Variability and predictability of decadal mean temperature and precipitation over China in the CCSM4 last millennium simulation
Ying, K and Frederiksen, CS and Zheng, X and LOU, J and Zhao, T, Variability and predictability of decadal mean temperature and precipitation over China in the CCSM4 last millennium simulation, Climate Dynamics, 51, (7-8) pp. 2989-3008. ISSN 0930-7575 (2018) [Refereed Article]
The modes of variability that arise from the slow-decadal (potentially predictable) and intra-decadal (unpredictable) components of decadal mean temperature and precipitation over China are examined, in a 1000 year (850-1850 AD) experiment using the CCSM4 model. Solar variations, volcanic aerosols, orbital forcing, land use, and greenhouse gas concentrations provide the main forcing and boundary conditions. The analysis is done using a decadal variance decomposition method that identifies sources of potential decadal predictability and uncertainty. The average potential decadal predictabilities (ratio of slow-to-total decadal variance) are 0.62 and 0.37 for the temperature and rainfall over China, respectively, indicating that the (multi-)decadal variations of temperature are dominated by slow-decadal variability, while precipitation is dominated by unpredictable decadal noise. Possible sources of decadal predictability for the two leading predictable modes of temperature are the external radiative forcing, and the combined effects of slow-decadal variability of the Arctic oscillation (AO) and the Pacific decadal oscillation (PDO), respectively. Combined AO and PDO slow-decadal variability is associated also with the leading predictable mode of precipitation. External radiative forcing as well as the slow-decadal variability of PDO are associated with the second predictable rainfall mode; the slow-decadal variability of Atlantic multi-decadal oscillation (AMO) is associated with the third predictable precipitation mode. The dominant unpredictable decadal modes are associated with intra-decadal/inter-annual phenomena. In particular, the El Nino-Southern Oscillation and the intra-decadal variability of the AMO, PDO and AO are the most important sources of prediction uncertainty.