Data on Spectral decomposition of high-frequency CO₂ concentration

Data was collected at different locations along the hillslope-riparian-stream transect in the Västrabäcken Catchment during 2015–2016. Västrabäcken Catchment is a 0.13-km2 forested subcatchment of the 68-km2 Krycklan Catchment (64°14′N, 19°46′E), located approximately 60 km northwest of Umeå, Sweden.

Stream and groundwater sampling was carried out according to a hillslope-riparian-stream sampling design, where observations were conducted at different locations along a 15-m transect following the assumed hydrological flow paths toward the stream. Measurements were taken in the hillslope podzol soils (15 m from the stream), in the riparian zone peat soils (1.5 m from the stream), and in the stream itself according to the design presented by Leith et al. (2015) and Campeau et al. (2018) using Vaisala CARBOCAP GMP221 nondispersive infrared (NDIR) CO2 sensor (range of 0–5%) enclosed with a water-tight, gas-permeable Teflon membrane (Johnson et al., 2010). At both riparian and hillslope locations, three groundwater tubes were installed, with the first two tubes used for groundwater sampling at 0- to 0.5-m and 0.5- to 1-m depth and the third groundwater tube, perforated along its entire 1-m length, was used for water table depth measurements (pressure transducer 1400, MJK Automation, Sweden) and groundwater temperature measurements (temperature sensors TO3R, TOJO Skogsteknik, Sweden). All sensors were connected to a data logger (CR1000, Campbell Scientific, USA) that recorded measurements at 60-min intervals. The system ran for 2 years (2015–2016) from snowmelt in early May until November. PAR was measured as part of the research infrastructure Integrated Carbon Observations System approximately 150 m from the soil-stream transect. The measurements were made at 1.1-m height using a SQ-110 Sun Calibration Quantum Sensor (Apogee Instruments Inc., USA).

Due to the need for simultaneous observations that are evenly distributed in time, the collected 2-year time series were screened for gaps caused by environmental constraints (mainly due to freezing of the stream water) and occasional instrumental malfunctions. By selection of gap-free periods, two time series consisting of 2,966 and 3,877 hourly observations were extracted covering the period of 10 June 2015 20:00 to 12 October 2015 09:00 and 16 May 2016 00:00 to 24 October 2016 12:00, respectively. These time series (hereafter denoted as 2015 and 2016) were used to calculate the PSDs as well as the wavelet power and coherence spectrum. In addition, several subsets of the time series (hereafter denoted time windows), with the length of 480 observations (i.e., 20 days), were extracted from the two time series. These time windows were obtained by sequentially shifting the first observation in the window by 120 observations (i.e., 5 days). This sliding window approach resulted in a set of partly overlapping time windows, 21 and 29 time windows for the Years 2015 and 2016, respectively, which subsequently were used to calculate the PSDs.

For further information, please see manuscript "Spectral Decomposition Reveals New Perspectives on CO2 Concentration Patterns and Soil-Stream Linkages" Riml et al (2019).

The dataset was originally published in DiVA and moved to SND in 2024.