Drivers of carbon fluxes in Alpine tundra: a comparison of three empirical model approaches
In high mountains, the effects of climate change are manifesting most rapidly. This is especially critical for the high-altitude carbon cycle, for which new feedbacks could be triggered. However, mountain carbon dynamics is only partially known. In particular, models of the processes driving carbon fluxes in high-altitude grasslands and Alpine tundra need to be improved. Here, we propose a comparison of three empirical approaches using systematic statistical analysis, to identify the environmental variables controlling $CO_2$ fluxes. The methods were applied to a complete dataset of simultaneous in situ measurements of the net $CO_2$ exchange, ecosystem respiration and basic environmental variables in three sampling sites in the same catchment. Large year-to-year variations in the gross primary production (GPP) and ecosystem respiration (ER) dependences on solar irradiance and temperature were observed,. We thus implemented a multi regression model in which additional variables were introduced as perturbations of the standard exponential and rectangular hyperbolic functions for ER and GPP, respectively. A comparison of this model with other common modelling strategies, showed the benefits of this approach, resulting in large explained variances (83% to 94%). The optimum ensemble of variables explaining the inter- and intra-annual flux variability included solar irradiance, soil moisture and day of the year for GPP, and air temperature, soil moisture, air pressure and day of the year for the ER, in agreement with other studies. The modelling approach discussed here provides a basis for selecting drivers of carbon fluxes and understanding their role in high-altitude Alpine ecosystems, also allowing for future short-range assessments of local trends.