Improved wavelet method for accurate high-resolution ecosystem flux estimation and time-derivative analysis

We present two major advancements to the wavelet-based, non-stationary flux estimation method of Destouet et al. (2025), enabling accurate calculation of high-resolution (1-minute) ecosystem fluxes and their time-derivatives.

We introduce first a scale-dependent estimation process that explicitly accounts for frequency-dependent eddy correlation times. By assigning different averaging times to each frequency band, we enhances the isolation of turbulent scales, reduces flux estimation errors, and improves the separation of local turbulence from larger scales by eliminating spurious correlations around the 'spectral gap'. This advancement is particularly valuable for wavelet-based flux partitioning, as it preserves high-quality flux estimates while retaining small-scale eddies, such as those hypothesized to transport soil respiration through forest canopies.

Second, our method now enables the computation of flux time-derivatives, allowing analysis of turbulent transport dynamics and ecosystem responses to environmental changes. As a first application, we present how to optimally determine the averaging time required for observed turbulent fluxes to represent underlying ecosystem fluxes. This is achieved by analysing the co-variation of flux time-derivatives with variables such as incoming radiation and carbon storage, which reflect underlying ecosystem dynamics.

These improvements together refine high-resolution flux estimation and unlock new opportunities to investigate ecosystem dynamics from flux towers. They have been implemented in the open-source TurbulenceFlux.jl package, which is readily available for community use.

Reference:

Destouet G, Besic N, Joetzjer E, and Cuntz M (2025) Turbulent transport extraction in time and frequency and the estimation of eddy fluxes at high resolution, Atmospheric Measurement Techniques 18(13):3193–3215, doi:10.5194/amt-18-3193-2025