Modeling Peak Season Net Radiative Forcing from Carbon Dioxide (CO2) and Methane (CH4) in Arctic Tundra Ecosystems

Changes in climate have the potential to significantly affect the net radiative forcing of arctic ecosystems by altering the ecosystem gas exchange of CO2 and CH4. Not only do CO2 and CH4 respond non-linearly to different environmental gradients and thresholds, but these gases also differ in their relative impact on net radiative forcing. Based on IPCC 2007 calculations using a 100-year time horizon, the global warming potential of CH4 is 25 times greater than CO2. As result of these differences, it is essential to understand the combined radiative forcing effect of both gases as a response to gradients in ecosystem properties in order to understand how ecosystem radiative forcing changes over space and time.

During the peak growing season of summer from 2005-2007, we visited 16 sites throughout the Beringia region in Chukotka, Wrangel Island and Alaska. Plot level measurements were recorded for three to four common land cover types to determine the controls on component fluxes (CO2, CH4). Data were used to create simple radiative forcing models based on parameters available from relatively inexpensive and readily available satellite remote sensing and automatic weather station platforms.

We constructed flux models using multiple regression, incorporating a subset of the following variables: air temperature, water table depth, active layer, soil moisture, normalized difference vegetation index (NDVI), photochemical reflectance index (PRI), water balance index (WBI), and normalized difference surface water index (NDSWI). Preliminary ecosystem flux models explained 70% and 61% of the variation in ecosystem CO2 and CH4 exchange, respectively.

When applied to large spatial areas, these models will likely improve our understanding of how spatial heterogeneity in arctic landscapes affects ecosystem CO2 exchange, CH4 exchange, and radiative forcing. Additionally, the simplicity of the model parameters increases the possibilities of using these models with existing and future datasets.