Atmospheric boundary layer height estimation using a Kalman filter and a frequency-modulated continuous-wave radar

Abstract

An adaptive solution based on an Extended Kalman Filter (EKF) is proposed to estimate the Atmospheric Boundary-Layer Height (ABLH) from Frequency-Modulated Continuous-Wave (FMCW) S-band weather-radar returns. The EKF estimator departs from previous works, in which the transition interface between the Mixing-Layer (ML) and the Free-Troposphere (FT) is modeled by means of an erf-like parametric function. In contrast to lidar remote sensing where aerosols give strong backscatter returns over the whole ML, clear-air radar reflectivity returns (Bragg scattering from refractive turbulence) shows strongest returns from the ML-FT interface. In addition, they are corrupted by “insect” noise (impulsive noise associated with Rayleigh scatter ing from insects and birds), all of which requires a specific treatment of the problem and the measurement noise for the clear-air radar case. The proposed radar-ABLH estimation method uses: (i) a first pre-processing of the reflectivity returns based on median filtering and threshold-limited decision to obtain “clean” reflectivity signal, (ii) a modified EKF with adaptive range intervals as time tracking estimator, and (iii) ad-hoc modelling of the observation noise covariance. The method has successfully been implemented in clear-air, single-layer, convective boundary layer conditions. ABLH estimates from the proposed radar-EKF method have been cross-examined with those from a collocated lidar ceilometer yielding a correlation coefficient as high as rho = 0.93 (mean signal-to-noise ratio, SNR = 18 (linear units), at the ABLH) and in relation to the classic threshold method.