Using High Resolution LiDAR Data and a Flux Footprint Parameterization to Scale Evapotranspiration Estimates to Lower Pixel Resolutions

Over the last several decades the hydrologically sensitive Boreal Plains ecoregion of Western Canada has experienced significant warming and drying. To better predict implications of land cover changes on evapotranspiration (ET) and future water resources in this region, high resolution light detection and ranging and energy balance data are used here to spatially parameterize the Penman-Monteith ET model. Within a 5 km x 5 km area of peatland ecosystems, riparian boundaries, and upland mixed-wood forests, the influence of land cover heterogeneity on the accuracy of modeled ET is examined at pixel sizes of 1, 10, 25, 250, 500, and 1,000 m, representing resolutions common to popular satellite products (SPOT, Landsat, and MODIS). Modeled ET was compared with tower-based eddy covariance measurements using a weighted flux footprint model. Errors range from 10% to 36% of measured fluxes and results indicate that sensors with small pixel sizes (1 m) offer significantly better accuracy in large heterogeneous flux footprints, while a wider range of pixel sizes (<25 m) can be suitably applied to smaller homogeneous footprints. Mid (250 m) and coarse (>500m) pixel sizes offered significantly less accuracy, although changes in pixel size within this range offered comparable results.