This article was submitted to Atmospheric Science, a section of the journal Frontiers in Earth Science
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
This study reveals the impacts of cumulus parameterization schemes (CPSs) on the simulation of interannual and interdecadal variations of Meiyu in the middle and lower reaches of the Yangtze River valley (Yangtze Meiyu) by using a regional climate model. A multiple 55-year (1955-2009) simulation of Yangtze Meiyu is conducted using RegCM4.6 with different CPSs, including the Emanuel scheme, Kuo scheme, and Grell scheme. It is found that all the CPSs have good performances in simulating the interannual variation of Yangtze Meiyu rainfall amounts. However, a large bias is found in the simulated interdecadal variation of Yangtze Meiyu rainfall amount by the three CPSs. Because the Emanuel CPS has good performance in the simulation of both interannual and interdecadal variations of Yangtze Meiyu, its overall performance in the simulation of the total amount of Yangtze Meiyu is the best among the three CPSs. The second best CPS is the Grell scheme, and the worst is the Kuo scheme. The model convergence in simulating the distribution of Yangtze Meiyu shows an obvious characteristic of interdecadal variation. During the years with strong summer monsoon and northward rain belt, the simulated distributions of Yangtze Meiyu by the three CPSs are quite different, and the model convergence is weak. On the contrary, during the years with weak summer monsoon and southward rain belt, the simulated distributions of Yangtze Meiyu by the three CPSs are similar to each other, and the model convergence is strong. As a result, when the well-known interdecadal change happens in the late 1970s, the monsoon changes from strong to weak, whereas the model convergence changes from weak to strong.
The Meiyu front is an important component of the East Asian summer monsoon system. Climatologically, as the summer monsoon advances northward in mid-June, the Western Pacific Subtropical High (WPSH) jumps northward to around 25°N, and the Meiyu front to the north of the WPSH remains quasi-stationary in the middle and lower reaches of the Yangtze River. The dry and cold air from the north converges with the warm and moist monsoon southwesterly flow, resulting in large-scale Meiyu (or plum rain) in the middle and lower reaches of the Yangtze River (hereafter Yangtze Meiyu), which usually lasts until early to mid-July (
Yangtze Meiyu demonstrates significant interannual and interdecadal variations (Tanaka, 1997). Some studies have revealed that the interannual variation of Yangtze Meiyu is mainly characterized by a biennial oscillation (TBO), and this TBO characteristic is more obvious in the middle and lower reaches of the Yangtze River than in other regions of China (
Whether the climate model can realistically reproduce characteristics of interannual and interdecadal variations of the East Asian monsoon is of great significance for future climate prediction, climate model improvement, and climate change assessment. Many studies have evaluated the capability of global climate models (GCMs) for simulating interannual and interdecadal changes of the East Asian monsoon (
Biases of the simulated summer precipitation in the RCMs are mainly related to the cumulus parameterization schemes (CPSs). This is because cumulus convections are very strong in the Meiyu frontal system, and convective precipitation accounts for a large proportion of total precipitation (
The simulation of summer precipitation and circulation in East Asia is very sensitive to the choice of CPS in the RCM. Because the triggering mechanisms, moist convective processes, and the feedback processes to the large-scale environmental field in different CPSs differ from each other, the precipitation and circulation simulated by different CPSs are also significantly different (
Previous studies have focused more on the impacts of CPSs on the simulation of seasonal precipitation in East Asia, but few studies provide detailed analysis about the impacts of CPSs on the simulated interannual and interdecadal variations of Yangtze Meiyu. Therefore, the performances of different CPSs in simulating the interannual and interdecadal variations of Yangtze Meiyu are evaluated, and the model bias and model convergence of the simulation by different CPSs are also investigated in this study. These results will be helpful for the prediction of future change of Yangtze Meiyu and the understanding of the results from a regional climate model. The rest of the article is organized as follows: Section 2 describes the datasets and methods and
