Modeling Strategy for a Multi-Layer Mixed Forest Using Remote Sensing Data

Abstract

An important objective of the European Euroflux and Ecocraft projects (EU Fourth Framework programme) is the description of the carbon balance of temperate forests. A main focus point of the present study is the quantification of the net ecosystem productivity (NEP) of a mixed Scots pine (Pinus sylvestris) and oak (Quercus robur) forest in Brasschaat (Belgium), during a two year period (1996–1998) using remote sensing techniques and data integration in a multi-layer carbon cycling model (CARDYN). The forest is composed of five dominant species and three canopy layers, and our aim is to simulate at the reflectance and carbon flux levels, taking the vertical and horizontal spatial heterogeneity of the forest into account. Our strategy is based both on the comparison of information obtained with remote sensing of surface reflectance, with a spatial resolution of 1 km (NOAA-AVHRR sensor) and on the comparison of the modeled carbon fluxes with measured carbon fluxes of eddy covariance determined footprints, over a two years period. We focus on the temporal variations of reflectances in the visible (VIS) and near infrared (NIR) wavelength bands during the phenological cycle. Footprint dimensions are determined from a tower (41 m above the forest floor) equiped with eddy covariance analysis instrumentation. This allows for the calculation of CO₂ exchanges dynamics for a specific area (footprint area) on an hourly basis. An important technical problem is the co-localisation of the footprint area with the NOAA image pixel corresponding with the footprint. To address this problem, we are developing a Geographic Information System (GIS). For the issue of surface cover heterogeneity (multi-species, multi-layer cover types) we propose a modeling protocol based on the description of each forest patch around the tower, within a radius of 3 km. A first model validation is performed at the physical level by comparing simulated reflectances (in VIS and NIR) using a radiative transfer model for typical mixed forest (Scattering by Arbitrarily Inclined Leaves, SAIL model) with measured reflectances (NOAA-AVHRR sensor). If a good agreement is obtained this should justify, by inversion, the determination of CO₂ fluxes with remote sensing and its comparison with simulated CO₂ fluxes obtained with the CARDYN modeling approach.