“...we bring together theory and observations, this work is a synthesis of knowledge.”

The eddy covariance technique is used at research sites around the world (e.g., FLUXNET) to measure the surface-atmosphere exchange of CO₂. Our paper offers some simple relationships by which the random uncertainty in these measurements can be estimated. Information about this uncertainty is needed so that statistically rigorous confidence intervals on the fluxes (particularly annual flux sums, indicating the year’s carbon sequestration) can be calculated and reported.

Our results are also especially important as eddy flux data are increasingly being used to constrain large-scale carbon cycle models using "data-model fusion." This technique requires information about measurement uncertainties. We have less confidence in data with larger uncertainties, so we weight observations with larger uncertainties less than those with smaller uncertainties. With our results, this weighting can be done in an objective manner.

Because we bring together theory and observations, this work is a synthesis of knowledge. More than anything, though, we present useful results that have direct application to the quantitative carbon-cycle analyses which are currently popular.

Our results allow us to specify the uncertainty in our measurements of the rates at which vegetation takes up CO₂ from the atmosphere during the day (photosynthesis) and releases CO₂ back to the atmosphere at night (respiration). Over the course of the year, the balance between these two fluxes represents the carbon sequestration.

We can now specify upper and lower limits (our "confidence interval") on our estimates of the annual carbon sequestration. This research is important given increasing concerns (reported in the mass media as well as the scientific literature) about the effects of rising atmospheric CO₂ on the climate system.

My coauthor, David Hollinger of the USDA Forest Service, and I realized that we needed this information about the flux measurement uncertainty to do the kinds of analyses we wanted to do. We’d already done some related work (Hollinger & Richardson, Tree Physiology, 2005), using data from just our own research site in Maine, but we realized that we could extend the analysis to a wider range of sites, and show that the patterns we had already documented were robust and in agreement with theoretical predictions.

The biggest problem along the way was that our manuscript was rejected by the first journal we sent it to—one reviewer suggested it was "much ado about nothing." That was a great disappointment, but it made us realize that we needed to frame the paper better, so that the practical significance was more clearly stated. So the reviewer’s negative comments ultimately resulted in a far more accessible paper.

My career to date has followed some strange twists and turns (I was at one point enrolled in a Ph.D. program in economics!), so it's hard to predict where I might be in five or ten years. As the length of our CO₂ flux time series increases, we have the opportunity to analyze the data with a whole different set of questions in mind, so I can see this being an important area of research for quite some time.

A more recent interest is in the field of phenology, which has to do with the timing of seasonal life cycle events such as flowering and budburst by plants, and migration and breeding by animals. Phenology is a sensitive indicator of climate change, and earlier leaf-out and delayed autumn senescence has important implications for the carbon cycle. I am also interested in getting more into feedbacks between vegetation and climate, particularly with regard to albedo, the Earth’s reflectance of the Sun’s radiation back to space.

Only to the extent that CO₂ flux measurements and carbon cycle models are being used to inform policy decision-making, particularly with regard to carbon accounting and greenhouse gas mitigation.