Regional climate model 4.6 (RegCM4.6), which is developed by the International Centre for Theoretical Physics (ICTP) in Italy, is used in this study. It is the latest version of the regional climate model and upgraded from its last version (RegCM3). It has more advanced physical schemes and computing solutions than RegCM3. Compared to the previous versions, RegCM4.6 has an improved performance in several respects (
Due to the large interannual variability of Yangtze Meiyu, the duration of the rainy season and the time of entering and exiting the rainy season are different from year to year. Climatologically, the Yangtze Meiyu season begins in mid-June and ends in mid-July every year. For the sake of convenience, we define each year from June 15 to July 15 as the Yangtze Meiyu season. RegCM4.6 is used to simulate climate in the Yangtze Meiyu season for a total of 65 years from 1951 to 2015. All experiments are initialized at 6:00 UTC 5 June and end at 00 UTC 16 July each year. The first 10 days are taken as a spin-up time, and results for the period 15 June to 15 July are analyzed. The grid interval is 30 km, and the model domain is centered at (116°E, 30°N). There are 72 and 56 grid numbers in the east–west and north–south directions, respectively. The area (112°E∼122°E, 28°N∼33°N) is taken as the middle and lower reaches of the Yangtze River. The model domain and the area of the middle and lower reaches of the Yangtze River are displayed in
Model domain, topography (shaded, unit: m), and the area of the middle and lower reaches of the Yangtze River.
The harmonic analysis method proposed by
The observed monthly precipitation at 752 stations in China and NCEP reanalysis data are interpolated to the model grids to verify the performance of the model. The precipitation dataset is released by the national meteorological information center, which can be downloaded freely from the Internet site at
The area-averaged rainfall amount in the middle and lower reaches of the Yangtze River is defined as the Yangtze Meiyu rainfall amount. The variations in Yangtze Meiyu rainfall amount from simulations with the three CPSs and from observations are displayed in
Correlation coefficients between observated and simulated Yangtze Meiyu by various CPSs and the 55-year mean biases of the simulation (unit: mm/day).
Emanuel | Grell | Kuo | |
---|---|---|---|
Correlation coefficient | 0.77 | 0.74 | 0.73 |
Mean biases | 0.74 | 1.97 | -1.97 |
Since all the three CPSs show systematic errors in the simulation of Yangtze Meiyu, observed and simulated normalized anomalies of Yangtze Meiyu simulated are calculated (
Next, the impacts of the CPSs on simulating the interannual variation of Yangtze Meiyu are analyzed.
Correlation coefficients of interannual and interdecadal changes of simulated Yangtze Meiyu with observations.
Emanuel | Grell | Kuo | |
---|---|---|---|
Interannual | 0.80 | 0.73 | 0.86 |
Interdecadal | 0.71 | 0.61 | 0.50 |
Spacial correlation coefficients between simulations and observations of Yangtze Meiyu averaged over the 55-year period (a) and its interannual (b) and interdecadal (c) variations.
Original | Interannual | Interdecadal | |
---|---|---|---|
Emanuel | 0.30 | 0.34 | 0.29 |
Grell | 0.45 | 0.42 | 0.41 |
Kuo | 0.22 | 0.34 | 0.29 |
In addition, comparison of interdecadal variations of Yangtze Meiyu simulated by various CPSs shows that the correlation coefficient between observations and simulations by the Emanuel CPS is the highest (0.71), followed by the Grell CPS (0.61), and the worst is the Kuo CPS (0.50). The above results indicate that the capability of the Emanuel CPS is much better than that of the other two CPSs for the simulation of interdecadal variations of Yangtze Meiyu. Because the Emanuel CPS performs better in the simulation of both interannual and interdecadal variations of Yangtze Meiyu, its overall performance in the simulation of the total amount of Yangtze Meiyu is also the best among the three CPSs. Although the Kuo scheme can simulate the interannual variation of Yangtze Meiyu well, its overall performance in simulating the variation of Yangtze Meiyu rainfall amount is dragged down by its poor performance in simulating the interdecadal variation of Yangtze Meiyu rainfall amount. This further illustrates that various CPSs in the RCM have quite different responses to interdecadal variation signals in the large-scale forcing, although their responses to signals of interannual variation are similar.
