Advanced Search
Article Contents

Spatial Characteristics of Extreme Rainfall over China with Hourly through 24-Hour Accumulation Periods Based on National-Level Hourly Rain Gauge Data


doi: 10.1007/s00376-016-6128-5

  • Hourly rainfall measurements of 1919 national-level meteorological stations from 1981 through 2012 are used to document, for the first time, the climatology of extreme rainfall in hourly through 24-h accumulation periods in China. Rainfall amounts for 3-, 6-, 12- and 24-h periods at each station are constructed through running accumulation from hourly rainfall data that have been screened by proper quality control procedures. For each station and for each accumulation period, the historical maximum is found, and the corresponding 50-year return values are estimated using generalized extreme value theory. Based on the percentiles of the two types of extreme rainfall values among all the stations, standard thresholds separating Grade I, Grade II and Grade III extreme rainfall are established, which roughly correspond to the 70th and 90th percentiles for each of the accumulation periods. The spatial characteristics of the two types of extreme rainfall are then examined for different accumulation periods. The spatial distributions of extreme rainfall in hourly through 6-h periods are more similar than those of 12- and 24-h periods. Grade III rainfall is mostly found over South China, the western Sichuan Basin, along the southern and eastern coastlines, and in the large river basins and plains. There are similar numbers of stations with Grade III extreme hourly rainfall north and south of 30°N, but the percentage increases to about 70% south of 30°N as the accumulation period increases to 24 hours, reflecting richer moisture and more prolonged rain events in southern China. Potential applications of the extreme rainfall climatology and classification standards are suggested at the end.
  • 加载中
  • Atomic Energy Regulatory Board of India, 2008: Extreme Values of Meteorological Parameters. Atomic Energy Regulatory Board of India, Mumbai, 37 pp.
    Chen J., Y. G. Zheng, X. L. Zhang, and P. J. Zhu, 2013: Distribution and diurnal variation of warm-season short-duration heavy rainfall in relation to the MCSs in China. Acta Meteorologica Sinica,27, 868-888, doi: 10.1007/s13351-013-0605-x.10.1007/s13351-013-0605-x2cfd5cc1979a2f2650a1129b7a990310http%3A%2F%2Flink.springer.com%2F10.1007%2Fs13351-013-0605-xhttp://d.wanfangdata.com.cn/Periodical/qxxb-e201306008Short-duration heavy rainfall (SDHR) is a type of severe convective weather that often leads to substantial losses of property and life. We derive the spatiotemporal distribution and diurnal variation of SDHR over China during the warm season (April–September) from quality-controlled hourly raingauge data taken at 876 stations for 19 yr (19912-2009), in comparison with the diurnal features of the mesoscale convective systems (MCSs) derived from satellite data. The results are as follows. 1) Spatial distributions of the frequency of SDHR events with hourly rainfall greater than 10–40 mm are very similar to the distribution of heavy rainfall (daily rainfall 82 50 mm) over mainland China. 2) SDHR occurs most frequently in South China such as southern Yunnan, Guizhou, and Jiangxi provinces, the Sichuan basin, and the lower reaches of the Yangtze River, among others. Some SDHR events with hourly rainfall 82 50 mm also occur in northern China, e.g., the western Xinjiang and central-eastern Inner Mongolia. The heaviest hourly rainfall is observed over the Hainan Island with the amount reaching over 180 mm. 3) The frequency of the SDHR events is the highest in July, followed by August. Analysis of pentad variations in SDHR reveals that SDHR events are intermittent, with the fourth pentad of July the most active. The frequency of SDHR over mainland China increases slowly with the advent of the East Asian summer monsoon, but decreases rapidly with its withdrawal. 4) The diurnal peak of the SDHR activity occurs in the later afternoon (1600–1700 Beijing Time (BT)), and the secondary peak occurs after midnight (0100–0200 BT) and in the early morning (0700–0800 BT); whereas the diurnal minimum occurs around late morning till noon (1000–1300 BT). 5) The diurnal variation of SDHR exhibits generally consistent features with that of the MCSs in China, but the active periods and propagation of SDHR and MCSs differ in different regions. The number and duration of local maxima in the diurnal cycles of SDHR and MCSs also vary by region, with single, double, and even multiple peaks in some cases. These variations may be associated with the differences in large-scale atmospheric circulation, surface conditions, and land-sea distribution.
    Chen L. X., Q. G. Zhu, H. B. Luo, J. H. He, M. Dong, and Z. Feng, 1991: East Asian Monsoon. China Meteorological Press, Beijing, 362 pp. (in Chinese).
    Chen X. C., K. Zhao, and M. Xue, 2014: Spatial and temporal characteristics of warm season convection over Pearl River Delta region,China, based on 3 years of operational radar data. J. Geophys. Res., 119, 12 447-12 465, doi: 10. 1002/2014JD021965.10.1002/2014JD02196502feb9f0ccf76057b32dff154651390ehttp%3A%2F%2Fonlinelibrary.wiley.com%2Fdoi%2F10.1002%2F2014JD021965%2Fpdfhttp://onlinelibrary.wiley.com/doi/10.1002/2014JD021965/pdfThis study examines the temporal and spatial characteristics and distributions of convection over the Pearl River Delta region of Guangzhou, China, during the May-September warm season, using, for the first time for such a purpose, 3 years of operational Doppler radar data in the region. Results show that convective features occur most frequently along the southern coast and the windward slope of the eastern mountainous area of Pearl River Delta, with the highest frequency occurring in June and the lowest in September among the 5 months. The spatial frequency distribution pattern also roughly matches the accumulated precipitation pattern. The occurrence of convection in this region also exhibits strong diurnal cycles. During May and June, the diurnal distribution is bimodal, with the maximum frequency occurring in the early afternoon and a secondary peak occurring between midnight and early morning. The secondary peak is much weaker in July, August, and September. Convection near the coast is found to occur preferentially on days when a southerly low-level jet (LLJ) exists, especially during the Meiyu season. Warm, moist, and unstable air is transported from the ocean to land by LLJs on these days, and the lifting along the coast by convergence induced by differential surface friction between the land and ocean is believed to be the primary cause for the high frequency along the coast. In contrast, the high frequency over mountainous area is believed to be due to orographic lifting of generally southerly flows during the warm season.
    Chen X. C., K. Zhao, M. Xue, B. W. Zhou, X. X. Huang, and W. X. Xu, 2015: Radar-observed diurnal cycle and propagation of convection over the Pearl River Delta during Mei-Yu season. J. Geophy. Res. Atmos.,120, 12 557-12 575, doi: 10.1002/2015JD023872.10.1002/2015JD023872667205f58e038eeb7bdcd37c55a642a8http%3A%2F%2Fonlinelibrary.wiley.com%2Fdoi%2F10.1002%2F2015JD023872%2Ffullhttp://onlinelibrary.wiley.com/doi/10.1002/2015JD023872/fullUsing operational Doppler radar and regional reanalysis data from 2007-2009, the climatology and physical mechanisms of the diurnal cycle and propagation of convection over the Pearl River Delta (PRD) region of China during the Mei-Yu seasons are investigated. Analyses reveal two hot spots for convection: one along the south coastline of PRD and the other on the windward slope of mountains in the northeastern part of PRD. Overall, convection occurs most frequently during the afternoon over PRD due to solar heating. On the windward slope of the mountains, convection occurrence frequency exhibits two daily peaks, with the primary peak in the afternoon and the secondary peak from midnight to early morning. The nighttime peak is shown to be closely related to the nocturnal acceleration and enhanced lifting on the windward slope of southwesterly boundary layer flow, in the form of boundary layer low-level jet. Along the coastline, nighttime convection is induced by the convergence between the prevailing onshore wind and the thermally induced land breeze in the early morning. Convection on the windward slope of the mountainous area is more or less stationary. Convection initiated near the coastline along the land breeze front tends to propagate inland from early morning to early afternoon when land breeze cedes to sea breeze and the prevailing onshore flow.
    Coles S., 2001: An Introduction to Statistical Modeling of Extreme Values. Springer,London, 223 pp.10.1007/978-1-4471-3675-0726293a625cd4893ac321a6dc41967fahttp%3A%2F%2Fwww.researchgate.net%2Fpublication%2F230663933_An_introduction_to_statistical_modeling_of_extreme_values_-_springerhttp://www.researchgate.net/publication/230663933_An_introduction_to_statistical_modeling_of_extreme_values_-_springerDirectly oriented towards real practical application, this book develops both the basic theoretical framework of extreme value models and the statistical inferential techniques for using these models in practice. Intended for statisticians and non-statisticians alike, the theoretical treatment is elementary, with heuristics often replacing detailed mathematical proof. Most aspects of extreme modeling techniques are covered, including historical techniques (still widely used) and contemporary techniques based on point process models. A wide range of worked examples, using genuine datasets, illustrate the various modeling procedures and a concluding chapter provides a brief introduction to a number of more advanced topics, including Bayesian inference and spatial extremes. All the computations are carried out using S-PLUS, and the corresponding datasets and functions are available via the Internet for readers to recreate examples for themselves. An essential reference for students and researchers in statistics and disciplines such as engineering, finance and environmental science, this book will also appeal to practitioners looking for practical help in solving real problems. Stuart Coles is Reader in Statistics at the University of Bristol, UK, having previously lectured at the universities of Nottingham and Lancaster. In 1992 he was the first recipient of the Royal Statistical Society's research prize. He has published widely in the statistical literature, principally in the area of extreme value modeling.
    Davis R. S., 2001: Flash flood forecast and detection methods. Severe Convective Storms, C. A. Doswell III, Ed., American Meteorological Society, 481- 525.
    Ding Y. H., J. Y. Zhang, 2009: Heavy Rain and Flood. China Meteorological Press, 290 pp. (in Chinese)
