Climate sensitivity of global terrestrial ecosystems' subdaily carbon, water, and energy dynamics.

Abstract:

Abstract:

Under the context of global climate change, it is important to understand the direction and magnitude of different ecosystems respond to climate at the global level. In this study, we applied dynamical process network (DPN) approach combined with eco-climate system sensitivity model and used the global FLUXNET eddy covariance measurements (subdaily net ecosystem exchange of CO₂, air temperature, and precipitation) to access eco-climate system sensitivity to climate and biophysical factors at the flux site level. For the first time, eco-climate system sensitivity was estimated at the global flux sites and extrapolated to all possible land covers by employing artificial neural network approach and using the MODIS phenology and land cover products, the long-term climate GLDAS-2 product, and the GMTED2010 Global Grid elevation dataset. We produced the seasonal eco-climate system DPN maps, which revealed how global carbon dynamics driven by temperature and precipitation. We also found that the eco-climate system dynamical process structures are more sensitive to temperature, whether directly or indirectly via phenology. Interestingly, if temperature continues rising, the temperature-NEE coupling may increase in tropical rain forest areas while decrease in tropical desert or Savanna areas, which means that rising temperature in the future could lead to more carbon sequestration in tropical forests whereas less carbon sequestration in tropical drylands. At the same time, phenology showed a positive effect on the temperature-NEE coupling at all pixels, which suggests increased greenness may increase temperature driven carbon dynamics and consequently carbon sequestration globally. Precipitation showed relatively strong influence on the precipitation-NEE coupling, especially indirectly via phenology. This study has the potential to conduct eco-climate system short-term and long-term forecasting.