OOS 82-9
Uncertainty of net primary productivity and net ecosystem productivity of terrestrial ecosystems in China

Friday, August 14, 2015: 10:50 AM
310, Baltimore Convention Center
Junjiong Shao, School of Ecological and Environmental Sciences, East China Normal University, Shanghai, China
Xuhui Zhou, School of Ecological and Environmental Sciences, East China Normal University, Shanghai, China
Background/Question/Methods

Despite the importance of net primary productivity (NPP) and net ecosystem productivity (NEP) to the welfare of human beings, the estimated NPP and NEP in China were with striking uncertainty. Investigating the main sources of uncertainty of estimated NPP and NEE in China is essential to improving the predictive ability of future models. To this end, we synthesized the estimations of NPP and NEP in China in recent 30 years.

Results/Conclusions

Our results showed that the total NPP and NEP of China were 3.35 ± 1.25 and 0.14 ± 0.094 Gt C yr-1, respectively. Differences in methods and temporal-spatial patterns are two main sources of the large uncertainty. Among models, the largest uncertainty was introduced by the light use efficiency models (50%) and process models (76%) for NPP and NEP, respectively. Among the global change factors, nitrogen addition had the largest effect on NEP (0.11 ± 0.028Gt C yr-1), while land cover and land use change introduced large uncertainty (217%). The interannual pattern of NPP was similar among diverse studies and increased by 0.01103 Gt C yr-1 (r2 = 0.37, P = 0.004) during 1981 - 2000. Among the biomes in China, the estimated NPP and NEP of forests and shrublands had the largest uncertainty. In China’s NPP, large uncertainty occurred in areas with high productivity due to the high spatial heterogeneity. Our results suggest that, to significantly reduce the uncertainty in the estimated NPP and NEP, model structure needs to be largely improved based on empirical results, to which end the coordinated distributed experiments might be a practical approach, and the consistence of temporal-spatial patterns among models should be examined for validating the approach.