Multi-Scale Observations of Urban Surface-Atmosphere Energy Exchange: Representativeness, Sensitivity, and Measurement Uncertainty

Urban processes are poorly represented in NWP and climate models due to heterogeneous surfaces, complex morphology, and multi-scale interactions, leading to errors in surface-atmosphere energy exchange across spatial and temporal scales. Satellite derived urban fluxes also face similar issues from coarse resolution and sub-pixel heterogeneity. Addressing this requires multi-scale, multi-height observations, which are rare and only a few urban sites worldwide provide such comprehensive long-term data. This study investigates urban energy fluxes in Enschede, The Netherlands using a net-radiometer (at micro-scale) with an eddy-covariance (EC, at neighborhood-scale) and a Scintillometer (SC, at city-scale) for one year in 2020. Enschede is a city in Eastern Netherlands and has a temperate oceanic climate, an area dominated by built-up (up to 84%), and predominantly compact, mid-rise local climate zone (LCZ-2). For the year 2020 this study examines annual variability, consistency, and representativeness of turbulent fluxes across scales, evaluates uncertainties from methodological choices, and assesses flux sensitivity to key input parameters.

The results show that turbulent fluxes, sensible (Q_H) and latent heat flux (Q_E), generally follow the seasonal and diurnal patterns of net radiation (Q^*). Observed EC fluxes at two heights show consistent vertical divergence due to the flow distortion caused by the building on which the EC instruments are installed and consequent roughness sublayer effects. Q_H estimates from EC tower and Scintillometer show comparable magnitudes, whereas Q_E derived from Scintillometer is approximately twice that obtained from EC indicating differences in sampled footprint.  Our results show that, in urban areas accounting for stability-dependent effective height is critical while processing Scintillometer data, as neglecting this factor introduces substantial errors (up to 150 Wm^-2 in Q_H and 80 Wm^-2 for Q_E under unstable conditions). This issue is particularly important because the associated uncertainty increases with unstable stratification, which often coincides with the overpass times of many satellites. Turbulent fluxes estimated by the Scintillometer depend strongly on two parameters: surface roughness (z₀) and displacement height (z_d). Our results show that these affect the results in opposite ways with higher z₀ increases fluxes, while higher z_d decreases them. Different aerodynamic parameterizations used to estimate these parameters often make comparable errors in both, which cancel out and give reasonable flux values. However, this result is site-specific and cannot be relied on in general. In practice, z₀ mainly affects friction velocity (u_*) and has a smaller impact when measurements are taken well above the surface. In contrast, z_d directly influences flux estimates and has a much stronger effect. Therefore, estimating z_d accurately is more important than estimating z₀. This long-term dataset supports urban process modeling, satellite product validation, and offers guidance on uncertainty and sensitivity for designing and interpreting future urban flux measurements.