PIAHSProceedings of the International Association of Hydrological SciencesPIAHSProc. IAHS2199-899XCopernicus GmbHGöttingen, Germany10.5194/piahs-369-97-2015Climatological features and trends of extreme precipitation during 1979–2012 in Beijing, ChinaXuZ. X.zongxuexu@vip.sina.comChuQ.Key Laboratory of Water and Sediment Sciences, Ministry of Education, College of Water Sciences, Beijing Normal University, Beijing, 100875, ChinaJoint Center for Global Change Studies, Beijing, 100875, ChinaZ. X. Xu (zongxuexu@vip.sina.com)11June20153693699710215April201515April2015This work is licensed under a Creative Commons Attribution 3.0 Unported License. To view a copy of this license, visit http://creativecommons.org/licenses/by/3.0/This article is available from https://piahs.copernicus.org/articles/369/97/2015/piahs-369-97-2015.htmlThe full text article is available as a PDF file from https://piahs.copernicus.org/articles/369/97/2015/piahs-369-97-2015.pdf
In this study, three kinds of hourly precipitation series with the spatial
resolution of 0.1∘ are used to analyze the climatological features and
trends of extreme precipitation during the period of 1979–2012 in Beijing,
China. The results show that: (1) the spatial distribution of median annual
precipitation, with a range from 500 to 825 mm, is similar to that of
local topography, which increases from the northwest to the southeast.
Taking the urban area as a centre, the inter-annual precipitation in the
Beijing area displays an outward decreasing tendency at the maximum rate of
125 mm per decade (125 mm × 10 a-1); (2) extreme precipitation amount,
which accounts for 40–48 % of total precipitation amount, has a
similar spatial distribution to average annual precipitation; (3) the
spatial distribution of extreme precipitation days and threshold estimated
as the upper 95 percentile are significantly different from that of extreme
precipitation, with maximum values concentrated on the urban area and the
eastern mountain area, and minimum values in northwest; (4) extreme
precipitation days (Ex_pd95) show an opposite distribution to
extreme precipitation threshold (Ex_pv95), indicating that
areas with greater precipitation threshold may has less precipitation days,
and vice versa; (5) an apparent spatiotemporal decreasing tendency is
detected in extreme precipitation amount. The downward tendencies are also
found in extreme precipitation threshold. Unlike Ex_pv95, in
most of the study area, Ex_pd95 is virtually unchanged; (6) downward
trends of extreme precipitation is slightly smaller than that of
annual precipitation, and the reducing amplitude of north-eastern areas are
much higher than the areas in the southwest.
Introduction
Extreme weather and climatic events have drawn broader concerns during past
years, particularly on the regional and local scales. It has been recognized
that changes in extreme events are more likely to cause damages for human
lives and their properties than gradual changes (Bonsal et al., 2001). In order to
obtain a better understanding of potential risks for decision making in
terms of societal adaptation to future climate change, the detection and
attribution of past changes become increasingly significant (Madsen et al., 2014).
Extreme precipitation, which is regarded as the main factor contributing to
water security, reflects the homogeneity of temporal and spatial
distribution of precipitation. Extreme precipitation indices, such as
extreme precipitation threshold, extreme precipitation days, extreme
precipitation amount, are widely used to assess the variations of extremes
in several studies (You et al., 2014). As an identical classification of
extremes is not comparable among the areas with greatly varying climates,
empirical ranking methods are recommended to determine the extreme threshold
at different percentiles. Therefore, the formula introduced by Beard (1943)
has come into wide use because it is more suitable for studies on the
changing climate extremes (Folland and Anderson, 2002).
Map of Study Area. (The square grid represents the spatial
resolution of the assimilated data, which covers 205 grids in the whole
area. The areas in the northern area surrounded by the blue line stand for
the urban areas. Elevation is also shown in this figure.)
There have been plenty of studies on the analysis for variations and trends
of extreme precipitation over global or regional scales. With respect to
extreme precipitation, most of studies are based on in situ observations or
large scaled gridded data downscaled from climate models (Koteswara et al.,
2014; Li et al., 2015). In the local area, both the length of the data
series and the spatial representative are limited due to the finite number
of long-term observational stations. An urgent demand occurs for high
resolution datasets for extremes studies especially in the rapidly urbanized
regions with insufficient data. Under this circumstance, data assimilation
technique undergoes a rapid development. It supplies an alternative way to
study the impact of climate changes in these kinds of areas.
Beijing, the capital of China, has experienced tremendous changes due to the
accelerated development of socio-economics and the rapid expansion of
population during the past fifty years. However, negative consequences, such
as sever water scarcity, serious floods and urban water-logging, are all
along with the rapid growth of economics and urbanization. Therefore,
accurate quantifications of recent changes in extreme precipitation can be
benefit to clarify the mechanism of climate change and enhance
decision-making for sustainable development of water resources and
environment protection in Beijing.
