Comparison of the two biosphere models to estimate global terrestrial carbon fluxes

For understanding and predicting global warming, it is important to accurately estimate the global terrestrial carbon fluxes, especially Net Primary Production (NPP) and Net Ecosystem Production (NEP). However, as actual measurements of these fluxes in global scale are difficult to realize, the only possible way is to use biosphere models. Recently, many biosphere models have been proposed, but the confidence of carbon fluxes derived from these models are not clear. This study aims at comparing spatial and temporal patterns of NPP and NEP between two terrestrial biosphere models (Sasai et al. model and Sim-CYCLE) and three re-analysis climate datasets (NCEP/NCAR, NCEP/DOE, and ERA40). The two models differ in their concept; Sim-CYCLE is more prognostic and driven by sole climate data, while Sasai model is more diagnostic and driven by both climate and satellite data (fAPAR and LAI by NOAA/AVHRR). Both models incorporate a carbon cycle (photosynthesis, respiration, litter fall, and soil decomposition) and hydrology (evaportranspiration and runoff) scheme. Simulations were performed globally from 1982 to 1999, using the three datasets for air temperature, downward short and long wave radiation, precipitation, and humidity. By comparing the six results (two models by three datasets) of NPP and NEP, we expect to address the current uncertainty in our model simulations. First, comparison of climatic variables revealed that the three datasets were mostly consistent in large-scale patterns and interannual variability, but inconsistencies were found for several features (e.g., incoming solar radiation; NCEP/NCAR > NCEP/DOE > ERA40). Second, interannual changes in the estimated NPP and NEP were approximately comparable between models and input data (i.e., within _}1GtC/yr). Here, we focused on the nature of differences and found that the differences in model structure and forcing data affected the estimation of NPP and NEP, since photosynthesis and respiration are sensitive to climatic conditions. Therefore, when we derive some conclusions from simulations, we should pay attention to specific characteristics in models and datasets. Further model validations with observational data, such as flux measurements, are also required. Finally, intercomparison studies between models and datasets, as presented here, should carry implications for global carbon-cycle researches.