Measurement-based upscaling of pan Arctic net ecosystem exchange: the PANEEx project

• Main Authors: Mbufong, Herbert Njuabe, Kusbach, Antonin, Lund, Magnus, Persson, Andreas, Christensen, Torben Røjle, Tamstorf, Mikkel P., Connolly, John
• Format: Report
• Language: English
• Published: 2015
• Subjects: Arctic; net ecosystem exchange; upscaling; light response curves; leaf area index; satellite remote sensing; CO2 fluxes; Sink-source interactions; Tundra
• Online Access: https://pure.au.dk/portal/en/publications/a6377b5a-065d-4ddf-b7c0-349f8cdc4a59

The high variability in Arctic tundra net ecosystem exchange (NEE) of carbon (C) can be attributed to the high spatial heterogeneity of Arctic tundra due to the complex topography. Current models of C exchange handle the Arctic as either a single or few ecosystems, responding to environmental change in the same manner. In this study, we developed and tested a simple NEE model using the Misterlich light response curve (LRC) function with photosynthetic photon flux density (PPFD) as the main driving variable. Model calibration was carried out with eddy covariance carbon dioxide data from 12 Arctic tundra sites. The model input parameters (fcsat, Rd and α) were estimated as a function of air temperature (AirT) and leaf area index (LAI) and represent specific characteristics of the NEE-PPFD relationship, including the saturation flux, dark respiration and initial light use efficiency, respectively. While remotely sensed LAI is readily available as a MODIS Terra product (MCD15A3), air temperature was estimated from a direct relationship with MODIS land surface temperature (MOD11A1, LST). Therefore, no specific knowledge of the vegetation type is required. Preliminary results show that the model captures some of the spatial heterogeneity of the Arctic tundra but could reasonably estimate NEE for 5 of the 12 sites used in the calibration of the model. However, the model effectively estimates NEE in three disparate Alaskan ecosystems (heath, tussock and fen). We suggest that the poor agreement may result from the disparity between ground-based measured LAI (used in model calibration) and remotely sensed MODIS LAI (used in NEE estimation). We propose recalibrating the model using the relationships between remotely sensed vegetation indices and LRC parameters. The results presented herein are only preliminary.