Due to the fact that surface precipitation changes exhibit obvious regional
characteristics, few temporal and spatial studies with higher resolution
data have been made so far in Beijing. The main objective of this study is
to: (1) analyze the tempo-spatial variability of the annual extreme
precipitation based on assimilated datasets with high resolution in Beijing;
(2) qualitatively indicate the local-scale effects on extreme precipitation,
such as topography, urbanization and local climate. Jenkinson's ranking
formula and Theil–Sen Estimator are employed in this study. The findings
will probably contribute to reduce uncertainties on floods and droughts
induced by the variations of extreme precipitation.
Study area description
Beijing, the capital of the People's Republic of China, is composed of
16 districts, with most of the urban areas lying in the western area. It is
located at 39∘26-–41∘03′ N and
115∘25′–117∘30′ E with an area of 16 410.54 km2
(Beijing Statistics Bureau, 2010), 68 % of which is
mountain areas. It lies on the northwestern border of the North China Plain,
surrounded by Taihang Mountain on the west and Yan Mountain on the north and
northeast. Terrain tilts from northwest to southeast over the whole area.
Elevation varies significantly (60–2303 m) in mountian areas; while it
changes sligtly in Plain areas, with values from 10 to 60 m, as shown in Fig. 1.
The city is in the semi-humid warm continental monsoon climate zone. This
place experiences four distinct seasons, with a cold and dry winter
accompanied by northward wind blowing from high-latitude area, while a hot
and wet summer because of the east-southeast toward airflow from the
southern Pacific Ocean and the Indian Ocean. Due to the interaction of these
cold and hot airflows, the precipitation is maily concentrated in summer,
which accounts for 60–80 % of total precipitation amount.
Data and method description
In this study, a high resolution assimilated dataset (1979–2012) was used to
analyze the variation of extreme precipitation. For each grid, Jenkinson's
ranking formula was employed to estimate the 95th percentiles of daily
precipitation distribution. The temporal and spatial characteristics and
trends of surface annual extreme precipitation indices were then analyzed by
Theil–Sen slope estimator method. A brief introduction on the dataset and
methods are as follows.
Data description
In this study, three hourly assimilated datasets (1979–2012) with
0.1∘× 0.1∘ of spatial resolution were used to
analyze the variations of extreme precipitation. These datasets used a
global dataset produced by the Global Land Data Assimilation System (Rodell
et al., 2004) as the background field when station observaitons are
interpolated to grid points. Detailed data fusion technique may be found in
He and Yang (2011). Simple quality control was also carried out to ensure
that the time series is physically resonable by eliminating the data
exceeding 3 standard deviations.
Method descriptionJenkinson empirical ranking formula
According to Bonsal et al. (2001), daily precipitation for each year should
be firstly ranked in ascending order X1, X2 … Xm … Xn.
The cumulative probability P that a random value is
less than or equal to the rank of that value Xm is then estimated by:
P=(m-0.31)/(n+0.38).
This formula was proposed by Beard (1943) and presented in detail by
Jenkinson (1977). It is proved by Folland and Anderson (2002) that this
method performed as well as other empirical ranking formulas. But unlike
other methods, Jenkinson's ranking method has no assumption on underlying
distributions. That makes it more suitable to investigate the changing
climate extremes, since knowledge of distribution form can rarely be
obtained for those extremes.
Theil–Sen estimator
The Theil–Sen estimator is an unbiased estimator of the true slope in simple
linear regression. For many distributions of the response error, this
estimator has high asymptotic efficiency relative to least-squares.
Estimators with low efficiency require more independent observations to
attain the same sample variance of efficient unbiased estimators. Besides,
it is more robust because it is much less sensitive to outliers: it can
tolerate arbitrary corruption of up to 29.3 % of the input data without
degradation of the accuracy.
Results analysis and discussionClimatological features of extreme precipitation
Figure 2 shows the spatial distribution of extreme precipitation indices in
Beijing. The spatial distribution of median annual precipitation (PTV), with
a range from 500 to 825 mm, is opposite to that of local topography,
which increases from the northwest to the southeast. Results by using
Principal Component Analysis (PCA) method indicate that the local climate
and topography are two main factors influencing the spatial distributions of precipitation.
