Spectral decomposition reveals new perspectives on CO2 concentration patterns and soil‐stream linkages

The rapid development of novel technologies to obtain high-frequency observations has provided new possibilities to observe and understand carbon cycling in inland waters. This study investigates carbon dioxide (CO2) dynamics along a boreal soil-stream transect using a state-of-the-art data set in combination with a spectral methodology to identify controls on stream CO2. The spectral decomposition of hourly observations revealed intermediate (multiday) to long-term (monthly) patterns across the upslope-riparian-stream continuum, with similar power law increases in CO2 concentration fluctuations with increasing period. High-frequency CO2 variabilities, specifically diel CO2 concentration fluctuations, were also identified at all locations but were substantially amplified in the stream compared to in the riparian groundwater. Moreover, the spectral coherence between soil and stream CO2 fluctuations was inconsistent and restricted to episodic events. In contrast, we found a strong and consistent spectral coherence between the riparian groundwater level and stream CO2 concentration, indicating a hydrological control on stream CO2 dynamics. However, during some time spans even these patterns were obscured, suggesting that additional processes, such as CO2 evasion and in-stream metabolism, modulated the influence from riparian sources. The scales and patterns of temporal coherences (or lack thereof) between CO2 at different points in the catchment, as well as with other factors, for example, groundwater levels and Photosynthetically Active Radiation, provide new perspectives on the range of processes governing stream CO2 dynamics. Thus, this study highlights the potential of using spectral decomposition of high-resolution, spatially distributed data of different types to investigate biogeochemical transformations and pathways linking terrestrial and aquatic systems.