GCM characteristics explain the majority of uncertainty in projected 21st century terrestrial ecosystem carbon balance

[摘要] One of the largest sources of uncertainties in modelling of the futureglobal climate is the response of the terrestrial carbon cycle. Studies haveshown that it is likely that the extant land sink of carbon will weaken in awarming climate. Should this happen, a larger portion of the annual carbondioxide emissions will remain in the atmosphere, and further increase globalwarming, which in turn may further weaken the land sink. We investigate thepotential sensitivity of global terrestrial ecosystem carbon balance todifferences in future climate simulated by four general circulation models(GCMs) under three different CO₂ concentration scenarios. We find thatthe response in simulated carbon balance is more influenced by GCMs thanCO₂ concentration scenarios. Empirical orthogonal function (EOF)analysis of sea surface temperatures (SSTs) reveals differences among GCMsin simulated SST variability leading to decreased tropical ecosystemproductivity in two out of four GCMs. We extract parameters describing GCMcharacteristics by parameterizing a statistical emulator mimicking thecarbon balance response simulated by a full dynamic ecosystem model. Bysampling two GCM-specific parameters and global temperatures we create 60new "artificial" GCMs and investigate the extent to which the GCMcharacteristics may explain the uncertainty in global carbon balance underfuture radiative forcing. Differences among GCMs in the representation ofSST variability and ENSO and its effect on precipitation and temperaturepatterns explain the majority of the uncertainty in the future evolution ofglobal terrestrial ecosystem carbon in our analysis. We suggest that thecharacterisation and evaluation of patterns and trends in simulated SSTvariability should be a priority for the further development of GCMs, inparticular as vegetation dynamics and carbon cycle feedbacks areincorporated.

[学科分类] 地球化学与岩石