The effect of forest canopies and view angles on binary mapping of snow cover

Snow is an important component of the Earth's hydrologic and energy budgets and significant gains have been made in mapping snow properties using remote sensing. However, the presence of forests complicates mapping and monitoring of snow from space. The fraction of the snow under forest canopies that is visible from above, or the viewable gap fraction (VGF), limits the information content of satellite data with respect to snow. The VGF is dependent on viewing geometry, canopy structure and surface topography. Results from both hemispheric photos and a geometric optical (GO) model quantify decreases in VGF as the view zenith angle increases. Application of a GO model over a landscape shows that there is a substantial VGF decrease over the whole region when view zenith angle is increased from 0° (nadir) to 30°. Increasing the elongation in the shape of tree crowns increases the magnitude of the view zenith angle effect on VGF. Topography affects the VGF by changing local view zenith angle as a function of the slope and viewing direction.; The use of a geometric-optical radiative-transfer (GORT) model generally improves estimates of VGF relative to a simpler GO model by taking into account within-crown gaps. It particularly helps for sparse forests or forest canopies with low foliage densities. Airborne LIDAR data are useful for mapping forest cover and canopy structure parameters necessary to parameterize the GORT model, such as canopy height.; Results from an analytical hybrid GORT model show that the normalized difference snow index (NDSI), which is commonly used in snow mapping algorithms, decreases with increasing view zenith angle due primarily to decreasing VGF. For dense forests, NDSI decreases particularly rapidly as view zenith angle moves away from nadir and becomes negative for high view zenith angles. Observed relationships between NDSI calculated from MODIS and view zenith angle show patterns similar to model predictions. This result highlights the importance of forest structure on angular viewing effects in satellite data over snow and that taking into account forest structure and view angle effects will improve remote sensing of snow.