In addition to the analysis of the Yangtze Meiyu changes simulated by the three CPSs, the simulated atmospheric circulation is also analyzed. It is well known that precipitation can persist over the middle and lower reaches of the Yangtze River in the Meiyu season only when the appropriate large-scale circulation remains stable. Generally, three external conditions are required for maintaining the Yangtze Meiyu: a stable WPSH with its ridge line located near 28°N, a stable blocking high in the north, and continuous water vapor transport to the middle and lower reaches of the Yangtze River (
Yangtze Meiyu distribution (shaded) and 500-hPa geopotential height (contours, unit: gpm) averaged over the 55-year period from 1955 to 2009 for observations and simulations by the Emanuel, Grell, and Kuo schemes.
However, obvious differences can be found in precipitation distribution simulated by the three CPSs. The Grell and Emanuel schemes significantly overestimate precipitation, while the Kuo scheme underestimates precipitation. Besides, in addition to the large precipitation area in the middle and lower reaches of the Yangtze River, the Emanuel and Kuo schemes also produce a fictitious rain band to the north of the middle and lower reaches of the Yangtze River. The differences in the simulated distribution of Yangtze Meiyu indicate that in addition to large-scale circulation, the physical parameterization scheme used in the model also has great impacts on the precipitation simulation, especially in summer, when the large-scale forcing is weaker than that in the winter and local convection becomes more important for precipitation (Giorgi et al., 1999). Therefore, under the same large-scale atmospheric circulation, sdifferent CPSs yield different Yangtze Meiyu distributions.
In addition to the total Yangtze Meiyu amount, the performances of three CPSs in simulating the distribution of Yangtze Meiyu are also evaluated. The spatial correlation coefficients between the simulations and observations of annual Yangtze Meiyu and its interannual and interdecadal variation components over the middle and lower reaches of the Yangtze River are calculated (
Evolution of the spatial correlation coefficients between simulations and observations of annual Yangtze Meiyu rainfall
Evolution of the convergence index of the simulated spatial distribution of Yangtze Meiyu by the three cumulus parameterization schemes. The smaller the index, the stronger the convergence.
Why the model convergence changes in the late 1970s? To find the reason, the circulation patterns in some typical years with large and small differences in Yangtze Meiyu between observations and simulations are analyzed, respectively. Several representative years are selected to demonstrate their specific characteristics. The years 1964 (
Spatial distributions of precipitation (unit: mm/day) and 500-hPa geopotential height (unit: gpm) during the Meiyu season of 1964 from observations and simulations by the Emanuel, Grell, and Kuo schemes.
Spatial distributions of precipitation (unit: mm/day) and 500- hPa geopotential height (unit: gpm) during the Meiyu season of 1974 from observations and simulations by the Emanuel, Grell, and Kuo schemes.
Spatial distributions of precipitation (unit: mm/d) and 500-hPa geopotential height (unit: gpm) during the Meiyu season of 1992 from observations and simulations by the Emanuel, Grell, and Kuo schemes.
Spatial distributions of precipitation (unit: mm/d) and 500-hPa geopotential height (unit: gpm) during the Meiyu season of 1993 from observations and simulations by the Emanuel, Grell, and Kuo schemes.
Let us take a look at the circulation characteristics in the years with small differences in Yangtze Meiyu between observations and simulations. The years 1992 (
Through the above analysis, it can be found that the model convergence of Yangtze Meiyu distribution simulated by different CPSs is closely related to the intensity of the East Asain summer monsoon (EASM), the pattern of the monsoon circulation system, and the location of the rain belt each year. To demonstrate this relationship between the model convergence and the intensity of EASM, the East Asian summer monsoon index (EASMI) proposed by
Evolution of interdecadal components of the convergence index (black line with hollow circles) and monsoon index (green line with solid circles).
