Workshop on Analysis and Modeling of Automated Soil Respiration Measurements
September 11-12, 2007
Sponsor: Terrestrial Ecosystem Responses to Atmospheric and Climate Change (TERACC)
Conveners: Eric A. Davidson, Kathleen E. Savage, and Lindsey Rustad
Rationale: Respiration of terrestrial ecosystems is a major flux in the global carbon cycle and a potentially important mechanism of positive feedback to climate change. Respiration of roots and soil microorganisms constitutes 30-80% of total forest ecosystem respiration. While tremendous advances were made in the 20th century for characterizing the interacting effects of temperature, light, nutrients, and water in conceptual and numerical models of photosynthesis, respiration in ecosystems is typically still represented in most biogeochemical models by simple Q10 temperature functions that have been modified very little from their 19th century origins. The research community is well aware of the limitations of temperature functions: (1) the assumption of constant activation energies or temperature sensitivities of respiratory enzymes at all temperatures is incorrect; (2) variation in soil water content affects the diffusion of soluble substrates at low water content and the diffusion of oxygen at high water content, both of which can limit microbial respiration, (3) seasonal variation in soil water content is often confounded with the effects of temperature; (4) rapid changes in substrate availability that accompany wetting of dry soil, girdling of trees, and shading and clipping of grasses indicate the importance of substrate supply, independently of temperature.
With the advent of automated soil respiration measurement systems during the last decade, soil respiration is now being measured at high temporal resolution (e.g., hourly) by many researchers. Appropriate models that take full advantage of these emerging rich datasets may be able to help tease apart the interactions of temperature, moisture, and substrate supply on soil respiration rates at daily, weekly, seasonal, and annual scales. Such datasets with high temporal frequency also provides a wealth of information on the statistical distributions of thousands of respiration measurements. At the same time that this abundance of data affords many new possibilities for data analysis and modeling, it also presents new challenges of automating data quality assurance and quality control, determining appropriate statistical treatments, and applying effective modeling approaches. We can no longer visually inspect every soil respiration measurement profile, as was typically done with datasets of far fewer manual measurements. Similarly the assumptions underlying the use of ordinary least squares (OLS) regression can now be tested. Preliminary studies indicate that OLS model errors do not follow assumptions of normality and variance homogeneity and that a maximum likelihood approach may be warranted. Using the correct model fitting procedure is essential for parameter interpretation and for understanding the underlying process controls.
Objectives: The objectives of this 1.5 – 2.0 day workshop will be: (1) to present and evaluate data quality protocols for high frequency automated soil respiration data; (2) to examine appropriate statistical models for empirical fitting procedures; and (3) to explore the ways that automated soil respiration data can be best integrated into mechanistic models of ecosystem respiration and carbon dynamics.