    Dong Q., X. Chen, and T. X. Chen, 2011: Characteristics and changes of extreme precipitation in the Yellow-Huaihe and Yangtze-Huaihe Rivers Basins, China. J.Climate, 24, 3781- 3795.10.1175/2010JCLI3653.1184be61bf0a0937d3885a0032aec9b66http%3A%2F%2Fadsabs.harvard.edu%2Fabs%2F2011JCli...24.3781Dhttp://adsabs.harvard.edu/abs/2011JCli...24.3781DMany works suggest that the intensity of extreme precipitation might be changing under the background of global warming. Because of the importance of extreme precipitation in the Yellow--Huaihe and Yangtze--Huaihe River basins of China and to compare the spatial difference, the generalized Pareto distribution (GPD) function is used to fit the daily precipitation series in these basins and an estimate of the extreme precipitation spatial distribution is presented. At the same time, its long-term trends are estimated for the period between 1951 and 2004 by using a generalized linear model (GLM), which is based on GPD. High quality daily precipitation data from 215 observation stations over the area are used in this study. The statistical significance of the trend fields is tested with a Monte Carlo experiment based on a two-dimensional Hurst coefficient, H2. The spatial distribution of the shape parameter of GPD indicates that the upper reaches of the Huaihe River (HuR) basin have the largest probability of extreme rainfall events, which is consistent with most historical flood records in this region. Spatial variations in extreme precipitation trends are found and show significant positive trends in the upper reaches of Poyang Lake in the Yangtze River (YaR) basin and a significant negative trend in the mid- to lower reaches of the Yellow River (YeR) and Haihe River (HaR) basins. The trends in the HuR basin and the lower reaches of Poyang Lake in the YaR basin are nearly neutral. All trend fields are significant at the 5%% level of significance from the Monte Carlo experiments.
    Frich P., L. V. Alexand er, P. Della-Marta B. Gleason, M. Haylock, A. M. G. Klein Tank, and T. Peterson, 2002: Observed coherent changes in climatic extremes during the second half of the twentieth century. Climate Research,19, 193-212, doi: 10.3354/cr019193.10.3354/cr01919381589def43f2db979886ad985d85dfe7http%3A%2F%2Fdx.doi.org%2F10.3354%2Fcr019193http://dx.doi.org/10.3354/cr019193ABSTRACT A new global dataset of derived indicators has been compiled to clarify whether frequency and/or severity of climatic extremes changed during the second half of the 20th century, This period provides the best spatial coverage of homogenous daily series, which can be used for calculating the proportion of global land area exhibiting a significant change in extreme or severe weather. The authors chose 10 indicators of extreme climatic events, defined from a larger selection, that could be applied to a large variety of climates. It was assumed that data producers were more inclined to release derived data in the form of annual indicator time series than releasing their original daily observations. The indicators are based on daily maximum and minimum temperature series, as well as daily totals of precipitation, and represent changes in all seasons of the year. Only time series which had 40 yr or more of almost complete records were used, A total of about 3000 indicator time series were extracted from national climate archives and collated into the unique dataset described here. Global maps showing significant changes from one multi-decadal period to another during the interval from 1946 to 1999 were produced. Coherent spatial patterns of statistically significant changes emerge, particularly an increase in warm summer nights, a decrease in the number of frost days and a decrease in intra-annual extreme temperature range. All but one of the temperature-based indicators show a significant change. Indicators based on daily precipitation data show more mixed patterns of change but significant increases have been seen in the extreme amount derived from wet spells and number of heavy rainfall events. We can conclude that a significant proportion of the global land area was increasingly affected by a significant change in climatic extremes during the second half of the 20th century. These clear signs of change are very robust; however, large areas are still not represented, especially Africa and South America.
    Gao R., X. K. Zou, Z. Y. Wang, and Q. Zhang, 2012: The Atlas of Extreme Weather and Climate Events in China. China Meteorological Press, 188 pp. (in Chinese)
    Garrett C., P. Müller, 2008: Supplement to extreme events. Bull. Amer. Meteor. Soc.,89, ES45-ES56, doi: 10.1175/2008 BAMS2566.2.10.1186/1297-9686-1-2-147e9ba96dc-a767-474a-b390-0d8fbd231762d9ab4b8984e218a535a800c6dbc7cabahttp%3A%2F%2Fadsabs.harvard.edu%2Fabs%2F2008BAMS...89S..45Grefpaperuri:(bc0a5375a143e8c99b4989c626993a5b)http://adsabs.harvard.edu/abs/2008BAMS...89S..45GNo Abstract available.
    Hosking J. R. M., 1990: L-moments: Analysis and estimation of distributions using linear combinations of order statistics. Journal of the Royal Statistical Society. Series B, 52, 105- 124.10.2307/23456533144146b-84bc-4955-85aa-75d7fdaaf1c3995ca37fb39f5659a2106def6521cc66http%3A%2F%2Fwww.jstor.org%2Fstable%2F2345653refpaperuri:(0195b52ba5700af69c360fc41047e23a)http://www.jstor.org/stable/2345653L-moments are expectations of certain linear combinations of order statistics. They can be defined for any random variable whose mean exists and form. the basis of a general theory which covers the summarization and description of theoretical probability distributions, the summarization and description of observed data samples, estimation of parameters and quantiles of probability distributions, and hypothesis tests for probability distributions. The theory involves such established procedures as the use of order statistics and Gini's mean difference statistic, and gives rise to some promising innovations such as the measures of skewness and kurtosis described in Section 2, and new methods of parameter estimation for several distributions. The theory of L-moments parallels the theory of (conventional) moments, as this list of applications might suggest. The main advantage of L-moments over conventional moments is that L-moments, being linear functions of the data, suffer less from the effects of sampling variability: L-moments are more robust than conventional moments to outliers in the data and enable more secure inferences to be made from small samples about an underlying probability distribution. L-moments sometimes yield more efficient parameter estimates than the maximum likelihood estimates.
    IPCC, 2013: Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change,T. F. Stocker et al., Eds., Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, 1535 pp.59dceddb1377b53b578b73a8eefab495http%3A%2F%2Fboris.unibe.ch%2F71452http://boris.unibe.ch/71452This is the report of Working Group I of the Intergovernmental Panel on Climatic Change (IPCC) Fourth Assessment, which describes progress in understanding of the human and natural drivers of climatic change, observed climatic change, climate processes and attribution and estimates of projected future climate change. It builds on past IPCC assessments and incorporates new findings from the past...
    Li J., R. C. Yu, and W. Sun, 2013a: Calculation and analysis of the thresholds of hourly extreme precipitation in mainland China. Torrential Rain and Disasters,32, 11-16, doi: 10.3969/j.issn.1004-9045.2013.01.002. (in Chinese)
    Li J., R. C. Yu, and W. Sun, 2013b: Duration and seasonality of hourly extreme rainfall in the central eastern China. Acta Meteor. Sinica,27, 799-807, doi: 10.1007/s13351-013-0604-y.10.1007/s13351-013-0604-y63593d9ec521267a1f376d47b2de517dhttp%3A%2F%2Fd.wanfangdata.com.cn%2FPeriodical%2Fqxxb-e201306003http://d.wanfangdata.com.cn/Periodical/qxxb-e201306003Compared with daily rainfall amount, hourly rainfall rate represents rainfall intensity and the rainfall process more accurately, and thus is more suitable for studies of extreme rainfall events. The distribution functions of annual maximum hourly rainfall amount at 321 stations in China are quantified by the Gen-eralized Extreme Value (GEV) distribution, and the threshold values of hourly rainfall intensity for 5-yr return period are estimated. The spatial distributions of the threshold exhibit significant regional differ-ences, with low values in northwestern China and high values in northern China, the mid and lower reaches of the Yangtze River valley, the coastal areas of southern China, and the Sichuan basin. The duration and seasonality of the extreme precipitation with 5-yr return periods are further analyzed. The average duration of extreme precipitation events exceeds 12 h in the coastal regions, Yangtze River valley, and eastern slope of the Tibetan Plateau. The duration in northern China is relatively short. The extreme precipitation events develop more rapidly in mountain regions with large elevation differences than those in the plain areas. There are records of extreme precipitation in as early as April in southern China while extreme rainfall in northern China will not occur until late June. At most stations in China, the latest extreme precipitation happens in August-September. The extreme rainfall later than October can be found only at a small por-tion of stations in the coastal regions, the southern end of the Asian continent, and the southern part of southwestern China.
    Luo Y. L., Y. Gong, and D.-L. Zhang, 2014: Initiation and organizational modes of an extreme-rain-producing mesoscale convective system along a Mei-yu front in East China. Mon. Wea. Rev., 142, 203- 221.10.1175/MWR-D-13-00111.1f7e06671004c87149ca403b878e4ceb8http%3A%2F%2Fwww.researchgate.net%2Fpublication%2F273635799_Initiation_and_organizational_modes_of_an_extreme-rain-producing_mesoscale_convective_system_along_a_Mei-yu_front_in_East_Chinahttp://www.researchgate.net/publication/273635799_Initiation_and_organizational_modes_of_an_extreme-rain-producing_mesoscale_convective_system_along_a_Mei-yu_front_in_East_ChinaAbstract The initiation and organization of a quasi-linear extreme-rain-producing mesoscale convective system (MCS) along a mei-yu front in east China during the midnight-to-morning hours of 8 July 2007 are studied using high-resolution surface observations and radar reflectivity, and a 24-h convection-permitting simulation with the nested grid spacing of 1.11 km. Both the observations and the simulation reveal that the quasi-linear MCS forms through continuous convective initiation and organization into west–east-oriented rainbands with life spans of about 4–10 h, and their subsequent southeastward propagation. Results show that the early convective initiation at the western end of the MCS results from moist southwesterly monsoonal flows ascending cold domes left behind by convective activity that develops during the previous afternoon-to-evening hours, suggesting a possible linkage between the early morning and late afternoon peaks of the mei-yu rainfall. Two scales of convective organization are found during the MCS's development: one is the east- to northeastward “echo training” of convective cells along individual rainbands, and the other is the southeastward “band training” of the rainbands along the quasi-linear MCS. The two organizational modes are similar within the context of “training” of convective elements, but they differ in their spatial scales and movement directions. It is concluded that the repeated convective backbuilding and the subsequent echo training along the same path account for the extreme rainfall production in the present case, whereas the band training is responsible for the longevity of the rainbands and the formation of the quasi-linear MCS.
