Estimating daytime subcanopy respiration from conditional sampling methods applied to mufti-scalar high frequency turbulence time series

This study presents a new method to estimate daytime respiration from the subcanopy of forests directly from conventional eddy covariance (EC) measurements. The method primarily considers the respiration signal from root, litter and microbial respiration, which are known to be the main components of ecosystem respiration, R,, as well as decomposition of coarse and fine woody detritus, and respiration from low understory vegetation (forbs, herbs and grasses). The conceptual framework is based on the premise that upward moving air parcels carry a specific and unambiguous signal in their CO2 and water vapour composition, which can be separated and distinguished into respiration and photosynthesis. The model employed a combination of conditional sampling methods, quadrant analysis and relaxed eddy accumulation with hyperbolic deadbands to identify respiration events and to quantify their flux contribution. Datasets from five sites, most of which had multiple sampling heights, were selected to test this technique among contrasting ecosystems and canopy structures. Respiration signals were successfully identified in daytime data of all sites. A hyperbolic deadband of size H = 0.25 applied to the plane constructed from perturbations of carbon dioxide c and water vapour q densities effectively separated the signals of respiration from photosynthesis. The time fraction of respiration events was <= 10% during daytime. The global correlation coefficient r(c,q) was found to be a universal predictor of this time fraction and was therefore used as a filter to identify periods of a meaningful and extractable respiration signal. Coherent structures, defined as ramp-shaped flow pattern observed in time series in and above forest canopies, are likely to be the underlying transport mechanism for these respiration events. Daytime subcanopy R-e estimates derived from the new method agreed with those derived from: (i) the intercept of light-response curves and (ii) Soil CO2 efflux chambers for three of the five sites. Limitations were posed by the dense, multi-layered deciduous canopy and the intense vertical turbulent mixing at one coniferous site. In addition, refixation of respired CO2 by the understorey (CO2 recycling) may cause an underestimation of daytime R-e or pose a limitation to the method proposed here. An indicator relating the canopy shear length scale, L-s, to the adjustment length scale, L-d, was proposed to predict the skill of the new method, and found to be useful in four of the five sites. Analysis of vertical coupling in the plant canopy using exchange regimes could explain the failure of the new method for the remaining site. (C) 2008 Elsevier B.V. All rights reserved.