sources to produce more accurate and consistent results. Although Compressed Sensing has been employed for data fusion in related areas, to the best of our knowledge we are the first to employ Compressed Sensing for the creation of air quality data products to combine surface retrievals with model forecasts. We evaluate the accuracy of the proposed CS technique using 4 Ceilometers installed and operating in the greater Baltimore-Washington corridor on the east coast of the United States. We find that the CS technique can estimate the spatial extent of PBL height with accuracy exceeding that of a baseline Gaussian Mixture Model (GMM) fusion technique.