Investigations of BOREAS spatial data in support of regional ecosystem modeling

Publisher Website
Google Scholar
Abstract
Simulation models are commonly used to scale up water, energy, and trace gas flux estimates from study plots to an entire region or biome, such as the North American boreal forest. As a means to validate ecosystem model predictions at the scale of the typical experimental area (100 m2 to 1 km2), it is necessary to quantify potential errors in spatial data layers used as model inputs, independently from prediction errors resulting from incorrect model design or flawed process algorithms. Our analysis of land cover, hydrology, and soil maps generated as part of the Boreal Ecosystem‐Atmosphere Study (BOREAS) suggests that coverages for plant functional types, derived from a combination of Landsat Thematic Mapper and the advanced very high resolution radiometer (AVHRR) satellite images, provide the highest quality information to define spatial model parameters. This information is critical for boreal ecosystem simulations of small patch types (e.g., dry conifer, fen, and disturbed sites), which are frequently obscured at 1‐km pixel resolution. We find that soil property information for the dominant class, as contained in the regional BOREAS soil maps, does not appear to be a highly consistent indicator of hydrologic dynamics for fen and other boreal wetlands at the study area level. It appears instead that accurate regional modeling analyses for methane and other biogenic trace gas fluxes will depend largely on relatively fine scale remote sensing to resolve the high level of undifferentiated pixel mixing seen among boreal forest types and fen/bog areas in aggregated 1‐km AVHRR land cover data.