Comparing global models of terrestrial net primary productivity (NPP):: analysis of the seasonal atmospheric CO2 signal

Eight terrestrial biospheric models (TBMs) calculating the monthly distributions of both net primary productivity (NPP) and soil heterotrophic respiration (R-H) in the Potsdam NPP Model Intercomparison workshop are used to simulate seasonal patterns of atmospheric CO2 concentration. For each model, we used net ecosystem productivity (NEP=NPP-R-H) as the source function in the TM2 atmospheric transport model from the Max-Planck Institute for Meteorology. Comparing the simulated concentration fields with detrended measurements from 25 monitoring stations spread over the world, we found that the decreasing seasonal amplitude from north to south is rather well reproduced by all the models, though the amplitudes are slightly too low in the north. The agreement between the simulated and observed seasonality is good in the northern hemisphere, but poor in the southern hemisphere, even when the ocean is accounted for. Based on a Fourier analysis of the calculated zonal atmospheric signals, tropical NEP plays a key role in the seasonal cycle of the atmospheric CO2 in the whole southern hemisphere. The relatively poor match between measured and predicted atmospheric CO2 in this hemisphere suggests problems with all the models. The simulation of water relations, a dominant regulator of NEP in the tropics, is a leading candidate for the source of these problems.