In spring, when the mountainous snow cover becomes patchy, the strong multi-scale surface heterogeneity influences atmospheric heat exchange processes. At valley scale, thermally driven winds advect warm air towards snow-covered regions at higher elevations. Additionally, the difference in albedo between bare ground and adjacent snow patches causes strong surface temperature differences on a (sub-)meter scale. Consequently, the structure of the near-surface boundary layer is spatio-temporally highly variable. During spring 2021, we recorded data in a comprehensive field campaign in the Dischma valley close to Davos (CH). The dataset includes multiple eddy-covariance measurements at different measurement heights. The topographic setting allows to group the measurements into up valley and down valley flow systems. Using a multi-resolution spectral decomposition of turbulent flux variables, we investigate the structure and dynamics of near-surface turbulence. During up valley flows, the advection of warm air induces stable internal boundary layers (SIBL) adjacent to snow patches with near-neutral static stability above. The strong stability withing the SIBL eventually leads to decoupling of the near-surface turbulence from the submeso-scale motions aloft. In contrast, during down valley flows, stability dampens turbulence similarly at all measurement heights. In concert with those findings, measurements utilizing the IR-screen setup during the same campaign yield high spatio-temporally resolved air temperature profiles during phases of different stability. Furthermore, the screen measurements visualize the near-surface boundary layer dynamics. In the next step, we will use the gained process understanding to test new physical parameterizations especially for warm air advection in hectometer scale coupled snow-atmosphere models.