|Other Abstract||Water vapor is one of the most abundant and important greenhouse gases in the atmosphere. It plays an important role in the process of global material transportation, energy transmission and water cycle. The feedback of water vapor can amplify the warming effect of other greenhouse gases and has a very important impact on weather and climate change. Therefore, a thorough study of the regularity and causes of atmospheric water vapor changes and regional climate effects will help us to understand the regional climate change and its response mechanism under the background of global warming. Based on the latest homogenized humidity sounding data, this paper systematically evaluates the simulation capability of Phase 5 of the Coupled Model Intercomparison Program (CMIP5) for atmospheric water vapor change in China using various statistical analysis and dynamic diagnostic methods. On this basis, the future changes of water vapor under different emission scenarios are analyzed, and then the climate feedback of water vapor is further studied. Finally, simulated experiments under different external forcing are used to detect the water vapor changes over China. The main conclusions are as follows:Most CMIP5 models can well simulate the seasonal variation characteristics of water vapor observed in China from 1970 to 2015, and can basically reproduce the mean distribution of observed water vapor. Especially in most areas of eastern China, the relative deviation of multi-models from observed values is less than 30%. However, in Western China, especially in the Qinghai-Tibet Plateau, the multi-model is 30-60% lower than the observed results. The root mean square errors of multi-model and observation values show an increasing distribution from northwest to southeast, and their spatial correlation coefficients are low. The positive correlation regions are obviously larger than the negative correlation regions. The correlation coefficients of most models in Xinjiang are generally higher than those in other regions. Compared with single model, the multi-model mean has larger mode correlation coefficients, smaller root mean square error and standard deviation. From 1970 to 2015, the observed water vapor values in most areas of China show a significant growth trend, especially in the Southeast coast, the eastern part of northeast China and Xinjiang. The multi-model can well simulate the growth trend of water vapor change in China. At the same time, the multi-model has a good consistency with the observed water vapor in time series. Empirical orthogonal decomposition (EOF) analysis shows that the first two modes of EOF and their corresponding principal component time series are in good agreement with the observed results, and MIROC-ESM model performs best in simulating the first two modes of EOF.Under different representative concentration pathways (RCP) emission scenarios, estimated results of CMIP5 multi-model show that the future change of atmospheric water vapor content over China has a significant growth trend, and the growth rate basically shows a decreasing distribution from southeast to northwest. In the future, dryer areas are drier and wetter areas are wetter in China. Compared with RCP4.5 scenario, the predicted results under RCP8.5 scenario show that the range of water vapor increase is wider and the growth rate is larger. By the end of the 21st century, the water vapor increase under RCP8.5 scenario is about twice as large as that under RCP4.5 scenario. Most of the models show a decreasing trend of frequency of the extreme dry over the whole China, especially in Northwest China. Compared with RCP4.5, the trend of decreasing extreme dry frequency in RCP8.5 is more significant. However, the frequency of extreme wet increased significantly throughout the country, and the growth rate showed a decreasing trend from southwest to northeast. In the high emission scenario (RCP8.5), the frequency of extreme wet increased more significantly.|
The historical simulation of water vapor feedback effect of CMIP5 models over most areas of China is basically consistent with the observed results. The percentage of water vapor anomaly is in good agreement with the change of surface temperature, which is more obvious in the multi-model simulation results, and the correlation between water vapor content and relative humidity is lower. Over the whole country, the slopes of fitting equation between the percentage of water vapor anomaly and surface temperature anomaly are 6.7%/K and 6.6%/K, respectively, slightly lower than the theoretical results (7%/K) of Clausius-Clapeyron (C-C) equation. The results of observation and simulation in southeastern and northern regions are basically the same, both lower than the theoretical results of 7%/K, while over the Qinghai Tibet Plateau, the simulation results are basically the same. The results were significantly lower than those observed. The future prediction results show that, under the RCP4.5 scenario, the fitting equation slopes (6.9%/K and 7.1%/K) of water vapor anomaly and surface temperature anomaly over the whole country and the Qinghai Tibet Plateau are basically consistent with the theoretical results of C-C equation, and slightly lower than those of C-C equation in the southeast and North regions. Compared with RCP4.5 scenario, the predicted results under RCP8.5 scenario are significantly larger and higher than the theoretical results of C-C equation. This means that in the high emission scenario, the future change of water vapor in China has a stronger positive feedback effect.Under total forcing (ALL), multi-model can basically reproduce the growth trend of observed water vapor. The increase of water vapor caused by greenhouse gas forcing (GHG) is more obvious, while natural forcing (NAT) can reproduce the impact of Pinatubo volcano eruption, but overall, it contributes little to the increase of water vapor in China and can not explain the change of water vapor. The single-signal detection results based on the optimal fingerprint method show that the multi-model simulations under total forcing underestimates 12% of the observed results over the whole China, while overestimates 7.5% of the observed results in eastern China. The results of two-signal detection and attribution show that the change of water vapor in China is mainly attributed to human factors (ANT). The contribution of human activities to water vapor changes over the whole country and eastern China is 90% and 94%, respectively, while that of natural factors is less than 10%. The trends of water vapor increase after the correction of the optimal scaling factor show that the rate of water vapor increase caused by human-induced forcing is 1.11mm/40yr in the whole China and 1.23mm/40yr in eastern China, respectively. Natural forcing can only cause 0.12mm/40yr and 0.04mm/40yr growth rates, respectively. Based on the detection and attribution results of GHG, ANTnoGHG (human factors except GHG) and NAT signals, greenhouse gases are the most important anthropogenic factors causing water vapor increase in China. The contribution rates of GHG to water vapor increase are 133% and 182% in the whole country and eastern China, respectively, while other human activities and natural factors have not been significantly detected. Based on the optimal scaling factor of ALL forcing obtained from single signal detection attribution, the future prediction results of CMIP5 are adjusted The results show that the predicted results after nationwide adjusting are larger than the raw simulations, while in eastern China, the adjusted results are smaller than the raw simulations.