It is well known that the distribution of the summer rain belt is closely related to the strength of the EASM: when the EASM is strong and the WPSH is located further northeast, the monsoon southwesterlies can reach North China without being blocked by the WPSH. As a result, more precipitation occurs in North China and less precipitation occurs in the middle and lower reaches of the Yangtze River, and the monsoon rain belt shifts northward than its normal position. On the contrary, when the EASM weakens and the WPSH shifts to the southwest, the monsoon southwesterlies are blocked by the WPSH and can only reach the Yangtze–Huai River basin, where they converge with the cold air and form precipitation. In this situation, more precipitation occurs in the Yangtze–Huai River basin and less precipitation occurs in North China. The rain band shifts to the south of its normal position (
In this study, the impacts of CPSs on the simulation of interannual and interdecadal variations of Yangtze Meiyu and the model convergence are investigated by using RegCM4. A multiple 55-year (1955-2009) simulation of Yangtze Meiyu is conducted with different CPSs, including the Emanuel scheme, Kuo scheme, and Grell scheme.
It is found that all the CPSs have good performances and strong convergence in simulating the interannual variation of Yangtze Meiyu rainfall amount. However, large bias and weak model convergence are found in the simulated interdecadal variation of Yangtze Meiyu rainfall amount by the three CPSs. Because the Emanuel CPS has good performance in the simulation of both interannual and interdecadal variations of Yangtze Meiyu, its overall performance in the simulation of the total amount of Yangtze Meiyu is also the best among the three CPSs. The second best CPS is the Grell CPS. and the worst is the Kuo CPS.
Atmospheric circulations can be realistically reproduced by all the three CPSs and are basically the same due to the strong constraint of the lateral boundary condition. The simulated circulation patterns are mainly determined by the strong forcing of lateral boundary conditions, instead of the internal physical process scheme (such as the CPS). However, due to the different triggering mechanisms and moist convective process of different CPSs, the simulated Yangtze Meiyu by different CPSs differs from each other. For example, the convective activity in the Kuo scheme is initiated when the moisture convergence in a column exceeds a given threshold and the vertical sounding is convectively unstable. The Grell scheme parameterization considers clouds as two steady-state circulations: an updraft and a downdraft. Its convection is activated when a lifted parcel attains moist convection. The Emanuel scheme assumes that the mixing in clouds is highly episodic and inhomogeneous and considers convective fluxes based on an idealized model of sub–cloud-scale updrafts and downdrafts. Its convection is triggered when the level of neutral buoyancy is greater than the cloud base level. Therefore, CPS has great impacts on precipitation simulation, especially in summer, when the large-scale forcing is weaker than that in the winter and local convection becomes more important for precipitation.
The model convergence in simulating the distribution of Yangtze Meiyu shows an obvious characteristic of interdecadal variation. During the years with strong summer monsoon and northward rain belt, the model convergence is weak by using different CPSs. On the contrary, during the years with strong and southward rain belt, the model convergence is strong. As a result, with the well-known interdecadal change happening in the late 1970s, the monsoon changes from strong to weak, whereas the model convergence changes from weak to strong.
It is noteworthy that the model convergence of different CPSs will change with the interdecadal variation of monsoon climate background. Weak model convergence will reduce the credibility of the ensemble prediction results in future climate change prediction. Therefore, it is very important to predict the future summer monsoon climate accurately and evaluate the model convergence of RCM cautiously. Then, we can determine the reliability of the ensemble prediction results.
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: NCEP/NCAR Reanalysis 1 [
YHu conducted the numerical experiments and analyzed the results. ZZ designed the numerical experiments and revised this paper. YHa analyzed the interannual and interdecadal variations of Meiyu in simulation. YS provided the data for this work. YZ and LY revised this paper and polish the English.
This work is sponsored jointly by the National Key Research and Development Project (2018YFC1505803), National Natural Science Foundation of China (41675077, 41975090, and 41605072), Scientific Research Program of National University of Defense Technology (ZK17-03–22), and Jiangsu Collaborative Innovation Center for Climate Change.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.