Extreme precipitation threshold (Ex_pv95) calculated as the
upper 95 percentile (15.0–32.5 mm day-1) is slightly smaller than that
estimated by You et al. (2014) in Beijing. It is likely due to the
elimination of extremes by using standard deviation method. Extreme
precipitation days (Ex_pd95) stand for the total time when
daily precipitation is greater than Ex_pv95 in each year. As
it can be seen from Fig. 1, Ex_pv95 presents an apparent
opposite distribution to Ex_pd95, which means that areas with
greater precipitation threshold may have less precipitation days and vice
versa. The maximum Ex_pv95 appears at most of urban area and
some districts in the northeast plain area, while with the least values in
the north-western areas. Ex_pd95 in the piedmont areas is
nearly 7 days, which is the largest value in the whole study area. This is
not only related to the interaction of the warm and cold airflows influenced
by the local monsoon climate, but also due to significant uplift effect of
terrain, resulting in systematic intensification of precipitation process in
these areas.
Extreme precipitation amount (Ex_ptv95) is defined as the
total amount of daily precipitation which exceeds Ex_pv95.
Figure 1 shows that extreme precipitation amount has a parallel spatial
distribution to average annual precipitation, with maximum values
concentrated on urban area and the eastern mountain area, and minimum values
in the north-western area. It accounts for 40–48 % of total
precipitation amount within only 5 to 7 days, which indirectly suggests the
inhomogeneous temporal characteristics of precipitation. It is worthwhile to
notice that Ex_ptv95 of the urban areas occupies the largest
proportion of total precipitation amount. Moreover, the total precipitation
also has the maximum value of 825 mm, displaying a strong feature of urban
wet island effect. The reason for this is partly owing to the effect of
urbanisation in terms of urban heat island, the obstacles of high-rise
buildings and the increase of condensation nucleus.
Climatological annual-median values of extreme precipitation based
on 3 h gridded data during 1979–2012. (From the left to the right and from
the up to the bottom, the figures referred to (a) median annual
precipitation (PTV); (b) extreme precipitation threshold (Ex_pv95);
(c) extreme precipitation days (Ex_pd95); (d) extreme
precipitation amount (Ex_ptv95); (e) extreme precipitation
proportion (Ex_maxper95); (f) extreme precipitation intensity
(Ex_pi95) calculated at upper 95th percentile by Jenkinson's formula, respectively).
Trends of extreme precipitation based on 3 h gridded data during
1979–2012. (The variables are same as Fig. 2.)
Extreme precipitation intensity (Ex_pi95) is an important
measurement of extreme precipitation, since larger Ex_pi95
implies higher risk caused by extreme precipitation. It is clear that the
spatial characteristic of Ex_pi95 is similar to that of
Ex_pv95, which suggests that areas with larger
Ex_pv95 may experience heavy storm. The maximum values appear
at the urban area and some north-eastern areas, which is just under 70 mm day-1.
This means that these regions should be paid more attention for
the threat of extreme precipitation.
Trends of extreme precipitation
A significant downward trend can be found in both PTV and Ex_ptv95
in Beijing, with sharply decreasing rate (90–110 mm × 10 a-1) in
the urban areas (see Fig. 3). Chu et al. (2015) found that the dramatic fall
somehow related to the rising temperature. Although the total water vapour
amount increases because of the rise of evaporation, the capacity of
atmosphere to hold water presents faster upward trend. An apparent
spatial-temporal decreasing trend is detected in Ex_ptv95
values. The downward tendencies are also found in extreme precipitation
threshold and days, which are more pronounced in Miyun and Mentougou districts.
According to the formula given by Jenkinson, the decrease of
Ex_pv95 indicates reducing daily precipitation intensity,
while the increase of Ex_pv95 represents a rise of daily
precipitation. As it can be seen from Fig. 3, the northern and
north-eastern districts have experienced an upward tendency in daily
intensity, while the regions in north and east fell with the value of
3.0 mm day-1× 10 a-1. Unlike Ex_pv95, in most of the study
area, Ex_pd95 is virtually unchanged, the decreasing
amplitude of which is less than 0.55 day × 10 a-1. These slight changes are
detected in the areas where Ex_pv95 increased a lot, which
leads to significant decrease of Ex_ptv95.
Compared the trends of PTV with that of Ex_ptv95, it is clear
that the downward rate of PTV is nearly 30 mm × 10 a-1 greater than the
rate of Ex_ptv95. On one hand, it suggests that
Ex_ptv95 contributes the largest part in PTV. On the other
hand, the proportion of Ex_ptv95 (Ex_maxper95)
varied slightly during this period, indicating that the risk of extreme
precipitation was still high, especially in the areas with the increase of Ex_pv95.
Conclusions
The spatial distribution of median annual precipitation increases from
the northwest to the southeast. Results obtained by using Principal
Component Analysis (PCA) method indicate that the local climate and
topography are two main factors influencing the spatial distributions of
precipitation in Beijing.
