Improved performance of a PRECIS ensemble in simulating near-surface air temperature over China

Abstract

The near-surface air temperature over China is simulated from 1950 to 2099 using the PRECIS model from the Met Office Hadley Centre at a 25-km resolution. In order to reflect the different parametric and structural uncertainties in future temperature projections, the PRECIS model is driven by five lateral boundary conditions, which include a four-member HadCM3-based perturbed-physics ensemble (i.e., HadCM3Q0, Q1, Q7 and Q13) and an ECHAM5 model. For the present climate, PRECIS reasonably reproduces the spatial patterns of near-surface air temperatures over most regions in China, except for some underestimation in the west. The annual cycles of mean temperature are well captured but its magnitude is slightly underestimated throughout the year. Future temperature projections are further analyzed for three successive 30-year periods throughout the twenty-first century. Despite more uncertainties with time, the ensemble results demonstrate that the temperature over China is likely to continue to increase throughout the twenty-first century, with different spatial-time variation. There is an apparent increasing pattern along with the latitude for seasonal temperature. Through comparison with the driving GCMs, PRECIS ensemble shows smaller biases in most regions of China, except for in the west plateau. The cause is that RCMs could inherit some errors from the driving GCMs in addition to their own errors. These errors could be magnified unintentionally in downscaling over high elevations and have been propagated into future climate projections. However, there is no apparent relationship between projected changes and model biases (i.e., larger bias does not necessarily lead to bigger changes in temperature). These results could be directly used to analyze the impacts caused by climate warming on agriculture, energy and other related sectors in China.