HR: 17:45h
AN: B22E-08 [PDF]
TI: Matching MODIS Products to Flux Towers: the first step in bottom-up scaling
AU: * Schmid, H
EM: hschmid@indiana.edu
AF: Indiana University, Dept. of Geography, Atm. Sc.
701 E. Kirkwood Ave., Bloomington, IN 47405 United States
AU: Wayson, C
EM: cwayson@indiana.edu
AF: Indiana University, Dept. of Geography, Atm. Sc.
701 E. Kirkwood Ave., Bloomington, IN 47405 United States
AU: Heinsch, F
EM: faithann@ntsg.umt.edu
AF: University of Montana, School of Forestry
437 Science Complex, Missoula, MT 59812 United States
AU: Running, S W
EM: swr@ntsg.umt.edu
AF: University of Montana, School of Forestry
437 Science Complex, Missoula, MT 59812 United States
AB:
Bottom-up scaling of tower based ecosystem fluxes to a large region involves several steps. In essence, the bottom-up
approach to scaling constitutes a defensible strategy to fill the space between a set of in-situ observational nodes (i.e.,
the flux towers) with an estimate of the exchange that is matched to measured values at the nodes. To ensure that the
space-filling process is responsive to variations of biophysical parameters related to land-cover and ecosystem type, the
scaling strategy uses a suitable ecosystem exchange model as its aggregation tool. Here, we apply a bottom-up scaling
strategy to gross photosynthetic exchange of carbon dioxide (GPE), and use MODIS derived 8-day composites at a 1 km
resolution as the aggregation tool. As a first step in the scaling strategy, the MODIS derived GPE composites must be matched
to corresponding estimates from the flux towers, to root them on the flux towers as their observational benchmarks. This
paper addresses problems and issues associated with matching MODIS products to flux tower derived GPE at the hand of 7 km x 7
km MODIS product subsets centered on AmeriFlux towers in Indiana (MMSF~flux) and Michigan (UMBS~flux).
Waypoints along our route to achieve matching include, (i) separation of directly measured net ecosystem exchange fluxes into
ecosystem respiration and GPE; (ii) high resolution assessment of vegetation indexes (VI) in the area around the flux tower
likely to be covered by the flux footprint, based on IKONOS or Landsat scenes. (iii) The high resolution VI will be overlaid
with computed flux footprints to examine the area-to-area representativeness of flux measurements over various time scales.
In particular, we will examine whether the averaging power of the spatially evolving flux footprint over an 8-day integration
period (matching the MODIS time scale) is usually sufficient to provide fluxes with acceptable spatial representativeness to
serve as a benchmark for MODIS products. (iv) We will compare the footprint weighted integrations of vegetation index
drivers (e.g., NDVI, LAI) to those of the 49 MODIS breakout "pixels" of the 7 km x 7 km subsets. Because of residual
uncertainty in the geopositioning of the MODIS pixels, and variations in the flux footprint location, it is not certain which
of the 49 breakout elements matches the area in the composite tower flux footprint best. Is it always the same MODIS pixel
that achieves the best match? (v) Based on the foregoing, we will examine the ratio of the MODIS derived flux to the measured
flux under the condition that the footprint weighted vegetation index matches that of MODIS (i.e., by using a selection of
data, based on given representativeness criteria). Finally, this analysis will allow us to derive a regression function to
calibrate the MODIS derived fluxes to a spatially representative subset of tower flux measurements.
DE: 0315 Biosphere/atmosphere interactions
DE: 0330 Geochemical cycles
DE: 0400 Biogeosciences
DE: 1640 Remote sensing
SC: Biogeosciences [B]
MN: 2003 Fall Meeting