Biosphere-atmosphere exchange of CO ₂ in relation to climate: a cross-biome analysis across multiple time scales

Abstract. The biosphere-atmosphere flux of CO2 responds to climatic variability at time scales from seconds to years and longer. Quantifying the strength of the interaction between the flux and climate variables at multiple frequencies is necessary to begin understanding the climatic controls on the dynamics of the terrestrial carbon cycle. Orthonormal wavelet transformation (OWT) can quantify the interaction between flux and microclimate at multiple frequencies while expressing time series variance in few energetic wavelet coefficients, offering a low-dimensional view of the measured climate-flux interaction. The variability of the net ecosystem exchange of CO2 (NEE), gross ecosystem productivity (GEP) and ecosystem respiration (RE), and their co-variability with dominant climatic drivers, are explored with a global dataset consisting of 253 eddy covariance research sites. Results demonstrate that the NEE and GEP wavelet spectra are similar amongst plant functional types (PFT) at weekly and shorter time scales, but significant divergence appeared among PFT at the biweekly and longer time scales, at which NEE and GEP are relatively less variable than climate. The RE spectra rarely differ among PFT across time scales. On average, RE spectra had greater low frequency (monthly to interannual) variability than NEE, GEP and climate. The low frequency Fourier coefficients of eight sites with more than eight years of data were compared against CANOAK ecosystem model simulations. Both measurements and theory demonstrate that "multi-annual" spectral peaks in flux may emerge at low (4+ years) time scales. Biological responses to climate and other internal system dynamics provide the likely explanation for observed multi-annual variability, but data records must be lengthened and measurements of ecosystem state must be made, and made available, to disentangle the mechanisms responsible for these patterns.