Deriving GEP seasonality: issues posed by the absence of CO2 profile measurements
Restrepo-Coupe, University of Arizona, firstname.lastname@example.org
Saleska, University of Arizona, email@example.com
Rocha, University of São Paulo, firstname.lastname@example.org
Rafael, University of São Paulo, email@example.com
Brad, University of Arizona, firstname.lastname@example.org
It is widely understood that when the diurnal cycle of Net ecosystem exchange (NEE) is estimated in tall-stature ecosystems using eddy covariance methods, it is essential that the measurements account for changes in the amount of CO2 in the canopy air space (called the “storage flux”, Sco2). Less widely appreciated is the possibility that estimates of Sco2 may also be important for accurate characterization of the seasonal cycle of components of NEE, Gross ecosystem productivity (GEP) and ecosystem respiration (Reco). Unfortunately, at many EC sites in the Amazon, Sco2 time series are discontinuous or missing.
This study evaluates the importance of Sco2 for estimating GEP seasonality, and quantifies biases that rise in its absence, using data from four LBA-flux tower sites (Tapajós Km67, Santarém Km83, Reserva Jarú RJA, and Manaus Km34). Our goal is to establish a reliable dataset of tower-based GEP that can be used to calibrate and test remote sensing indices (the Enhanced Vegetation Index, EVI, from MODIS) and achieve an accurate integrated estimate of basin-wide photosynthetic flux. We found that in the absence of Sco2 observations, above-canopy flux measurements by themselves underestimate ‘true’ GEP and artificially alter the seasonal trends. Further, the amplitude of the annual cycle, and the onset of low and high GEP periods is affected differently at each site. We also evaluated various models for estimating Sco2in the absence of observations, in terms of their ability to recover the amplitude and phasing of the GEP annual cycle. We found that most of the Sco2-filling models were able to reduce error in the amplitude of the seasonal cycle and improve annual estimates of GEP in general. Moreover, we were able to determine a 'best' method at each site based on its ability to accurately model the above-mentioned seasonal cycle and the onset of low and high GEP periods.