Application Of The Piecewise Approach In Short-term Climate Sensitivity Experiments

The numerical experiments with two simple models have documented that the piecewise modeling approach(PW)used in numerical sensitivity experiments can effectively improve simulation precision,but for the moment it is not applied to complex models and real sensitivity cases.For this reason,we design a short-term climate sensitivity experiment to study the influence of anomalous snow cover over the Tibetan Plateau in spring 2003 on the summertime climate of China and assess the effectivity and ability of PW applied to a complex model.Firstly,we perform a series of ideal numerical simulations for the above sensitivity experiment with the Weather Research and Forecasting(WRF)model to test whether PW outperforms the conventional continuous modeling approach(CONT),and to investigate the impact of update period and analysis error on PW.Then a real sensitivity experiment is performed with PW in which NCEP FNL analysis is used to update reference and perturbed states to explore the influence of anomalous snow cover over the Tibetan Plateau in spring 2003 on the summertime climate of China.Furthermore,the simulated 2-m temperature and percipitation changes(the difference between reference and perturbed states)caused by anomalous snow cover are compared with CMAP analysis and NCEP R2 reanalysis.The main conclusions are listed as follows:(1)In the ideal numerical sensitivity experiments,we compare the simulated reference and perturbed states and its difference(change field)from CONT and PW experiments.It can be found from statistic on the simulated error and correlation coefficient and spatial distribution characteristic that PW is significantly superior to CONT.PW can effectively reduce the error accumulation caused by a long-term continuous integrate,and can significantly improve the simulation accuracy of reference and perturbed state,and consequently improve the simulationaccuracy of the change field.(2)We further investigate the influence of update period on PW.The update period is set to12-hour,24-hour and 48-hour,respectively.It can be seen from the statistical results of correlation coefficient and root mean square error and the spatial distribution of simulated error that the shorter update period is,the better simulated results are.(3)We also examine the effect of the analysis error on PW.The simulations with two analysis error are used to compare with the simulation with perfect analysis.The experimental results show that the simulation with perfect analysis is the best,and the smaller analysis error is,the better simulated results are.For the imperfect analysis data,the shorter update period is,the better PW is,as we can see from statistical results betwen the simulated fields and its "truth".(4)The NCEP FNL analysis data is preprocessed by spectral nudging method prior to its real simulation in PW.The results with real simulations indicate that the increase of snow cover over the Tibetan Plateau in spring will lead to the enhancement of the Western Pacific Subtropical High,precipitation increase and temperature decrease in the Yangtze-Huaihe river basin,precipitation decrease and temperature increase in South China.The climatic change characteristic is consistent with observed climate anomalies in some areas of China,it implies that China climate anomalies in 2003 have some relationship with the increase of snow cover over the Tibetan Plateau in spring.The piecewise approach can reveal this relations,but it can not be seen from the simulated results with CONT.(5)Compared with observations(the existing reanalysis or analysis data),PW can get more more reasonable simulations than CONT,especially for detail description of the change fields in spatial distribution.