    Ma Y., X. Wang, and Z. Y. Tao, 1997: Geographic distribution and life cycle of mesoscale convective system in China and its vicinity. Progress in Natural Science, 7, 701- 706.10.1007/s002690050078fc184ddc8713047fd86c7602bd6c1f3dhttp%3A%2F%2Fwww.cnki.com.cn%2FArticle%2FCJFDTotal-ZKJY199705009.htmhttp://www.cnki.com.cn/Article/CJFDTotal-ZKJY199705009.htm正 A census of mesoscale convective systems (MCS) has been extended from mesoscale convective complexes (MCCs) to more general meso-a scale convective systems (Ma CSs) and meso-β scale convective systems (Mβ CSs). 234 Ma CSs and 585 Mβ CSs were found in China and its vicinity during the summers of 1993-1995 by the GMS satellite infrared images. The geographic distribution with higher representative shows that the Ma CSs occurred in three favorable zones. One of them, the middle-to-lower reaches basin of the Yellow River and the Yangtze River, were not found in the past researches of MCC census. There are two kinds of life cycles of Ma CS: one is similar to the life cycle of MCC occurring at night and dissipating early in the morning; the other occurs in the afternoon and dissipates in the evening.
    Sen Roy, S., 2009: A spatial analysis of extreme hourly precipitation patterns in India. International Journal of Climatology,29, 345-355, doi: 10.1002/joc.1763.10.1002/joc.1763fdd24c667c2148211a1ffb11894c3c30http%3A%2F%2Fonlinelibrary.wiley.com%2Fdoi%2F10.1002%2Fjoc.1763%2Ffullhttp://xueshu.baidu.com/s?wd=paperuri%3A%2863b2f9c15f0f257d5aa28b61348575c9%29&filter=sc_long_sign&tn=SE_xueshusource_2kduw22v&sc_vurl=http%3A%2F%2Fonlinelibrary.wiley.com%2Fdoi%2F10.1002%2Fjoc.1763%2Ffull&ie=utf-8&sc_us=6643647539805733401Station level hourly precipitation data from 1980 to 2002 spread across the Indian subcontinent were analysed for trends in extreme hourly precipitation events. The analyses were conducted for the main seasons of winter, dry-summer, and wet-summer monsoon seasons, respectively. The results of the study indicated rising trends in extreme heavy precipitation events, mostly in the high-elevation r...
    Tao S. Y., 1980: Heavy Rains in China. China Science Press, 225 pp. (in Chinese)
    Tao Z. Y., Y. G. Zheng, 2013: Forecasting issues of the extreme heavy rain in Beijing on 21 July 2012. Torrential Rain and Disasters,32, 193-201, doi: 10.3969/j.issn.1004-9045.2013.03.001. (in Chinese)aec6a1ce8453060a3b33859d41fe4348http%3A%2F%2Fen.cnki.com.cn%2FArticle_en%2FCJFDTOTAL-HBQX201303001.htmhttp://en.cnki.com.cn/Article_en/CJFDTOTAL-HBQX201303001.htmBased on the upper air sounding,NWP model forecasts,satellite,radar and surface weather data,the extreme heavy rain event in Beijing during July 21-22,2012 is analyzed and its forecasting process is summarized.The results are as follows.(1) The"7-21"heavy rain was produced by a typical mesoscale convective complex(MCC) that occurred in the left front of a low-level jet in the warm and moist southerly,on the right of the entrance jet stream,with deep warm advection and clockwise wind direction with altitude,which are very favorable for the occurrence of MCC.(2) A variety of numerical models have predicted the heavy rain with skill in various degrees,and the lead time is up to 3-4 d.The short-term forecasting with lead time 1-2 d can be more accurately predict the occurrence area and intensity of heavy rainfall.But the prediction of heavy rain's start and end times is noticeably delayed about 6 h.(3) Satellite and radar monitoring indicates that 12 h before the arrival of a large-scale rain band,the initial warm convection has occurred ahead of the front.According to the movement,enhancement and organization of radar echoes,it can be extrapolated that the first stage of the heavy rain will affect Beijing at around noon,and thus timely corrections can be made to the numerical predictions.Using comprehensive analysis of radar echo and the surface weather(such as wind,dew point) fields,we can roughly determine the convective instability condition and the uplift condition favorable for the initial occurrence of convection,which can provide the basis for echo extrapolation forecasts.
    Wang Y., Z. W. Yan, 2011: Changes of frequency of summer precipitation extremes over the Yangtze river in association with large-scale oceanic-atmospheric conditions. Adv. Atmos. Sci.,28, 1118-1128, doi: 10.1007/s00376-010-0128-7.10.1007/s00376-010-0128-70b0a3b8cb98132132121d516767eaf5chttp%3A%2F%2Fd.wanfangdata.com.cn%2FPeriodical_dqkxjz-e201105013.aspxhttp://d.wanfangdata.com.cn/Periodical_dqkxjz-e201105013.aspxChanges of the frequency of precipitation extremes (the number of days with daily precipitation exceeding the 90th percentile of a daily climatology,referred to as R90N) in summer (June-August) over the mid-lower reaches of the Yangtze River are analyzed based on daily observations during 1961-2007.The first singular value decomposition (SVD) mode of R90N is linked to an ENSO-like mode of the sea surface temperature anomalies (SSTA) in the previous winter.Responses of different grades of precipitation events to the climatic mode are compared.It is notable that the frequency of summer precipitation extremes is significantly related with the SSTA in the Pacific,while those of light and moderate precipitation are not.It is suggested that the previously well-recognized impact of ENSO on summer rainfall along the Yangtze River is essentially due to a response in summer precipitation extremes in the region,in association with the East Asia-Pacific (EAP) teleconnection pattern.A negative relationship is found between the East Asian Summer Monsoon (EASM) and precipitation extremes over the mid-lower reaches of the Yangtze River.In contrast,light rainfall processes are independent from the SST and EASM variations.
    Yu R. C., J. Li, 2012: Hourly rainfall changes in response to surface air temperature over eastern contiguous China. J.Climate, 25, 6851- 6861.10.1175/JCLI-D-11-00656.1cff43ede691866c68834b8bacd2d889chttp%3A%2F%2Fadsabs.harvard.edu%2Fabs%2F2012JCli...25.6851Yhttp://adsabs.harvard.edu/abs/2012JCli...25.6851YNot Available
    Zhai P. M., X. B. Zhang, H. Wan, and X. H. Pan, 2005: Trends in total precipitation and frequency of daily precipitation extremes over China. J. Climate,18, 1096-1108, doi: 10.1175/ JCLI-3318.1.2d61d9895684e47d5a5e3b375e6105edhttp%3A%2F%2Fjxb.oxfordjournals.org%2Fexternal-ref%3Faccess_num%3D10.1175%2FJCLI-3318.1%26link_type%3DDOIhttp://xueshu.baidu.com/s?wd=paperuri%3A%28a54cd39e921360c2e30ad9d967965bc5%29&filter=sc_long_sign&tn=SE_xueshusource_2kduw22v&sc_vurl=http%3A%2F%2Fjxb.oxfordjournals.org%2Fexternal-ref%3Faccess_num%3D10.1175%2FJCLI-3318.1%26link_type%3DDOI&ie=utf-8&sc_us=1612413880034890929
    Zhai P. M., A. J. Sun, F. M. Ren, X. N. Liu, B. Gao, and Q. Zhang, 1999: Changes of climate extremes in China. Climatic Change,42, 203-218, doi: 10.1023/A:1005428602279.10.1023/A:1005428602279ae7fd00741ae2c231dab1f3e272172d8http%3A%2F%2Fonlinelibrary.wiley.com%2Fresolve%2Freference%2FXREF%3Fid%3D10.1023%2FA%3A1005428602279http://onlinelibrary.wiley.com/resolve/reference/XREF?id=10.1023/A:1005428602279Changes in China's temperature and precipitation extremes have been studied by using observational data after 1950. The results reveal that mean minimum temperature has increased significantly in China during the past 40 years, especially in the winter in northern China. Meanwhile, nation-wide cold wave activity has weakened and the frequency of cold days in northern China has been reduced significantly. Mean maximum temperatures display no statistically significant trend for China as a whole. However, decreasing summer mean maximum temperatures are obvious in eastern China, where the number of hot days has been reduced. Seasonal 1-day extreme maximum temperatures mainly reflect decreasing trends, while seasonal 1-day extreme minimum temperatures are increasing. A statistically significant reduction of much above normal rain days in China has been detected. Contrarily, an increasing trend was detected in much above normal of precipitation intensity (precipitation/number of precipitation days) during the past 45 years.
    Zhang H., P. M. Zhai, 2011: Temporal and spatial characteristics of extreme hourly precipitation over eastern China in the warm season. Adv. Atmos. Sci.,28, 1177-1183, doi: 10.1007/s00376-011-0020-0.10.1007/s00376-011-0020-0e8c4b593a69b036cacf1dbda66a132e5http%3A%2F%2Fwww.cnki.com.cn%2FArticle%2FCJFDTotal-DQJZ201105018.htmhttp://d.wanfangdata.com.cn/Periodical_dqkxjz-e201105017.aspxBased on hourly precipitation data in eastern China in the warm season during 1961 2000,spatial distributions of frequency for 20 mm h-1 and 50 mm h-1 precipitation were analyzed,and the criteria of short-duration rainfall events and severe rainfall events are discussed.Furthermore,the percentile method was used to define local hourly extreme precipitation; based on this,diurnal variations and trends in extreme precipitation were further studied.The results of this study show that,over Yunnan,South China,North China,and Northeast China,the most frequent extreme precipitation events occur most frequently in late afternoon and/or early evening.In the Guizhou Plateau and the Sichuan Basin,the maximum frequency of extreme precipitation events occursin the late night and/or early morning.And in the western Sichuan Plateau,the maximum frequency occursin the middle of the night.The frequency of extreme precipitation (based on hourly rainfall measurements) has increased in mostparts of eastern China,especially in Northeast China and the middle and lower reaches of the Yangtze River,but precipitation has decreased significantly in North China in the past 50 years.In addition,stations inthe Guizhou Plateau and the middle and lower reaches of the Yangtze River exhibit significant increasing trends in hourly precipitation extremes during the nighttime more than during the daytime.
    Zhang J. C., Z. G. Lin, 1985: Climate of China. Shanghai Science and Technology Press, 436 pp. (in Chinese)
    Zhao Y. Y., Q. H. Zhang, Y. Du, M. Jiang, and J. P. Zhang, 2013: Objective analysis of circulation extremes during the 21 July 2012 torrential rain in Beijing. Acta Meteorologica Sinica,27, 626-635, doi: 10.1007/s13351-013-0507-y.10.1007/s13351-013-0507-yc8011063f5914ee6fb04a38b20994c59http%3A%2F%2Fwww.cnki.com.cn%2FArticle%2FCJFDTotal-QXXW201305002.htmhttp://d.wanfangdata.com.cn/Periodical/qxxb-e201305002
