Climate impacts of land-use change in China and its uncertainty in a global model simulation

Results from a series of Australian Bureau of Meteorology Research Centre (BMRC) model simulations are analysed to explore the impacts of land-use vegetation change (LUC) in China on surface climate. Extensive model experiments are conducted to explore uncertainties in the modelling results caused by land-surface modelling and the model intrinsic noise. Results show moderate but statistically significant local impacts by LUC, with significantly different surface temperature responses in winter and summer due to two regimes dominating surface energy balance in the region. In winter, the model shows consistent patterns of a reduction of precipitation in its south/southeast region and a cooling in large parts of the country. The cooling is primarily attributed to the increase in surface albedo while the rainfall reduction is associated with circulation change due to reduced surface roughness. In summer, it reveals moderate impacts with a 10–20% rainfall reduction and above 0.5°C warming in the south/southeast region. The warming is largely the results of reductions in surface evaporation and rainfall simulated in the model. Spatially averaged results are analysed so as to explicitly characterise the model uncertainties. Results suggest that the model-simulated impacts of LUC are not significantly affected by complexity in its land-surface parameterisation, with the scatter between the runs using different surface complexity modes similar to that from a set of the model ensemble runs with different initial conditions. This is largely due to the fact that fundamental processes affected by LUC, including changes in surface albedo and roughness length, are similarly represented in these complexity modes, while weak land–air coupling in the model may also contribute to such results. Further analysis shows that the model LUC signal is much larger than its internal noise derived from ensemble runs and detailed analysis of responses in the model physics and dynamics is to be documented in another manuscript.