Effects of ICESat-2 Vegetation Product Spatial Sampling Rate on Satellite-Derived Momentum Aerodynamic Roughness Fields

Land surface model (LSM) parameterizations require information about vegetation roughness for purposes of characterizing fluxes of momentum, heat, and other scalars between the surface and atmosphere. It is feasible to estimate normalized momentum aerodynamic roughness for vegetation with remotely sensed leaf-area index alone; however LSMs ingest absolute roughness length and zero-plane displacement, which necessitates the use of vegetation height information. Spaceborne LIDAR presents the best data source for contemporary estimates of vertical vegetation structure at the regional or global scale at which LSMs run.

The ATLAS/ICESat-2 Land and Vegetation Product (ATL08) provides along-track terrain heights and canopy height metrics on a near-global basis with a nominal 91-day repeat cycle. Early ATL08 releases (i.e., data versions) included this height information at the scale of a processing segment only; this corresponds to a 100-meter along-track sampling distance. R005, the most recent ATL08 data version, provides the same height information as in previous releases, while adding 20-meter versions of the 98% relative canopy height metric (RH98) and terrain height. The RH98 canopy height is considered to be the best estimate of the true canopy height over a sampled ground surface distance.

Here, we show how the difference in sampling rate affects gridding of along-track RH98 data, and how that propagates into scaled roughness fields. Our spatial domain is the conterminous U.S., although our analysis here focuses mainly on a limited number of flux tower sites. The effects of quintupling the along-track sampling rate do not necessarily translate into consistent improvements in canopy and terrain height estimates. Atmospheric effects, signal-to-noise ratio, and differences in surface types can all affect the quality of height retrievals and derived datasets.