Katabatic winds strongly influence turbulent heat fluxes over glaciers, yet the turbulent length scales governing these fluxes remain poorly understood. The height of the wind speed maximum ( \(h_{\textrm{jet}}\) ) has been proposed as a relevant turbulent length scale for some katabatic flows, while scaling based on distance from the surface (law-of-the-wall scaling) may apply when \(h_{\textrm{jet}}\) exceeds the Obukhov length ( \(\varLambda \) ). However, these relations remain underexplored for glacier surfaces and katabatic flows more broadly. In this study, we use cospectral analysis of near-surface eddy covariance measurements and kite-borne wind profiles from a glacier site to identify dominant turbulent length scales and assess their relation with \(h_{\textrm{jet}}\) . We introduce a filtering method to remove non-turbulent contributions — such as internal gravity waves — from the cospectra, improving the reliability of length scale detection. For comparison, we investigate tower-based data from a non-glacierized site with deep katabatic flows. Our results reveal a one-to-one relation between the streamwise temperature flux length scale ( \(L_{uT}\) ) and the smaller of \(h_{\textrm{jet}}\) and \(\varLambda \) , defining two regimes: \(h_{\textrm{jet}}\) –scaling and \(\varLambda \) –scaling, with \(h_{\textrm{jet}}\) –scaling dominant at the glacier site. This relation indicates that it may be possible to detect katabatic jet height from near-surface observations from a single sensor. Neither site exhibited classical boundary layer scaling: turbulent mixing lengths scaled with distance from \(h_{\textrm{jet}}\) rather than height above the surface. These findings challenge conventional turbulence scaling assumptions for katabatic flows.
