SEARCH

SEARCH BY CITATION

Keywords:

  • CO2 sensors;
  • model performance;
  • rain pulse;
  • soil respiration;
  • temperature independent;
  • time series analysis;
  • wavelet analysis;
  • wavelet coherence;
  • wireless sensors

Abstract

Ecosystem processes are influenced by weather and climatic perturbations at multiple temporal scales with a large range of amplitudes and phases. Technological advances of automated biometeorological measurements provide the opportunity to apply spectral methods on continuous time series to identify differences in amplitudes and phases and relationships with weather variation. Here we used wavelet coherence analysis to study the temporal covariance between soil CO2 production and soil temperature, soil moisture, and photosynthetically active radiation (PAR). Continuous (hourly average) data were acquired over 2 years among three vegetation types in a semiarid mixed temperate forest. We showed that soil temperature and soil moisture influence soil CO2 production differently at multiple periods (e.g. hours, days, weeks, months, years), especially after rain pulse events. Our results provide information about the periodicity of soil CO2 production among vegetation types, and provide insights about processes controlling CO2 production through the study of phase relationships between two time series (e.g. soil CO2 production and PAR). We tested the performance of empirical models of soil CO2 production using the continuous wavelet transform. These models, built around soil temperature and moisture, failed at multiple periods across the measured dates, suggesting that empirical models should include other factors that regulate soil CO2 production at different temporal scales. Our results add a new dimension for the analysis of continuous time series of biometeorological measurements and model testing, which will prove useful for analysis of increasing sensor data obtained by environmental networks.