estimate of growing season, FAPAR at chlorophyll level and LUE at chlorophyll level using scaled-up MODIS daily data and flux tower data
Qingyuan
Zhang, GEST/UMBC &, GSFC/NASA, qingyuan@umbc.edu
(Presenting)
Elizabeth
M.
Middleton, GSFC/NASA, elizabeth.m.middleton@nasa.gov
Guillaume
G.
Drolet, Université Laval, Canada, guillaume.drolet.1@ulaval.ca
Hank
A.
Margolis, Université Laval, Canada, hank.margolis@sbf.ulaval.ca
Alan
A.
Barr, Meteorological Service of Canada, alan.barr@ec.gc.ca
We investigated the use of daily “land band” satellite imagery from the MODerate resolution Imaging Spectroradiometer (MODIS) for estimating the seasonal and annual photosynthetic light use efficiency (LUE) of a deciduous boreal forest over a five year period. Our study site was the Southern Old Aspen (SOA) site in the Canadian Carbon Program’s flux network, located in Prince Albert, Saskatchewan. The surrounding area of SOA is not spectrally homogeneous at the MODIS nadir resolution of 500 m, and even more heterogeneity is introduced by off-nadir observations that may observe different ground sectors and which can be nine times as large as those for the nadir observations. In order to achieve spectral characteristics of similar ground sectors over the forest to compare with tower measurements, we scaled up the nominal 500 m MODIS products to achieve grid cells representing uniform ground area of 2.5 km x 2.5 km (5x5 MODIS 500 m grid cell area). We then used the 5x5 scaled-up MODIS products in a coupled canopy-leaf radiative transfer model, PROSAIL2, to estimate the fraction of photosynthetically active radiation absorbed (FAPAR) by the physiologically active sector of the SOA forest canopy dominated by photosynthetic pigments (designated as FAPARchl), versus that by the whole SOA forest (FAPARcanopy). From the tower measurements, we determined 90 minute averages for PAR, gross ecosystem production (GEP), and gross LUE centered on the MODIS overpass times for all available relatively cloud-free MODIS nadir and off-nadir observations during the 2001-2005 period. We also calculated the amount of PAR absorbed (APAR) and LUE for the physiologically active foliage (APARchl, LUEchl) and for the entire aspen forest (APARcanopy, LUEcanopy). The flux tower GEP measurements were strongly related to the MODIS-derived estimates of APARchl (r2 = 0.78) but weakly related to estimates of APARcanopy (r2 = 0.33). Gross LUE (slope of GEP:APAR relationship) between 2001 to 2005 for LUEchl was 0.0241 mol C mol-1 PPFD whereas gross LUE for LUEcanopy was substantially lower at 0.0155 mol C mol-1 PPFD. Estimates of the average growing season length over the five-year period were made performing derivative analyses on the temporal curves, which differed among the GEP and the three spectral indices examined (normalized difference vegetation index, NDVI; the land surface water index; LSWI; and the enhanced vegetation index, EVI). NDVI and LSWI provided similar season lengths (~106 days) to GEP while the EVI described a one week lag and shorter season. However, EVI matched the GEP mid-summer peak whereas NDVI and LSWI did not. Our results also showed that growing season LUEchl (mol C mol-1 PPFD) varied across years: 2001-2002, 0.0243±0.002; 2003, 0.0225; 2004, 0.0310; and 2005, 0.0267. Our five-year LUEchl average for the growing season matches well with recently published tower-based LUE estimates at SOA for both seasonal phase and amplitude (Krishnan et al. 2006). We expect that this approach to estimating forest LUEchl from MODIS observations should provide a promising input to land surface models and atmospheric general circulation models for improved carbon dynamics of ecosystems.
NASA Carbon Cycle & Ecosystems Active Awards Represented by this Poster: