Model Evaluation Metrics Based On The Linkage Between Large-scale Circulation Indices And Regional Precipitation In The CMIP5 Climate Models And Its Application

East Asian summer monsoon, west Pacific subtropical high and subtropical westerly jet are three of the most prominent features tightly associated with the summer precipitation in China. The East Asian atmospheric circulations have been evaluated in 31 climate models participating in the Coupled Model Intercomparison Project Phase 5 (CMIP5) by comparing the results from four reanalysis datasets, including NCEP2, ERA-interim, MERRA and JRA-55. Based on the good relationships between four circulation indices and summer precipitation over mid-lower Yangtze River, new metrics are developed for model evaluation. The uncertainties in summer precipitation and extreme precipitation simulations over China are analyzed through intercomparing models grouped by the new metrics and Taylor skill score. The main results are as follows:1. Evaluation of summer circulation simulation over East AsiaThe climatology of summer atmospheric circulation over East Asia has been evaluated in 31 CMIP5 models. Although multi-model ensemble (MME) can reasonably reproduce the climatology of 850 hPa and 200 hPa zonal wind as well as 850 hPa geopotential height, most models underestimate the climatology of 500 hPa geopotential height.Based on the performance of CMIP5 models in simulating the climatology of East Asian summer atmospheric circulations, four circulation indices are selected to evaluate CMIP5 models, including East Asian summer monsoon index (EASMI), west Pacific subtropical high index (WPSHI), South China Sea subtropical high index (SCSHI), and East Asian subtropical jet index (EASJI). The relationship between each two circulation indices derived from the four reanalysis datasets is fairly consistent. EASMI, WPSHI and SCSHI are dependent indices, and EASJI is independent of the other three indices. The large uncertainties exist among models in simulating the amplitude of their interannual variability. However, the CMIP5 MME is close to the observation.2. The linkage between large-scale circulation and regional precipitation in the CMIP5 climate modelsThe correlation coefficients between large-scale circulations over East Asia and summer precipitation over mid-lower Yangtze River existing in the reanalysis datasets are persistent and statistically significant. These correlation coefficients can quantitatively reflect the physical progress by which the large-scale circulations affect the summer precipitation in China Based upon the above investigation, we use the correlation coefficients as the metrics for evaluating the model performance in simulating the summer precipitation in China.The analysis of climate models to represent the observed behaviour of past climate is as a necessary condition to be considered a viable tool for future projections. Therefore, before investigating the future projection of summer precipitation, it is necessary to evaluate the model performance in simulating the precipitation. We use the Taylor skill score to evaluate the CMIP5 models performance in simulating the precipitation in China and find that this metric can distinguish model skill in the historical simulations but can not narrow down the uncertainties in the projections of future changes. So we correlate the skill of models determined based on the metrics (relationships) with the precipitation skill then find that there are no statistically significant relationships between them. To address this issue, new metrics are proposed by combining the information from skill of models determined based on the relationships and the precipitation skill. We use these metrics to divide the CMIP5 models into three groups:The top 5 best models (Top5), the bottom 5 models (Bottom5) and the middle 21 models (Middle21). Therefore, the selected models which reasonably reproduce the relationships between four circulation indices and summer precipitation over mid-lower Yangtze River also can well simulate the summer precipitation in China. Further analysis revealed that these metrics can distinguish the skill of models in simulating precipitation in China, but also help to narrow down the uncertainty in the projections of future changes in summer precipitation in China.3. Extreme precipitation indices over China in the CMIP5 Climate Models: Model evaluation and future climate projectionsBased on the daily precipitation dataset from Chinese National Meteorological Centre, the extreme precipitation indices in China have, been evaluated and these metrics’ applicability in precipitation indices is discussed. The spatial distribution of summer precipitation amounts from very wet days (R95pTOT) in China is well reproduced by the CMIP5 MME, the simulated maximum number of consecutive wet days (CWD) is overestimated over China except northwest China and the maximum consecutive dry days (CDD) is underestimated over the region from the eastern Tibetan Plateau to southwest China. Most models’skills in simulating the first three indices are better than that of hydroclimatic intensity (HY-INT) index. Note that the models reasonably reproduces the precipitation in China and the relationships between the first three circulation indices and summer precipitation over mid-lower Yangtze River, and also can well represent the R95pTOT in China. Although these models can reasonably reproduce the CWD and CDD in China, they don’t show outstanding performance in 31 CMIP5 models compared to the R95pTOT simulations. Moreover, these metrics can not only distinguish R95pTOT skill in the historical climate, but also help to narrow uncertainty in the projection of future change in R95pTOT in China.The models display a wide range of skill in simulating the CWD, CDD and HY-INT. Thus, this study investigates the consistent responses of the last three extreme indices to global warming among the models by analyzing the probability of models which project consistent increases or decrease in extreme indices at each grid in China. We draw the conclusion that more than 60 percent of the models project an increase in CWD over some parts of northeast and north China and an increase in CDD over the area south to the Yangtze River and eastern Tibetan Plateau. Under global warming, the HY-INT does not display consistent responses in individual models.