Summary: | The carbon budgets of the atmosphere and terrestrial ecosystems are closely coupled by vertical gas exchange fluxes. Uncertainties remain with respect to high latitude ecosystems and the processes driving their temporally and spatially highly variable methane (CH4) exchange. Problems associated with scaling plot measurements to larger areas in heterogeneous environments are addressed based on intensive field studies on two nested spatial scales in Northern Siberia. CH4 fluxes on the microsite scale (0.1–100 m2) were measured in the Lena River Delta from July through September 2006 by closed chambers and were compared with simultaneous ecosystem scale (104–106 m2) flux measurements by the eddy covariance (EC) method. Closed chamber measurements were conducted almost daily on 15 plots in four differently developed polygon centers and on a polygon rim. Controls on CH4 emission were identified by stepwise multiple regression. In contrast to relatively low ecosystem-scale fluxes controlled mainly by near-surface turbulence, fluxes on the microsite scale were almost an order of magnitude higher at the wet polygon centers and near zero at the drier polygon rim and high-center polygon. Microsite scale CH4 fluxes varied strongly even within the same microsites. The only statistically significant control on chamber-based fluxes was surface temperature calculated using the Stefan–Boltzmann equation in the wet polygon centers, whereas no significant control was found for the low emissions from the dry sites. The comparison with the EC measurements reveals differences in controls and the seasonal dynamics between the two measurement scales, which may have consequences for scaling and process-based models. Despite those differences, closed chamber measurements from within the EC footprint could be scaled by an area-weighting approach of landcover classes based on high-resolution imagery to match the total ecosystem-scale emission. Our nested sampling design allowed for checking scaling results against measurements and to ...
|