Understanding landscape level carbon fluxes: an upscaling approach of two small scale models in an arctic lowland ecosystem, Alaska

Submitted by charlotte.schilt on Mon, 10/08/2018 - 15:37

Understanding arctic carbon exchange patterns is highly important because of its sensitivity in change and because the large amount of potential CO2 which is stored in the soil as organic matter. Key drivers are irradiance and Leaf Area Index (LAI) or vegetation specific characteristics. Other drivers can be the amount of ambient CO2, soil water constraints and air temperature. These key drivers plus air temperature form the base of two recently developed models that estimate Net Ecosystem Production (NEP) on a small scale (1m2). Both models are simple and exist of an Ecosystem Respiration (ER) part and a Gross Primary Production (GPP) part. The models differ in approach which can be based on vegetation types or on LAI.

What we want to know is how NEP fluctuates at landscape level in arctic lowland ecosystems. To learn about landscape level NEP fluctuations we compared these two small scale models with large scale eddy covariance measurements. We observed a large difference between measured and modeled CO2 flux. A period of five days forms an exception: here, measured CO2 flux and models coincide well at several vegetation types and leaf area indexes. The LAI based model is slightly more in favor compared with the vegetation type based model because its model performance is better at low leaf area.

Furthermore we looked for reasons of this large mismatch. We found that the ER part of the models is likely well predicted. We found that cumulative rainfall in the eddy covariance measuring year 2005 is 2.5 times less than in 2004 before and 3.8 times less than in 2003. This tells that soil water constraints are likely to be a reason for CO2 flux mismatch. Wind direction does not seem to influence CO2 flux measurements. A CO2 flux measurement footprint, determined by a combination of wind direction and wind speed, is addressed to specific vegetation types. This linking of footprints with area and its characteristics is an interesting exercise. Still, more analysis is necessary to be able to draw conclusions.

To conclude, it is clear that 2005 was an exceptional year in terms of rainfall. This becomes clear in landscape eddy covariance CO2 flux measurements. Water constraints are likely to be the main reason for the bad model performance. Adaptation of the models can help to overcome this unexpected bias. Variations thoughout the landscape are to be considered as possible causes of changes in CO2 flux measurements in further research. Linking footprints to area can be used for that. With these conclusions, describing large scale arctic lowland ecosystem carbon fluctuations has come both a step closer as a step further away.

Peter van Buuren


Mark van Wijk, Nico de Ridder