For many natural and agricultural landscapes, vegetation partially covers the ground surface, resulting in significant variations in soil heat flux between interspace areas and underneath vegetation. This is particularly apparent in arid and semiarid regions where vegetation cover is low and clustered or 'clumped' with large areas of exposed soil. Surface heterogeneity presents significant challenges to the use of standard micro-meteorological measurement techniques for estimating surface energy balance components. The objective of this study was to use an array of 20 soil heat flux plates and soil temperature sensors to characterize the spatial and temporal variability in soil heat flux as affected by vegetation and micro-topographic effects of mesquite dunes in the Jornada Experimental Range in southern New Mexico. Maximum differences in soil heat flux among sensors were nearly 300 W m^-2. Maximum differences among individual sensors under similar cover conditions (i.e., no cover or interdune, partial or open canopy cover and full canopy cover) were significant, reaching values of 200-250 W m^-2. The 'area-average' soil heat flux from the array was compared with an estimate using three sensors from a nearby micro-meteorological station. These sensors were positioned to obtain soil heat flux estimates representative of the three main cover conditions: namely, no cover or interdune, partial or open canopy cover, and full canopy cover. Comparisons between the array-average soil heat flux and the three-sensor system indicate that maximum differences on the order of 50 to nearly 100 W m^-2 are obtained in the early morning and mid-afternoon periods, respectively. These discrepancies are caused by shading from the vegetation and micro-topography. The array-derived soil heat flux also produced a significantly higher temporal varying soil heat flux/net radiation ratio than what has been observed in other studies under more uniform cover conditions. Results from this study suggest that, to determine the number and location of sensors needed for estimating area-average soil heat flux in this type of landscape, one needs to account not only for the clustering of the vegetation cover but also micro-topography. Published by Elsevier Science B.V.