Constraining Long Term Carbon Exchange over an Old-Growth Forest in Complex Terrain with ACASA

Our understanding of terrestrial CO2 fluxes comes primarily from flux towers which measure CO2, H2O, and energy exchange between the vegetated surface and the atmosphere with the eddy covariance (EC) technique. However flux measurements are limited in their real world application by atmospheric conditions and heterogeneous land use. Given that hundreds of towers across the globe directly measure biospheric CO2 exchange, the question then remains, why is uncertainty in the land CO2 fluxes so high? Flux towers suffer from logistical problems including instrument failure and power outages which lead to data gaps as well as random and systematic errors inherit to the EC technique: rain, lack of turbulence, inhomogeneous terrain, and inhomogeneous vegetation lead to erroneous flux estimates. These periods of missing or erroneous data must be gap-filled in order to estimate whether an ecosystem is a net annual carbon sink or source. Current gap-filling techniques are inadequate and over the course of a year can cause estimates to be off by 30-50 g C m^-2 yr^-1 or more. For an ecosystem that is a small annual sink or source of carbon (typical for a disturbed forest, grassland, or savannah ecosystem), this uncertainty is then of the same magnitude as the measured flux! In this study we describe an approach to gap filling that adds an advanced soil-canopy-atmosphere model to standard procedures to reduce the uncertainty in the biospheric CO2 flux. Wind River is one most complex canopies when compared to the majority of flux tower sites. The over height and bottom heavy leaf area distribution typically of coniferous old-growth forests still make it a unique and challenging flux tower site. We present continuous results for the past 15 years with an emphasis on error bias of long term carbon exchange estimates from flux measurements. Site characteristics include very stable nighttime conditions that lead to a 70% rejection rate by u* criteria for nighttime data. The overall footprint for the daytime flux measurements is flat but the topography in the surrounding mountains is highly complex, leading to Mountain-Valley circulations and extended heterogeneous nighttime footprints. While standard techniques have been deployed at this site to correct this nighttime bias we investigate the use of a highly sophisticated high order closure model to constrain the long term data set.