Satellite-based measurements of vegetation canopy structure have been in common use for the last decade but have never been used to estimate canopy's impact on wind sheltering of individual lakes. Wind sheltering is caused by slower winds in the wake of topography and shoreline obstacles (e.g. forest canopy) and influences heat loss and the flux of wind-driven mixing energy into lakes, which control lake temperatures and indirectly structure lake ecosystem processes, including carbon cycling and thermal habitat partitioning. Lakeshore wind sheltering has often been parameterized by lake surface area but such empirical relationships are only based on forested lakeshores and overlook the contributions of local land cover and terrain to wind sheltering. This study is the first to examine the utility of satellite imagery-derived broad-scale estimates of wind sheltering across a diversity of land covers. Using 30 m spatial resolution ASTER GDEM2 elevation data, the mean sheltering height, h_s, being the combination of local topographic rise and canopy height above the lake surface, is calculated within 100 m-wide buffers surrounding 76,000 lakes in the U.S. state of Wisconsin. Uncertainty of GDEM2-derived h_s was compared to SRTM-, high-resolution G-LiHT lidar-, and ICESat-derived estimates of h_s, respective influences of land cover type and buffer width on h_sare examined; and the effect of including satellite-based h_s on the accuracy of a statewide lake hydrodynamic model was discussed. Though GDEM2 h_s uncertainty was comparable to or better than other satellite-based measures of h_s, its higher spatial resolution and broader spatial coverage allowed more lakes to be included in modeling efforts. GDEM2 was shown to offer superior utility for estimating h_s compared to other satellite-derived data, but was limited by its consistent underestimation of h_s, inability to detect within-buffer h_s variability, and differing accuracy across land cover types. Nonetheless, considering a GDEM2 h_s-derived wind sheltering potential improved the modeled lake temperature root mean square error for non-forested lakes by 0.72 °C compared to a commonly used wind sheltering model based on lake area alone. While results from this study show promise, the limitations of near-global GDEM2 data in timeliness, temporal and spatial resolution, and vertical accuracy were apparent. As hydrodynamic modeling and high-resolution topographic mapping efforts both expand, future remote sensing-derived vegetation structure data must be improved to meet wind sheltering accuracy requirements to expand our understanding of lake processes.