We compared the capacity of woody versus grassy vegetation covers to buffer high temperatures during heat waves by partitioning turbulent heat between latent (λE) and sensible (H) fluxes, and quantifying advection using the Priestley‐Taylor coefficient (α), for a16‐year old grassland and an adjoining 6‐year old plantation. We found that because λE dominated (>65%) the turbulent flux in the plantation and was at least twice as large as on the grassland (λE< 35% of the turbulent flux) during heat waves, the ambient temperature over the plantation was up to 5 °C lower in the afternoon, and averaged 1.2 °C lower for the whole day, compared with the grassland. Both vegetation covers emitted significant amounts of H when soil water availability was limited and also in winter when canopy cover was mostly inactive because of dormancy in the grassland and mutual shading in the plantation due to low solar angle. During the winter, advection of additional energy from surrounding vegetation suppressed λE and reduced α to <1.0 in both vegetation covers. Advection enhanced λE during periods of frequent rainfalls in summer, with mean α rising to 2.6 in the grassland and 3.4 in the plantation. Consistently low λE but high H made the grassland a source rather than a sink for advective energy, while the plantation was the opposite. The broadleaved evergreen woody vegetation consistently maintained a larger λE than the grassland in this mid‐latitude environment, contrary to the smaller λE observed over mostly coniferous forests at high (northern) latitudes (>35°). Annual evapotranspiration was 54% lower from the grassland (384 mm) than from the plantation (834 mm). Woody vegetation covers dominated by broadleaved species are therefore preferred for buffering extreme high temperatures during heat waves and recommended for rehabilitating degraded landscapes in urban areas. We also present functions for approximating α for soil water limited conditions.