An Improved Technique for Coupling Remote Sensing With Tower Based Carbon Flux Estimates

Eddy covariance system provides temporally continuous but spatially limited measurements of carbon flux (C-flux) from terrestrial ecosystems. On the other hand, remotely sensed imagery provides spatially continuous data that are temporally snapshots at best. A third way of estimating C-flux is to use process-based simulation models. This study is aimed at estimating the C-flux of Morgan-Monroe State Forest, a mixed hardwood deciduous forest in South Central Indiana, using multiple techniques in order to couple remotely sensed data with eddy covariance measurements. In addition to tower-based eddy covariance data, photosynthesis data from the Moderate-resolution Imaging Spectroradiometer (MODIS) sensor and outputs from Biome-BGC model simulation, we are collecting time series of hyperspectral data (AŸA›A›ƒ_sAªA.ƒ_onear-surfaceAŸA›A›ƒ_sAª? data) from the top of the tower. Also, we are collecting leaf area index (LAI) data using a Ceptometer along two transects radiating 100m northwest and southwest from the tower. An annual series of eight-day composite images from NASAAŸA›A›ƒ_sAªA›ƒ_zA›s MODIS sensor are also used to estimate image-based NPP of a 49 km AŸ’'A›ƒ,ªƒ__ 49 km area of the forest around the flux tower. The preliminary estimates from last yearAŸA›A›ƒ_sAªA›ƒ_zA›s (2002) eddy covariance, model result and MODIS imagery showed discrepancies among the outputs. We expect that the addition of AŸA›A›ƒ_sAªA.ƒ_onear-surfaceAŸA›A›ƒ_sAª? spectral data during the current year (2003) will enable us to bridge these discrepancies. Here we present a description of the AŸA›A›ƒ_sAªA.ƒ_onear surfaceAŸA›A›ƒ_sAª? spectral data collection system, its difficulties and rewards, and show some promising results in bridging the gap between AŸA›A›ƒ_sAªA.ƒ_ospectral vs. fluxAŸA›A›ƒ_sAª? realms using data from this yearAŸA›A›ƒ_sAªA›ƒ_zA›s growing season.