    Zheng Y. G., J. Chen, 2013: A climatology of deep convection over South China and the adjacent waters during summer. Journal of Tropical Meteorology, 19, 1- 15.10.1175/JTECH-D-12-00105.1ec2f402d82c3fae3a1b5a4cabaa7d7f6http%3A%2F%2Fwww.cqvip.com%2FQK%2F85390X%2F201301%2F1002057481.htmlhttp://www.cnki.com.cn/Article/CJFDTotal-RQXB201301003.htmDue to the higher temporal and spatial resolution and the better integrality of long-term satellite infrared(IR) Brightness Temperature(TBB) data,a climatology of deep convection during summer over South China and the adjacent waters is presented in this paper based on the 1-hourly infrared IR TBB data during June-August of 1996-2007(except 2004).The results show that the geographic distribution of deep convection denoted by TBB over South China and the adjacent waters are basically consistent with previous statistical results based on surface thunderstorm observations and low-orbit satellite lightning observations.The monthly,ten-day,five-day and diurnal variations of deep convection in this region are focused on in this paper.There are 5 active deep-convection areas in June-August.The monthly variations of the deep convection are closely associated with the large-scale atmospheric circulations.The deep convection over the land areas of South China is more active in June while that over the South China Sea is more active in July and August.The development of deep convection is prominently intermittent and its period is about 3 to 5 five-day periods.However,the deep convection over the coastal areas in South China remains more active during summer and has no apparent intermittence.The ten-day and five-day variations of deep convection show that there are different variations of deep convection over different areas in South China and the adjacent waters.The tendency of deep convection over the land areas of South China is negatively correlated with that over the South China Sea.The diurnal variations of deep convection show that the sea-land breeze,caused by the thermal differences between land and sea,and the mountain-valley breeze,caused by the thermal differences between mountains and plains or basins,cause deep convection to propagate from sea to land in the afternoon and from land to sea after midnight,and the convection over mountains propagates from mountains to plains after midnight.The different diurnal variations of deep convection over different underlying surfaces show that not only there are general mountainous,marine and multi-peak deep convection,but also there is longer-duration deep convection over coastal areas and other deep convection triggered and maintained by larger-scale weather systems in South China during summer.
    Zheng Y. G., J. Chen, and P. J. Zhu, 2008: Climatological distribution and diurnal variation of mesoscale convective systems over China and its vicinity during summer. Chinese Sci. Bull.,53, 1574-1586, doi: 10.1007/s11434-008-0116-9.10.1007/s11434-008-0116-9acd920c76e21caadef6faf173899cc13http%3A%2F%2Flink.springer.com%2F10.1007%2Fs11434-008-0116-9http://www.cnki.com.cn/Article/CJFDTotal-JXTW200810018.htmThe climatological distribution of mesoscale convective systems (MCSs) over China and its vicinity during summer is statistically analyzed, based on the 10-year (1996―2006, 2004 excluded) June-August infrared TBB (Temperature of black body) dataset. Comparing the results obtained in this paper with the distribution of thunderstorms from surface meteorological stations over China and the distribution of lightning from low-orbit satellites over China and its vicinity in the previous studies, we find that the statistic characteristics of TBB less than -52℃ can better represent the spatiotemporal distribution of MCSs over China and its vicinity during summer.The spreading pattern of the MCSs over this region shows three transmeridional bands of active MCSs, with obvious fluctuation of active MCSs in the band near 30°N. It can be explained by the atmospheric circulation that the three bands of active MCSs are associated with each other by the summer monsoon over East Asia. We focus on the diurnal variations of MCSs over different underlying surfaces, and the result shows that there are two types of MCSs over China and its vicinity during summer. One type of MCSs has only one active period all day long (single-peak MCSs), and the other has multiple active periods (multi-peak MCSs). Single-peak MCSs occur more often over plateaus or mountains, and multi-peak MCSs are more common over plains or basins. Depending on lifetimes and active periods, single-peak MCSs can be classified as Tibetan Plateau MCSs, general mountain MCSs, Ryukyu MCSs, and so on. The diurnal variation of multi-peak MCSs is very similar to that of MCCs (mesoscale convective complexes), and it reveals that multi-peak MCSs has longer life cycle and larger horizontal scale, becomes weaker after sunset, and develops again after midnight. Tibetan Plateau MCSs and general mountain MCSs both usually develop in the afternoon, but Tibetan Plateau MCSs have longer life cycle and more active MαCSs. Ryukyu MCSs generally develop after midnight, last longer time, and also have more active MαCS. The abundant moisture and favorable large-scale environment over Indian monsoon surge areas lead to active MCSs and MαCSs almost at any hour all day during summer. Due to local mountain-valley breeze circulation over the Sichuan Basin, MCSs are developed remarkably more often during the nighttime, and again there are also more active MαCSs. Because of local prominent sea-land breeze circulation over Guangxi and Guangdong, the MCSs over this region propagate from sea to land in the afternoon and from land to sea after midnight. The statistic characteristics of TBB less than -52℃ clearly display the different climatological characteristics of MCSs owing to the thermal difference among water, land and rough terrain. Not only the large-scale atmospheric circulation but also the local atmospheric circulation caused by the thermal difference among water, land and rough terrain, to a great extent, determines the climatological distribution of MCSs over China and its vicinity during summer.
    Zheng Y. G., J. Chen, and Z. Y. Tao, 2014: Distribution characteristics of the intensity and extreme intensity of tropical cyclones influencing China. J. Meteor. Res.,28, 393-406, doi: 10.1007/s13351-014-3050-6.10.1007/s13351-014-3050-6ec49f9d24cf20ae376c1d169cf2bbb6chttp%3A%2F%2Fwww.cqvip.com%2FQK%2F88418X%2F201403%2F50299543.htmlhttp://d.wanfangdata.com.cn/Periodical/qxxb-e201403006To address the deficiency of climatological research on tropical cyclones(TCs) influencing China, we analyze the distributions of TCs with different intensities in the region, based on the best-track TC data for1949-2011 provided by the Shanghai Typhoon Institute. We also present the distributions of 50- and 100-yr return-period TCs with different intensities using the Gumbel probability distribution. The results show that TCs with different intensities exert distinctive effects on various regions of China and its surrounding waters. The extreme intensity distributions of TCs over these different regions also differ. Super and severe typhoons mainly influence Taiwan Island and coastal areas of Fujian and Zhejiang provinces, while typhoons and TCs with lower intensities influence South China most frequently. The probable maximum TC intensity(PMTI) with 50- and 100-yr return periods influencing Taiwan Island is below 890 hPa; the PMTI with a50-yr return period influencing the coastal areas of Fujian and Zhejiang provinces is less than 910 hPa, and that with a 100-yr return period is less than 900 hPa; the PMTI with a 50-yr return period influencing the coastal areas of Hainan, Guangdong, and the northern part of the South China Sea is lower than 930 hPa,and that with a 100-yr return period is less than 920 hPa. The results provide a useful reference for the estimation of extreme TC intensities over different regions of China.
  • [1] Gang FU, Yawen SUN, Jilin SUN, Pengyuan LI, 2020: A 38-Year Climatology of Explosive Cyclones over the Northern Hemisphere, ADVANCES IN ATMOSPHERIC SCIENCES, 37, 143-159.  doi: 10.1007/s00376-019-9106-x
    [2] LI Jiawei, HAN Zhiwei, 2012: A Modeling Study of Seasonal Variation of Atmospheric Aerosols over East Asia, ADVANCES IN ATMOSPHERIC SCIENCES, 29, 101-117.  doi: 10.1007/s00376-011-0234-1
    [3] Ying LI, Dajun ZHAO, 2022: Climatology of Tropical Cyclone Extreme Rainfall over China from 1960 to 2019, ADVANCES IN ATMOSPHERIC SCIENCES, 39, 320-332.  doi: 10.1007/s00376-021-1080-4
    [4] ZHANG Chengwei, YU Fan, WANG Chenxi, YANG Jianyu, 2011: Three-dimensional Extension of the Unit-Feature Spatial Classification Method for Cloud Type, ADVANCES IN ATMOSPHERIC SCIENCES, 28, 601-611.  doi: 10.1007/s00376-010-9056-9
    [5] Juan HUO, Yongheng BI, Daren Lü, Shu DUAN, 2019: Cloud Classification and Distribution of Cloud Types in Beijing Using Ka-Band Radar Data, ADVANCES IN ATMOSPHERIC SCIENCES, , 793-803.  doi: 10.1007/s00376-019-8272-1
    [6] Yue SUN, Huiling YANG, Hui XIAO, Liang FENG, Wei CHENG, Libo ZHOU, Weixi SHU, Jingzhe SUN, 2024: The Spatiotemporal Distribution Characteristics of Cloud Types and Phases in the Arctic Based on CloudSat and CALIPSO Cloud Classification Products, ADVANCES IN ATMOSPHERIC SCIENCES, 41, 310-324.  doi: 10.1007/s00376-023-2231-6
    [7] Li Jun, Zhou Fengxian, Wang Luyi, 1992: Automatic Classification and Compression of GMS Cloud Imagery in Heavy Rainfall Monitoring Application, ADVANCES IN ATMOSPHERIC SCIENCES, 9, 458-464.  doi: 10.1007/BF02677078
    [8] Xiaohu XIONG, Weijian ZHOU, Shugang WU, Peng CHENG, Hua DU, Yaoyao HOU, Zhenchuan NIU, Peng WANG, Xuefeng LU, Yunchong FU, 2020: Two-Year Observation of Fossil Fuel Carbon Dioxide Spatial Distribution in Xi’an City, ADVANCES IN ATMOSPHERIC SCIENCES, 37, 569-575.  doi: 10.1007/s00376-020-9241-4
    [9] ZHANG Dingyuan, LIAO Hong, WANG Yuesi, 2014: Simulated Spatial Distribution and Seasonal Variation of Atmospheric Methane over China: Contributions from Key Sources, ADVANCES IN ATMOSPHERIC SCIENCES, 31, 283-292.  doi: 10.1007/s00376-013-3018-y
    [10] LIN Pengfei, LIU Hailong, ZHANG Xuehong, 2008: Effect of Chlorophyll-a Spatial Distribution on Upper Ocean Temperature in the Central and Eastern Equatorial Pacific, ADVANCES IN ATMOSPHERIC SCIENCES, 25, 585-596.  doi: 10.1007/s00376-008-0585-4
    [11] TANG Yanbing, GAN Jingjing, ZHAO Lu, GAO Kun, 2006: On the Climatology of Persistent Heavy Rainfall Events in China, ADVANCES IN ATMOSPHERIC SCIENCES, 23, 678-692.  doi: 10.1007/s00376-006-0678-x
    [12] Iman ROUSTA, Mehdi DOOSTKAMIAN, Esmaeil HAGHIGHI, Hamid Reza GHAFARIAN MALAMIRI, Parvane YARAHMADI, 2017: Analysis of Spatial Autocorrelation Patterns of Heavy and Super-Heavy Rainfall in Iran, ADVANCES IN ATMOSPHERIC SCIENCES, 34, 1069-1081.  doi: 10.1007/s00376-017-6227-y
    [13] ZHANG Huan, ZHAI Panmao, 2011: Temporal and Spatial Characteristics of Extreme Hourly Precipitation over Eastern China in the Warm Season, ADVANCES IN ATMOSPHERIC SCIENCES, 28, 1177-1183.  doi: 10.1007/s00376-011-0020-0
    [14] YUE Caijun, GAO Shouting, LIU Lu, LI Xiaofan, 2015: A Diagnostic Study of the Asymmetric Distribution of Rainfall during the Landfall of Typhoon Haitang (2005), ADVANCES IN ATMOSPHERIC SCIENCES, 32, 1419-1430.  doi: 10.1007/s00376-015-4246-0
    [15] Israel LOPEZ-COTO, Subhomoy GHOSH, Kuldeep PRASAD, James WHETSTONE, 2017: Tower-Based Greenhouse Gas Measurement Network Design——The National Institute of Standards and Technology North East Corridor Testbed, ADVANCES IN ATMOSPHERIC SCIENCES, 34, 1095-1105.  doi: 10.1007/s00376-017-6094-6
    [16] Dajun ZHAO, Yubin YU, Lianshou CHEN, 2021: Impact of the Monsoonal Surge on Extreme Rainfall of Landfalling Tropical Cyclones, ADVANCES IN ATMOSPHERIC SCIENCES, 38, 771-784.  doi: 10.1007/s00376-021-0281-1
    [17] Ming XUE, 2023: Preface to the Special Collection on the July 2021 Zhengzhou, Henan Extreme Rainfall Event, ADVANCES IN ATMOSPHERIC SCIENCES, 40, 335-336.  doi: 10.1007/s00376-022-2018-1
    [18] Xiao PAN, Tim LI, Ying SUN, Zhiwei ZHU, 2021: Cause of Extreme Heavy and Persistent Rainfall over Yangtze River in Summer 2020, ADVANCES IN ATMOSPHERIC SCIENCES, 38, 1994-2009.  doi: 10.1007/s00376-021-0433-3
    [19] Wenhua Gao, Chengyin Li, Lanzhi Tang, 2024: A numerical study of the impacts of hydrometeor processes on the “21.7” extreme rainfall in Zhengzhou, China, ADVANCES IN ATMOSPHERIC SCIENCES.  doi: 10.1007/s00376-024-3365-x
    [20] Philip E. BETT, Gill M. MARTIN, Nick DUNSTONE, Adam A. SCAIFE, Hazel E. THORNTON, Chaofan LI, 2021: Seasonal Rainfall Forecasts for the Yangtze River Basin in the Extreme Summer of 2020, ADVANCES IN ATMOSPHERIC SCIENCES, 38, 2212-2220.  doi: 10.1007/s00376-021-1087-x