Ex_pv95 presents an apparent opposite distribution to
Ex_pd95, which means that areas with greater precipitation
threshold may have shorter precipitation days. The maximum Ex_pv95
appears at most of urban areas and some districts in the northeast
plain area. The piedmont areas have the largest Ex_pd95
because of the effect of local monsoon climate and significant uplift of terrain.
Ex_ptv95 has a similar spatial distribution to average
annual precipitation, with maximum values concentrated on the urban area and
the eastern mountain area. It accounts for 40–48 % of total
precipitation amount within only 5 to 7 days, which indirectly suggests the
inhomogeneous temporal characteristics of precipitation.
The spatial characteristics of Ex_pi95 are similar to
that of Ex_pv95, with the maximum values appearing at the
urban area and some north-eastern areas. These areas may experience heavy
storm since larger Ex_pi95 implies higher risk caused by
extreme precipitation.
Significant downward trends are detected in both PTV and
Ex_ptv95, with sharply decreasing rate (90–110 mm × 10 a-1) in urban areas. This dramatic decrease is partly because the
rise of air temperature, which results in higher rising rate of the capacity
of atmosphere to hold water than the total water vapour amount.
The northern and north-eastern districts have experienced an upward
tendency in daily intensity, while the regions in north and east fell with
the value of 3.0 mm day-1× 10 a-1. Unlike Ex_pv95, in
most of the study area, Ex_pd95 is virtually unchanged.
Ex_ptv95 contributed the largest part in the decrease of
PTV. However, the proportion of Ex_ptv95 (Ex_maxper95) varied slightly during this period, indicating that the risk of
extreme precipitation was still high, especially in the areas with the
increase of Ex_pv95.
Acknowledgements
This study was supported by the research project from Beijing Natural
Science Foundation (No. 8141003). The authors thank the Data Assimilation
and Modelling Centre for Tibetan Multi-spheres (DAM) for providing high
resolution dataset of surface precipitation.
References
Beard, L. R.: Statistical analysis in hydrology, T. Am. Soc. Civ. Eng., 108, 1110–1121, 1943.
Beijing Statistics Bureau: Beijing Statistical Yearbook 2010, China Statistics Press, Beijing, 2010.
Bonsal, B. R., Zhang, X., Vincent, L. A., and Hogg, W. D.: Characteristics of
daily and extreme temperatures over Canada, J. Climate, 14, 1959–1976, 2001.
Chu, Q., Xu, Z. X., Peng, D. Z., Yang, X. J., and Yang, G.: Trends of surface
humidity and temperature during 1951–2012 in Beijing, China, in: Proceedings
of the 3rd Remote Sensing and Hydrology Symposium, August 2014, Guangzhou, China, in press, 2015.
Folland, C. and Anderson, C.: Estimating changing extremes using empirical
ranking methods, J. Climate, 15, 2954–2959, 2002.He, J. and Yang, K.: China Meteorological Forcing Dataset, Cold and Arid
Regions Science Data Center, Lanzhou, 10.3972/westdc.002.2014.db, 2011.
Jenkinson, A. F.: The analysis of meteorological and other geophysical
extremes, Synoptic Climatology Branch, Meteorological Office, Bracknell,
Berkshire, UK, 44 pp., 1977.
Koteswara, R. K., Patawardhan, S. K., Kulkarni, A., Kamala, K., Sabade, S. S.,
and Kumar, K. K.: Projected changes in mean and extreme precipitation
indices over India using PRECIS, Global Planet. Change, 113, 77–90, 2014.
Li, J. F., Zhang, Q., Chen, Y. Q., and Singh, V. P.: Future joint probability
behaviors of precipitation extremes across China: Spatiotemporal patterns
and implications for flood and drought hazards, Global Planet. Change, 124, 107–122, 2015.
Madsen, H., Lawrence, D., Lang, M., Martinkova, M., and Kjeldsen, T. R.:
Review of trend analysis and climate change projections of extreme
precipitation and floods in Europe, J. Hydrol., 519, 3634–3650, 2014.
Rodell, M., Houser, P. R., Jambor, U., GottschalcK, J., Mitchell, K., Meng, C. J.,
Arsenault, K., Cosgrove, B., Radakovich, J., Entin, K., Walker, J. P., Lohmann, D.,
and Toll, D.: The Global Land Data Assimilation System, Am. Meteorol. Soc., 3, 381–394, 2004.
You, H. L., Liu, W. D., and Ren, G. Y.: Variation characteristics of
precipitation extremes in Beijing during 1981–2010, Clim. Environ. Res., 19, 69–77, 2014.