Impacts and uncertainties of climate/CO2 change on net primary productivity in Xinjiang, China (2000–2014): A modelling approach

Abstract Dryland in Xinjiang, China has been threatened in recent decades by rapid climate change. However, due to large uncertainties in spatial climate datasets, the impacts of climate change and rising CO 2 on net primary productivity (NPP) in Xinjiang have remained unclear. These uncertainties in climate will inevitably lead to uncertainties in model estimated NPP in Xinjiang. The uncertainties can be assessed by running an Arid Ecosystem Model (AEM) using multiple climate datasets. Such an approach allows to disentangle the relative contributions of individual climate factors on NPP change using numerical simulations and factorial analysis. The average annual NPP for Xinjiang from 2000 to 2014 was 155.90±2.74 g·C/(m 2 ·year), 189.99±1.80 g·C/(m 2 ·year), 213.04±8.93 g·C/(m 2 ·year) by using the MERRA, ERA-Interim and CFSR datasets respectively to drive AEM. Our multiple simulations show a consistent temporal pattern of the regional NPP during 2000–2014 that increased during 2008–2011, and decreased during 2005–2006 and 2013–2014. However, we found large uncertainties in the spatial pattern of NPP change during this time, particularly in the Taklimakan Desert and northern Xinjiang (except for the Tacheng where NPP increased). All simulations indicated that areas surrounding the Taklamakan Desert and ecosystems at high latitude (>47°) and northwestern Xinjiang were dominated by precipitation change. However, there are large uncertainties in the dominant climate driver in other areas. To assess uncertainties in ecosystem NPP assessment, efforts should be made to improve the confidence in climate data in the northern Xinjiang. This can be achieved by establishing more rain gauges.