Get Citation+

Export:  

Share Article

Manuscript History

Manuscript received: 05 May 2016
Manuscript revised: 26 July 2016
Manuscript accepted: 18 August 2016
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Spatial Characteristics of Extreme Rainfall over China with Hourly through 24-Hour Accumulation Periods Based on National-Level Hourly Rain Gauge Data

  • 1. National Meteorological Centre, Beijing 100081, China
  • 2. School of Atmospheric Sciences, Nanjing University, Nanjing 210023, China
  • 3. Center for Analysis and Prediction of Storms, Oklahoma University, Norman OK 73072, USA
  • 4. University of Illinois at Urbana-Champaign, Champaign, IL 61820, USA
  • 5. Peking University, Beijing 100871, China

Abstract: Hourly rainfall measurements of 1919 national-level meteorological stations from 1981 through 2012 are used to document, for the first time, the climatology of extreme rainfall in hourly through 24-h accumulation periods in China. Rainfall amounts for 3-, 6-, 12- and 24-h periods at each station are constructed through running accumulation from hourly rainfall data that have been screened by proper quality control procedures. For each station and for each accumulation period, the historical maximum is found, and the corresponding 50-year return values are estimated using generalized extreme value theory. Based on the percentiles of the two types of extreme rainfall values among all the stations, standard thresholds separating Grade I, Grade II and Grade III extreme rainfall are established, which roughly correspond to the 70th and 90th percentiles for each of the accumulation periods. The spatial characteristics of the two types of extreme rainfall are then examined for different accumulation periods. The spatial distributions of extreme rainfall in hourly through 6-h periods are more similar than those of 12- and 24-h periods. Grade III rainfall is mostly found over South China, the western Sichuan Basin, along the southern and eastern coastlines, and in the large river basins and plains. There are similar numbers of stations with Grade III extreme hourly rainfall north and south of 30°N, but the percentage increases to about 70% south of 30°N as the accumulation period increases to 24 hours, reflecting richer moisture and more prolonged rain events in southern China. Potential applications of the extreme rainfall climatology and classification standards are suggested at the end.

1. Introduction
  • Extreme weather and climate events are receiving increasing attention due to their great threat to people's lives and properties. For example, extremely heavy rainfall can cause human casualties, urban flooding, river overflow, landslides, and other forms of disastrous consequences. Extreme weather and climate events are usually defined as low-probability events for particular times and locations, often with a probability of occurrence lower than 10% (e.g., IPCC, 2013). Therefore, the probability for an extreme event is usually discussed in terms of percentiles, and the 95th percentile is commonly used as the threshold (e.g., Frich et al., 2002;Zhai et al., 2005). To date, there have been numerous studies on extreme weather and climate events, but most have focused on their detection, spatial distributions, and climate change characteristics (e.g., Frich et al., 2002; Garrett and Müller, 2008; Sen Roy, 2009). Within China, Zhai et al. (1999, 2005) studied the spatial distributions of extreme daily temperature and rainfall, as well as their climatological trends of change, based on a dataset of 349 meteorological stations during 1951-95 and another dataset of 740 stations during 1951-2000. (Gao et al., 2012) detailed the spatial distributions of a number of extreme weather and climate events in China, including the extreme daily and 3-day precipitation, using a dataset from 1031 meteorological stations in China during 1951-2011.

    Figure 1.  The (a) topography of China, and (b) locations of stations with continuous observations of hourly rainfall for 1965-2012 (orange dots) and 1981-2012 (green dots). In (b), thick solid lines separate various regions marked by numbers: 1——South China; 2——Guizhou and Hunan provinces, and most parts of Jiangxi Province; 3——eastern Jiangxi Province and inland areas of Fujian and Zhejiang provinces; 4——coastal areas of Fujian and Zhejiang provinces; 5——the Sichuan Basin; 6——Hubei Province; 7——the Yangtze River-Huaihe River Basins; 8——the Huanghe River-Huaihe River Basins; 9——the Shandong Peninsula; 10——the North China Plain; 11——southern Liaoning Province.

    Due to the unavailability of long-term hourly rainfall data in China (Fig. 1), hardly any research exists prior to 2010 on extreme rainfall for accumulation periods shorter than 24 hours. An hourly rainfall event of ≥ 20 mm is commonly referred to as a short-duration heavy rainfall (SDHR) event, which is rare in China and the United States (Davis, 2001; Zhang and Zhai, 2011; Chen et al., 2013). (Zhang and Zhai, 2011) presented the temporal and spatial distributions and the climatological trend of extreme hourly rainfall with intensities greater than 20 mm h-1 and 50 mm h-1. The study focused on central and eastern China for May-September, using hourly rainfall data from 480 meteorological stations during 1961-2000. (Chen et al., 2013) documented the temporal and spatial characteristics of SDHR events of no less than 10, 20, 30, 40 and 50 mm h-1 over China during April-September using hourly rainfall data from 549 stations for 1991-2009. Neither study, however, analyzed the spatial distributions of extreme rainfall for different return periods of hourly rainfall. Using the probability distribution of the generalized extreme value (GEV; Coles, 2001), and based on hourly rainfall data from 465 and 321 stations in China, respectively, Li et al. (2013a, 2013b) presented the return values and their spatial characteristics for 2-, 5-, 10- and 50-yr return periods. However, they did not examine and analyze in detail the differences among extreme rainfall events for accumulation periods from hourly through 24 hours. Despite these previous investigations, issues and problems remain, as follows:

    (1) The meteorological station data used in previous studies were all very sparse, with the number of stations considered usually less than 600, meaning those studies may not have fully captured the extreme rainfall events produced by meso- or convective-scale systems.

    (2) There has been no research on the spatial distributions of extreme rainfall in accumulation periods of between 1 and 24 hours in China. Previous studies on extreme precipitation in China either focused on daily or hourly rainfall (e.g., Zhai et al., 1999, 2005; Dong et al., 2011; Wang and Yan, 2011; Gao et al., 2012; Yu and Li, 2012; Li et al., 2013a, 2013b); and the 3-, 6-, 12- and 24-h running cumulative rainfall amounts at each hour have not been examined. Here, the running accumulations are calculated in the same way as the moving average, except summation is taken instead of averaging. The use of daily rainfall, rather than 24-hour running accumulation from hourly rainfall, may underestimate extreme rainfall that straddles the recording day.

    (3) There is thus far no classification standard, based on statistically determined thresholds, for extreme rainfall in different accumulation periods in China.

    Because the occurrence of extreme rainfall at any single meteorological station carries a very low probability, the prediction of such highly improbable events is very difficult. However, if a dataset from a large number of meteorological stations covers a sufficiently long time period, it is possible to estimate the distributions of extreme events and thereby provide useful information for improving the prediction of such rare events. For these reasons, utilizing hourly rainfall data at 2420 national-level meteorological stations in China that cover the period 1951-2012, we document and investigate the spatial distributions of two types of extreme rainfall, the historical maximum and the estimated 50-yr return value (hereafter, 50-yr rainfall), for running accumulation periods of 1, 3, 6, 12 and 24 hours. Based on such long-term historical data covering a large portion of China, we establish standards of classification for extreme rainfall, in terms of threshold values that separate three grades of extreme rainfall, for different accumulation periods. The thresholds roughly correspond to the 70th and 90th percentiles of extreme rainfall amounts among the stations. Our study allows us to obtain the spatial characteristics and classify different regions based on their extreme rainfall, and it also provides important reference information for the estimation and prediction of extreme rainfall in China (Fig. 1).

    Following this introduction, in section 2 we describe the data and analysis methods used. In sections 3 and 4, we document and discuss the spatial distributions of historical rainfall maxima and 50-yr return values, respectively. Section 5 examines the regional distributions of extreme rainfall. A summary and conclusions are given in section 6.

2. Data and methods
  • The hourly rainfall dataset during 1951-2012 was obtained from the National Meteorological Information Center of the China Meteorological Administration. In this dataset, the rainfall was measured by either tipping-buckets or self-recording siphon rain gauges, or from automatic rain gauges, at 2420 national-level meteorological stations in mainland China. The data were subject to strict quality control by the data provider according to the following rules. For each individual rain gauge on any single day, the difference between the observed daily rainfall and the accumulated daily value from hourly rainfall was calculated. The hourly rainfall data were considered erroneous if this difference exceeded a threshold: For daily rainfall ≥ 5 mm, the threshold was 20% of the daily amount; and for daily rainfall <5 mm the threshold was 1 mm. All erroneous data are discarded in this study.

    The number of meteorological stations available in the hourly rainfall dataset increased over the study period. In the 1950s, there were less than 1000 stations, but the number increased to more than 2000 after 1980. The number of stations taking observations in July is around two to three times greater than that of January, because a number of stations in northern China routinely stop taking rainfall measurement in the freezing conditions of the winter season under certain regulations. In general, the densest observations occur in central and eastern China. Although the spatial and temporal coverages of the hourly rainfall dataset are not homogeneous, this dataset represents the most complete and accurate measurements of hourly rainfall in China to date.

