Modeling Soil CO2 Emissions From Ecosystems

We present a new soil respiration model, describe a formal model testing procedure, and compare our model with five alternative models using an extensive data set of observed soil respiration. Gas flux data from rangeland soils that included a large number of measurements at low temperatures were used to model soil CO₂ emissions as a function of soil temperature and water content. Our arctangent temperature function predicts that Q₁₀ values vary inversely with temperature and that CO₂ fluxes are significant below 0 °C. Independent data representing a broad range of ecosystems and temperature values were used for model testing. The effects of plant phenology, differences in substrate availability among sites, and water limitation were accounted for so that the temperature equations could be fairly evaluated. Four of the six tested models did equally well at simulating the observed soil CO₂ respiration rates. However, the arctangent variable Q₁₀ model agreed closely with observed Q₁₀ values over a wide range of temperatures (r² = 0.94) and was superior to published variable Q₁₀ equations using the Akaike information criterion (AIC). The arctangent temperature equation explained 16–85% of the observed intra-site variability in CO₂ flux rates. Including a water stress factor yielded a stronger correlation than temperature alone only in the dryland soils. The observed change in Q₁₀ with increasing temperature was the same for data sets that included only heterotrophic respiration and data sets that included both heterotrophic and autotrophic respiration.