<p>Horizontal convective rolls (HCRs) are elongated, counter-rotating circulations in the atmospheric boundary layer. Accurately quantifying their orientation and cross-roll wavelength is essential for evaluating simulations against observations, testing theoretical models, and improving boundary-layer parameterizations. This study evaluates, refines, and combines statistical methods for estimating HCR properties. Three methods are considered: i) the primary mode of two-dimensional (2D) Fourier analysis, ii) the volume flux ratio (VFR), a simplified version of the mass flux ratio (MFR), and iii) autocorrelation fields, with a newly developed automated analysis of autocorrelation contour centers. Statistical results are evaluated against two LES cases: i) updraft bands with known orientation and wavelength enforced by heterogeneous surface heating, and ii) narrow-mode HCRs over a homogeneous surface. The primary Fourier mode captures overall roll orientation well but can misidentify cross-roll wavelength, showing the primary spectral mode does not always match the quasi-linear structure of interest. VFR provides satisfactory estimates of overall roll orientation but requires vertical velocity fields at fixed heights. The major axes of center autocorrelation contours provide satisfactory estimates of overall roll orientation only when these axes span sufficient along-roll distance. Given a satisfactory roll orientation as input, the automated procedure analyzing autocorrelation contour centers successfully reproduces expected cross-roll wavelengths. An advantage of analyzing autocorrelation fields is their applicability to operational radar scans with only scalar fields at fixed elevation angles. An example operational radar scan is analyzed to show the importance of noise-reduction preprocessing strategies for autocorrelation fields to suggest the expected roll orientation and cross-roll wavelength.</p>