    For identifying extreme rainfall data series that cover the same climatological periods over China, we only select the stations that have at least 25 hourly-rainfall-observation days in the summer months (June, July and August) of each year. The reason for this screening is that China is significantly affected by the East Asian summer monsoon, and thus heavy rain and SDHR events mainly occur in summer (Ding and Zhang, 2009; Chen et al., 2013). Ultimately, 783 stations with continuous observations are selected for the period 1965-2012, and 1919 stations for the period 1981-2012, with the former being a subset of the latter (Fig. 1b). The average distance between the 1919 stations is about 50 km. The selected stations are mainly located in central and eastern China, east of 100°E, and only a few stations are situated in the Tibetan Plateau or in the western deserts, west of 100°E (Fig. 1b).

    To better capture extreme rainfall events, we use all available rainfall data from the 1919 stations for the period 1981-2012 to obtain the historical maximum and estimate the 50-yr return value at each station. Given that the observational periods of the 1919 stations cover more than 30 years, the rainfall data from these stations are regarded as carrying sufficient climatological information. To obtain the historical maximum rainfall series in different accumulation periods for each station, we first compute the 3-, 6-, 12- and 24-h running cumulative rainfall from the hourly rainfall data, and then find the historical rainfall maximum for each accumulation period from the complete series. This ensures a full account of extreme rainfall that straddles the rainfall accumulation periods. We obtain the spatial distributions of the historical rainfall maximum for both 1965-2012 and 1981-2012, and find that the spatial distributions of the two periods are similar for all accumulation periods, although their rainfall amounts are somewhat different. We show results from the latter period only because of its larger number of stations.

    Different regions of China are referred to in this paper. Figure 1a labels the provinces and four main rivers of China, while Fig. 1b divides and labels various regions. For brevity, we use the term "Northeast China" to refer to the provinces of Heilongjiang, Jilin and Liaoning. "North China" includes the cities of Beijing and Tianjin, and the provinces of Hebei and Shanxi; and "South China" comprises the provinces of Guangxi, Guangdong, and Hainan.

  • The historical maximum rainfall series in different accumulation periods at each station are considered random processes of extremes, and thus we use the GEV distribution to model the annual maxima, and then estimate the 50-yr rainfall amount for each station. The GEV distribution has been widely applied to extreme rainfall estimation (e.g., Coles, 2001; Gao et al., 2012; Li et al., 2013a, 2013b). The 50-yr rainfall is considered an extreme event according to the definition of extreme weather and climate events (e.g., IPCC, 2013). According to probability theory, for an event with a 50-yr return period, the probability of at least one such occurrence in 50 years is 63.6% (Atomic Energy Regulatory Board of India, 2008).

    The GEV cumulative probability distribution of variable z is defined as $$ G(z)=\exp\left\{-\left[1+\xi\left(\dfrac{z-\mu}{\sigma}\right)\right]^{-1/\xi}\right\} , $$ where G(z) is the probability that z is not exceeded (z means any value in the support of the distribution), and μ, σ and \(\xi\) are the location, scale and shape parameters, respectively. The parameters must satisfy \(1+\xi(z-\mu)/\sigma>0\), \(-\infty<\mu<\infty\), σ>0 and \(-\infty<\xi<\infty\) (Coles, 2001). Given an annual maximum sample series, one can estimate the parameters and then determine the cumulative probability function of the GEV either using the maximum likelihood method (Coles, 2001) or the L-moments method (Hosking, 1990). We choose the maximum likelihood method in our estimation. After obtaining the annual maximum rainfall series for a given accumulation period and a given station, we estimate parameters μ, σ and \(\xi\) of the GEV distribution, assuming the series is stationary. With the estimated GEV distribution function, we then estimate the rainfall amounts for different return periods.

    Two stations, Beijing in North China and Qingyuan in South China, are taken from those 783 stations for the period 1965-2012 as examples to show the reliability of the estimated GEV distribution function. The reason for choosing these two stations is that they both have relatively longer observational periods, and also represent different climate regions. For brevity, we show in Figs. 2 and 3 only the probability plots, the fitted GEV distributions and the 95% confidence intervals of hourly and 24-h rainfall, for the two stations respectively, in order to evaluate the goodness-of-fit of the fitted model.

    The fitted GEV distributions using the hourly, 3-, 6-, 12- and 24-h rainfall data of 1965-2012 and 1981-2012 (in Figs. 2 and 3, respectively, but without showing the fitted GEV distributions of 3-, 6- and 12-h rainfall) all agree well with the probability distributions of annual rainfall maxima. The fitted probability distributions using the two datasets are very similar. The confidence intervals for the estimated return level curves are wider for longer return periods, in particular for return periods longer than 50 years, which is not surprising. Therefore, in section 4, we only present the spatial distributions of estimated rainfall at the 50-yr return level, although all rainfall events with return periods no shorter than 50 years are considered extreme. In addition, due to the different lengths of the two datasets used, a number of differences between the fitted GEV distributions can also be seen. The 50-yr rainfall amounts from the fitted GEV distribution using the 1981-2012 dataset are higher than those using the 1965-2012 dataset, which may be related to the fact that the observational period of the former dataset is shorter, and the dataset features heavier rainfall amounts on average.

    The reliability of the estimated 50-yr rainfall across China is also tested by comparing the spatial distributions of the two estimates from the two datasets. They are found to be consistently similar for all of the different accumulation periods.

  • There is thus far no standard classification in China that is particularly designed for extreme rainfall in different accumulation periods; and all existing classifications are based on fixed amounts of rainfall, regardless of their accumulation periods. Rainfall amounts with different accumulation periods cannot be directly compared. Here, we propose a new standard categorization for classifying extreme rainfall according to their accumulation periods, and then further classify different regions based on their extreme rainfall classification. We suggest using the percentiles of the extreme rainfall over the 1919 stations in different accumulation periods to define the thresholds of classification. With the establishment of such standard thresholds, the extreme rainfall in different accumulation periods can be classified consistently, and thus the spatial distributions of the extreme rainfall in different accumulation periods can be compared, and the differences in extreme rainfall for different accumulation periods among various regions can be obtained.

    Figure 2.  Probability plots (a, b, e, f) and fitted GEV distributions (c, d, g, h) of (a-d) hourly and (e-h) 24-h rainfall at Beijing station, based on 1965-2012 (a, c, e, g) and 1981-2012 (b, d, f, h) data. Gray solid lines in (a, b, e, f) are the unit diagonals, and those in (c, d, g, h) indicate the 95% confidence intervals. Note that the vertical coordinate ranges in (c, d, g, h) are different, and the units are mm.

    Figure 3.  As in Fig. 2, but for Qingyuan station in Guangdong Province.

    For the historical maximum or the 50-yr rainfall in any accumulation period during 1981-2012, we first sort the extreme rainfall data series at the 1919 stations (there is only one extreme rainfall value at each station) in ascending order then determine the 70th and 90th percentile values across all stations. These values are given separately in Table 1 for the historical maximum and the 50-yr rainfall. As the historical maximum rainfall values differ slightly from their corresponding 50-yr return values, to facilitate the comparison of the spatial distributions between these two types of extreme rainfall and among different rainfall accumulation periods, we compute the threshold values of two levels for the extreme hourly, 3-, 6-, 12- and 24-h rainfall datasets mainly according to the 50-yr rainfall values in Table 1 (see Table 2). Table 2 shows that the threshold values for the low level (defined as Grade I precipitation) are located around the 69th percentile of the ordered historical maximum rainfall sequence, and around the 70th percentile of the ordered 50-yr rainfall sequence among the 1919 stations. Thresholds for the high level (defined as Grade III precipitation) correspond approximately to the 89th percentile of the ordered historical maximum rainfall sequence, and the 90th percentile of the ordered 50-yr rainfall sequence. Thus, three grades of extreme rainfall in Table 3 are proposed to classify and compare the spatial distributions among different types of extreme rainfall. In the following sections, we use the classification and threshold values defined above to examine the spatial distributions of extreme rainfall.

    Note that the Central Meteorological Office of China classifies daily rainfall of no less than 50 mm, 100 mm and 250 mm as heavy rainfall, very heavy rainfall, and extremely heavy rainfall, respectively (Ding and Zhang, 2009). Therefore, all the thresholds for Grade II and Grade III extreme rainfall, in different accumulation periods, as defined above, are much greater than that for the heavy rain threshold (50 mm) defined in China. Furthermore, except for the thresholds for Grade II (75 mm) and Grade III (95 mm) extreme hourly rainfall, all the other thresholds are greater than that of the very heavy rainfall threshold (100 mm). The threshold for Grade III extreme hourly rainfall (95 mm) approaches that of very heavy rainfall (100 mm), and the thresholds for Grade III extreme 12-h rainfall (260 mm) and Grade II extreme 24-h rainfall (230 mm) are close to that of extremely heavy rainfall (250 mm). Note that the threshold for Grade III extreme 24-h rainfall (305 mm) is much greater than that of extremely heavy rainfall (250 mm).

  • For the convenience of contour plotting, we utilize a grid of 0.75°× 0.75° latitude-longitude cells. We identify the maximum extreme rainfall amount within each of the cells for each accumulation period. For each grid cell, the maximum extreme rainfall amount is equal to the highest value among the stations within that grid cell. If no rainfall observation is found within a cell, that cell is assigned a missing value and is not contoured (the cell will be shown as white). Since the average distance among the 1919 stations is about 50 km, the 0.75° grid distance is somewhat greater than the average distance, so the use of this grid would smooth the spatial distribution somewhat where station density is high.

    The spatial distributions of the historical rainfall maxima and the estimated 50-yr rainfall are shown in Fig. 4 and Fig. 5, respectively, for different accumulation periods. Note that Figs. 4, 5 and 6 show only central and eastern China, as almost all of the 1919 stations used in this study lie over this region. Grades II and III are shown for all periods in dark blue and magenta colors, respectively. The 20 mm threshold is shown for hourly extreme rainfall, which corresponds to the definition of SDHR (Chen et al., 2013), while 50 mm is shown for all accumulation periods corresponding to the definition of daily heavy rainfall (Ding and Zhang, 2009) in China. In addition, the threshold value of Grade III extreme hourly rainfall (95 mm) is also presented for accumulation periods longer than 1 hour. In addition to the contour maps, stations with Grade II and Grade III extreme rainfall are plotted as light blue stars and yellow dots, respectively, in Figs. 4 and 5. While the contour maps are convenient for revealing the spatial distributions, in the next sections we focus our discussions more on the stations because they are more faithful to the original observations.

    As stated in the previous subsections, the spatial distributions of the historical maximum and the estimated 50-yr rainfall for the period 1965-2012 (not shown) are consistently similar to those for 1981-2012, regardless of their accumulation periods. However, considering that the latter data are taken from more stations, which can provide a finer-scale spatial representation, we only present the latter in this paper.

3. Spatial distributions of historical maximum rainfall
  • At a given station for a given accumulation period, the historical maximum rainfall represents the most extreme value that has been recorded in the dataset used. Overall, the spatial distributions of historical maximum rainfall are very uneven (Fig. 4). It is not surprising that the rainfall amounts over the southern part of China are larger than those over the northern part, over eastern China are larger than over western China, over the coastal areas are larger than over inland areas, over the southern coastal areas are larger than over the northern coastal areas, over the southern inland areas are larger than over the northern inland areas, and over the major plains and river valleys are larger than over the adjacent large plateaus and mountains. This has to do with the warm air and moisture supply, which is the richest from the south and from the ocean. Grade III historical maximum hourly, 3-, 6-, 12- and 24-h rainfall are most noticeable east and south of the black solid line in each panel of Fig. 4, which runs from southern Liaoning, through northern Hebei, Shanxi, Sichuan, and then to Yunnan Province. The areas with heavier historical maximum rainfall in different accumulation periods are mainly located in the coastal areas of China, South China, the Yangtze River-Huaihe River Basins, the Huanghe River-Huaihe River Basins, the western Sichuan Basin, and the North China Plain.

    The above spatial distributions share some similarities with those of heavy rainfall and SDHR occurrence frequency (Zhang and Lin, 1985; Chen et al., 2013) over central and eastern China. For example, both South China and the Sichuan Basin (Regions and in Fig. 1b) exhibit heavier historical maximum rainfall, a higher mesoscale convective system (MCS) frequency (Zheng et al., 2008), a higher heavy rainfall frequency, and heavier average annual precipitation, than other regions of China.

    Figure 4.  Color-filled contour maps of historical maximum (a) hourly, (b) 3-, (c) 6-, (d) 12- and (e) 24-h rainfall over central and eastern China for 1981-2012 (units: mm), mapped to a 0.75$^\circ$ latitude-longitude grid. The dark blue and magenta colors correspond to Grade II and Grade III of extreme rainfall, respectively, while three lower thresholds are also plotted. The stations with Grade II and Grade III extreme rainfall are marked by light blue stars and yellow dots, respectively (see legends). The thick black line in each panel marks the western boundary of stations that recorded Grade III extreme rainfall events (hourly rainfall of $\ge 95$ mm, 3-h rainfall of $\ge 155$ mm, 6-h rainfall of $\ge 205$ mm, 12-h rainfall of $\ge 260$ mm, or 24-h rainfall of $\ge 305$ mm).

    Figure 5.  As in Fig. 4, but for estimated 50-yr rainfall using the GEV distribution: (a) hourly rainfall; (b) 3-h rainfall; (c) 6-h rainfall; (d) 12-h rainfall; (e) 24-h rainfall.

    However, the spatial distributions of the historical maximum rainfall differ from those of MCS frequency, heavy rainfall frequency, and average annual precipitation (Zhang and Lin, 1985; Zheng et al., 2008; Chen et al., 2013) over the region between 25°N and 40°N in China, which includes Hunan, Jiangxi, Zhejiang provinces, the Huanghe River-Huaihe River Basins, the Shandong Peninsula, and the North China Plain (Fig. 1). For instance, Hunan, Jiangxi, and Zhejiang provinces exhibit higher MCS, heavy rainfall and SDHR frequencies, and heavier average annual precipitation (Zhang and Lin, 1985; Zheng et al., 2008; Chen et al., 2013), but they still have less intense historical maximum rainfall than the regions of the Huanghe River-Huaihe River Basins, the Shandong Peninsula, and the North China Plain.

    West of the thick black line in Fig. 4, most of the historical maximum hourly, 3-, 6-, 12- and 24-h rainfall amounts attain only Grade I (below 75 mm, 125 mm, 160 mm, 195 mm and 230 mm, respectively) according to our classification, although most of them are greater than 20 mm, the threshold of SDHR for hourly rainfall. Conversely, east of the line, there are several areas featuring historical maximum hourly, 3-, 6-, 12- and 24-h rainfall of no less than 95 mm, 155 mm, 205 mm, 260 mm and 305 mm (Grade III), respectively.

    Figure 4 shows that the stations with Grade II historical maximum hourly, 3-, 6-, 12- and 24-h rainfall are mostly concentrated over South China, the western Sichuan Basin, eastern Hubei Province, the coastal areas of Zhejiang and Fujian provinces, the Yangtze River-Huaihe River Basins (excluding the central Anhui Province), the Huanghe River-Huaihe River Basins, the North China Plain, and southern Liaoning Province. However, over Guizhou, Hunan, western Jiangxi, inland Zhejiang, and inland Fujian provinces, which are located between 25°N and 30°N, the stations with Grade II rainfall are sparse and scattered, although there are higher occurrence frequencies of SDHR events (Chen et al., 2013) and MCSs (Zheng et al., 2008).

    Furthermore, the densely distributed stations with Grade III historical maximum hourly, 3-, 6-, 12- and 24-h rainfall (Fig. 4) are located mainly over South China, the western Sichuan Basin, eastern Hubei Province, the coastal area of Zhejiang Province, the northern coastal area of Fujian Province, eastern Henan Province, the Huanghe River-Huaihe River Basins, the North China Plain, and parts of southern Liaoning Province. Whereas, over the area north of 30°N in China, the number of stations with Grade III historical maximum 12- or 24-h rainfall (≥ 260 mm or ≥ 305 mm) is significantly fewer than that with Grade III historical maximum hourly and 3-h rainfall (≥ 95 mm and ≥ 155 mm). However, over eastern and northern Jiangxi Province, there are more stations with Grade III 24-h rainfall than those with Grade III hourly, 3-, 6- and 12-h rainfall.

    For various regions labeled in Fig. 1b, the heaviest rainfall for a region is obtained as the maximum that has ever been recorded at any one station within the region. The heaviest hourly rainfall is above 140 mm over South China, and it is 135 mm and close to 140 mm over eastern Hubei Province, the Huanghe River-Huaihe River Basins, and southern North China. Therefore, there are only slight regional differences in historical maximum hourly rainfall amounts among southern North China, the Huanghe River-Huaihe River Basins, and South China. However, for the historical maximum 24-h rainfall, the heaviest rainfall is above 550 mm over South China, while over southern North China and the Huanghe River-Huaihe River Basins, it is only about 420 mm. Clearly, there are larger differences, in both relative and absolute values, among 24-h extreme rainfall across China than hourly extreme rainfall. This suggests that heavy rainfall events in southern China are longer-lasting than those in northern China.

    Apart from the spatial distributions of historical maximum rainfall, we are also interested in how the extreme rainfall is distributed in amounts. The most popular amounts of the historical maximum rainfall among the 1919 stations are determined by applying different bin-widths to different accumulation periods. Using 20 mm as the bin-width, stations with hourly extreme rainfall between 60 mm and 80 mm are most common, accounting for 40.8% of total stations. Using 50 mm as an interval, stations with 3-h extreme rainfall between 100 mm and 150 mm are most common (42.7%); stations with rainfall between 100 mm and 150 mm are most common for 6-h extreme rainfall (36.5%); and for 12-h extreme rainfall, 150-200 mm amounts are most common (27.7%). Using 100 mm as an interval of 24-h extreme rainfall, amounts between 100 mm and 200 mm are most common, accounting for 44.8% of total stations.

4. Spatial distributions of 50-yr return values
  • This section describes the spatial distributions of 50-yr return values for hourly, 3-, 6-, 12- and 24-h rainfall obtained from the fitted GEV distribution based on the 1981-2012 data. These spatial distributions are compared to those of the historical maximum rainfall in different accumulation periods.

    As given in Table 2, the numbers of stations with Grade II and Grade III 50-yr rainfall for different accumulation periods are less than those with their corresponding historical maximum rainfall. Nevertheless, the spatial patterns of the 50-yr return values for hourly, 3-, 6-, 12- and 24-hr rainfall are generally similar to those of the corresponding historical maximum rainfall. Similar to Fig. 4, over the areas east and south of the thick black line in each panel of Fig. 5, the estimated 50-yr rainfall return values at some stations can attain Grade III. Figures 5d and 5e clearly show that there are far fewer stations with Grade III 50-yr return values for 12- or 24-h rainfall than those for hourly and 3-h rainfall over the area north of 30°N in China.

    Similar to how we obtain the most common rainfall amounts in the historical maxima, we also examine the 50-yr return values. With a 20 mm bin-width for hourly rainfall, stations with 60-80 mm rainfall are most common, accounting for 42.4% of all stations. Using 50 mm as an interval for 3-, 6- and 12-h rainfall, stations with rainfall amounts of 100-150 mm, 100-150 mm and 150-200 mm are most common (44.2%, 35.7% and 27.4%), respectively. Using 100 mm as the interval, stations with 100-200 mm 24-h rainfall are most common, amounting to 44.1% of all stations. These statistics are all comparable to those of corresponding historical maximum rainfall.

    Similarly, for various regions labeled in Fig. 1b, the heaviest 50-yr return value for hourly rainfall is about 150 mm over South China. Over the Huanghe River-Huaihe River Basins, and southern North China, the heaviest 50-yr hourly rainfall is about 140 mm. Therefore, there is also only a slight difference in the 50-yr hourly rainfall amounts across these regions. However, for the 50-yr 24-h rainfall, the heaviest rainfall can be above 500 mm over South China; yet, it is only above 400 mm over the Huanghe River-Huaihe River Basins, and less than 400 mm over southern North China. These results also indicate that the absolute and relative differences in the 50-yr 24-h rainfall between South China and the regions of the Huanghe River-Huaihe River Basins and southern North China is larger than that in 50-yr hourly rainfall.

    Rainfall is the product of rainfall rate and duration; however, rainfall is also a complex nonlinear physical process, during which rainfall rates are usually non-uniform. Therefore, for any given site, the extreme cumulative rainfall amount in the accumulation period longer than 1 hour almost never equals the extreme hourly rainfall amount multiplied by the number of hours, and its average hourly rainfall intensity is usually less than the extreme hourly rainfall amount. As stated earlier, the regional heaviest historical maximum and the 50-yr hourly rainfall return value over the Huanghe River-Huaihe River Basins are close to those over South China; but if we consider the 50-yr rainfall return value in the accumulation periods that are greater than 3 hours, then the differences between these two regions significantly increase as the accumulation period increases. This is because, in South China, extreme rainfall tends to last longer (Chen et al., 2013; Li et al., 2013b). Overall, the results from the historical maximum rainfall and the estimated 50-yr rainfall are consistent.

    Figure 6.  Regional classification based on historical maximum and 50-yr rainfall amounts: (a) hourly rainfall; (b) 3-h rainfall; (c) 6-h rainfall; (d) 12-h rainfall; (e) 24-h rainfall (units: mm). White areas indicate no rainfall observation, and red areas indicate historical maximum rainfall reaching Grade III and 50-yr rainfall under Grade III.

5. Regional classification and differences in extreme rainfall
  • The similarity between the spatial distribution of the historical maximum rainfall and that of the estimated 50-yr rainfall suggests the results obtained in this paper are reliable. In this section, we further examine the spatial distributions of extreme rainfall of the three grades for different accumulation periods.

    Based on the historical maximum and 50-yr rainfall amounts over each 0.75°× 0.75° grid cell, we present a regional classification in Fig. 6. The main characteristics of the classified regions are summarized below:

    (1) The extreme rainfall reaching Grade II and Grade III is mainly observed east and south of the black lines in Figs. 4 and 5, which runs from southern Northeast China through Shanxi Province, then around the western edge of the Sichuan Basin towards the eastern slope of the Yunnan-Guizhou Plateau, more or less following the terrain elevation contour. However, Grade II is not reached over nearly half of the region between 25°N and 30°N for extreme 3-, 6- and 12-h rainfall especially.

    (2) Over Yunnan Province, eastern Inner Mongolia, and northern and central Northeast China, there are still a number of cells with Grade II extreme hourly rainfall (no less than 75 mm), but there are fewer cells with Grade II extreme 3-, 6-, 12- and 24-h rainfall. This shows that, over these areas, even if an SDHR event occurs and reaches Grade II extreme hourly rainfall, because of the shorter lifespan of convective systems producing the rainfall, the cumulative rainfall amounts in longer accumulation periods are less likely to attain Grade II.

    (3) For different accumulation periods, the spatial distributions of Grade III extreme rainfall are somewhat similar to each other. The similarity is greater among extreme hourly, 3- and 6-h rainfall, and less so for 12-h and 24-h rainfall.

    (4) The spatial distributions of Grade III extreme rainfall possess the following characteristics: they are situated over the lower latitudes (e.g., South China), along the southern and eastern coastlines, in the large Huanghe River-Yangtze River Basins, and over the lower-elevation side of the border region between plains or basins and plateaus or mountains (e.g., the west side of the Sichuan Basin, and the west side of the North China Plain).

    (5) Both South China and the Sichuan Basin exhibit not only heavier extreme rainfall, but also higher SDHR frequencies (Chen et al., 2013) and more heavy-rainfall days (Zhang and Lin, 1985).

    (6) Between 25°N and 30°N in China, there are fewer cells with Grade III extreme rainfall for different accumulation periods than in the regions of South China, the Yangtze River-Huaihe River Basin, and the Huanghe River-Huaihe River Basins. However, there are more cells with Grade III extreme 24-h rainfall than with hourly and 3-h rainfall (Fig. 4e, Fig. 5e and Fig. 6e) over some parts of this region, such as southern Anhui Province, eastern Jiangxi Province, and northwestern Hunan Province. This indicates that, although these regions do not exhibit Grade III extreme hourly rainfall, they can suffer more often from Grade III extreme 24-h rainfall. This phenomenon may be related to their terrain distributions or tropical weather systems, such as tropical cyclones, which affect these areas and cause long-duration rainfall.

    Rainfall rates in tropical systems are generally high, because they are usually associated with deep moist and organized convection (Davis, 2001). Extreme rainfall over South China is often associated with tropical systems that affect this region. Low-level southwesterly jets, land-sea breezes (Zheng et al., 2008; Zheng and Chen, 2013; Chen et al., 2015), and differential friction effects between the sea and land (Chen et al., 2014), have been found to provide additional local forcing and triggers for long-duration convection and precipitation near the coast. The extreme rainfall over the coastal areas of Zhejiang and Fujian provinces may be related to the frequent influence of tropical cyclones in these areas (Zheng et al., 2014), as well as land-sea breezes and differential friction effects present along the coast (Chen et al., 2014). The cause for the extreme rainfall over the Yangtze River-Huaihe River Basins, and the Huanghe River-Huaihe River Basins, appears to be due to the fact that these areas are situated at the edge of the summer monsoon and the subtropical high in summer, such that these regions experience long-duration Mei-yu rainfall. From the perspective of convective systems, the regions belong to the active MαCS (Meso-α-scale Convective System) and MβCS (Meso-β-scale Convective System) areas (Ma et al., 1997; Zheng et al., 2008), which will also have direct impacts. The extreme rainfall over the Sichuan Basin and the North China Plain is likely related to the northward migrating summer monsoon, which regularly influences these regions (Chen et al., 1991), as well as the impact of regional terrain. The heavier extreme rainfall for accumulation periods greater than 6 hours may be associated with nocturnal occurrences of heavy rainfall and SDHR over South China, the Sichuan Basin, the Yangtze River-Huaihe River Basins, and the Huanghe River-Huaihe River Basins (Chen et al., 2013); and nocturnal rainfall is often associated with MCSs that last longer.

    Our study does not try to document climate variability or seasonal cycles in the extreme rainfall, but these aspects could be potential topics for future research. There have been some studies (e.g., Zhai et al., 1999, 2005; Dong et al., 2011; Wang and Yan, 2011; Zhang and Zhai, 2011) on the climate variability of extreme daily or hourly rainfall over different regions of China, but not over the country as a whole. As the spatial distribution of rainfall in China is determined primarily by the advance and retreat of the summer monsoon (Tao, 1980; Ding and Zhang, 2009), heavy rainfall and SDHR events in China occur most frequently during the summer (June, July and August). The second highest heavy rainfall and SDHR frequency is in April and May, but their frequency then drops substantially in September (Tao, 1980; Ding and Zhang, 2009; Chen et al., 2013). For various regions, heavy rainfall and SDHR events in South China occur mainly in April, May, June, August and September; those in the middle and lower reaches of the Yangtze River appear mainly in June, July and August; and those over North China and Northeast China occur mainly in July and August. Therefore, we can speculate that extreme rainfall events in China occur mainly in summer, although their seasonal cycles may vary from region to region due to the influence of the summer monsoon. For example, historically, several extremely heavy rainfall events have occurred in summer, such as those of August 1963 in North China, August 1975 in Henan Province, August 1996 in North China, and July 2012 and 2016 in Beijing and Hebei Province, all of which caused heavy losses of life and serious damage to property.

  • To highlight the differences in the spatial distributions of extreme rainfall between the south and the north in China, the 30°N parallel is selected (light blue dashed line in Fig. 1) to divide China into northern and southern regions. Based on the historical maximum rainfall and the 50-yr return values, Fig. 7a compares Grade III extreme rainfall with different accumulation periods between these two regions. Figure 7a shows that the percentage of total stations with Grade III extreme rainfall south of 30°N increases significantly as the accumulation period increases, with the percentage increasing from about 49% to about 69% for the historical maximum rainfall, and from about 50% to about 72% for the 50-yr rainfall. In contrast, the percentages over the area north of 30°N significantly decrease as the accumulation period increases, from about 51% to about 31% for the historical maximum rainfall, and from about 50% to about 28% for the 50-yr rainfall.

    Figure 7.  Comparison of the percentages of the total stations with Grade III extreme rainfall over (a) south and north of 30$^\circ$N in China, and (b) the Beijing-Tianjin-Hebei area and Guangdong Province. Vertical axis: percentage (%); horizontal axis: accumulation period (h).

    Similarly, Fig. 7b shows the difference in the percentages of total stations with Grade III extreme rainfall between Guangdong Province and the Beijing-Tianjin-Hebei area (indicated by the light blue solid lines in Fig. 1). Although there are some differences between Fig. 7a and b, the trends along with the accumulation period in Fig. 7b for the two local regions are similar to those for the south and north of China shown in Fig. 7a. Again, these results show that long-duration rainfall events are much more prevalent in the southern part of China, for which the occurrence frequencies of hourly extremely rainfall are very similar. The northern and inner parts of China have climates of a more continental nature, which are capable of producing intense short-duration convection, but the lack of sustained moisture supply from the ocean tends to limit the duration of heavy rainfall.

6. Summary and conclusions
  • Based on the hourly rainfall data from 1919 national-level meteorological stations in China during the period 1981-2012, we first derive the 3-, 6-, 12- and 24-h running cumulative rainfall, and then estimate the GEV distributions using the hourly and different running cumulative rainfall series. Based on our analysis of these data, we propose a new classification for different accumulation periods to divide the extreme rainfall into three grades. The thresholds separating the three grades correspond to roughly the 70th and 90th percentiles of extreme rainfall among the stations. We analyze, compare and classify the spatial distributions of the historical maximum hourly, 3-, 6-, 12- and 24-h rainfall, and their corresponding estimated 50-yr return values over China.

    The coastal areas of southern and eastern China, the large river basins, the western Sichuan Basin, and the North China Plain, all exhibit heavier extreme rainfall for different accumulation periods. Furthermore, both South China and the western Sichuan Basin exhibit not only heavier extreme rainfall, but also higher occurrence frequencies of SDHR and more heavy-rainfall days. In general, the spatial distributions of Grade III extreme hourly, 3-, 6-, 12- and 24-h rainfall are similar, especially for hourly, 3- and 6-h rainfall. The distributions of 12- and 24-h rainfall are more different.

    The number of stations with Grade III extreme hourly rainfall over the area south of 30°N is nearly as many as that over the area north of 30°N in China. However, when considering the stations with Grade III extreme 6-, 12- and 24-h rainfall, the differences in the station numbers between these two areas increases significantly as the accumulation period becomes longer. This characteristic reflects the fact that the extreme hourly rainfall amounts of these two areas are almost equal, but the extreme rainfall events over the former area last longer than those over the latter area due to the effects of richer moisture, the low-level southwesterly jet, tropical cyclones, and so on.

    The spatial distributions of the 50-yr rainfall using the fitted GEV of static parameters are presented in this paper. They differ somewhat from those of the historical maximum rainfall over certain areas. The differences may be related to the fact that the fitted GEV parameters are static and thus cannot fully reflect climate variabilities in extreme rainfall. In addition, the detailed temporal characteristics of extreme rainfall for different accumulation periods, including long-term trends, seasonal cycles, and diurnal variations, are not presented in this study; they can be topics for further research. In future studies, an alternative method, the Generalized Pareto distribution, can be explored to investigate long-term trends or climate variabilities in extreme rainfall by defining non-stationary thresholds. Finally, although many studies have investigated the development mechanisms of heavy rainstorms in China (e.g., Tao, 1980; Ding and Zhang, 2009; Tao and Zheng, 2013; Zhao et al., 2013; Luo et al., 2014), there remain needs for further research on the weather patterns, the environmental characteristics, and the mesoscale and small-scale mechanisms, of extreme rainfall in China.

    Our current study provides only a climatological background for such specific research. Our climatological study, including the classification standards set based on long-term historical data for accumulation periods ranging from hourly through 24-h, also has the potential to help policy-makers draw up region-specific regulations and standards, including those on buildings, roads, reservoirs, dams, and other infrastructure types. The standards could also be adopted by the central and regional meteorological services for operational use.

Reference

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return