Abstract

Spatial and temporal characteristics of precipitation trends in the Zhujiang River basin, South China, are analyzed. Nonparametric trend tests are applied to daily precipitation data from 192 weather stations between 1961 and 2007 for the following indices: annual, monthly, and daily precipitation; annual and monthly number of rain days and precipitation intensity; annual and monthly maximum precipitation; 5-day maximum precipitation, number of rainstorms with >50 mm day−1, and peaks over thresholds (90th, 95th, and 99th percentile).

The results show that few stations experienced trends in the precipitation indices on an annual basis. On a monthly basis, significant positive and negative trends above the 90% confidence level appear in all months except December. Trends in the indices of monthly precipitation, rain intensity, rain days, and monthly maximum precipitation show very similar characteristics. They experience the most distinct negative (positive) trends in October (January). A change of the mean wind direction by 50° from east-southeast to east-northeast explains the downward trend in precipitation in October. Dry October months (months with low precipitation indices) can be observed when the mean wind direction is east-northeast (arid) instead of the prevailing mean wind direction, east-southeast (moist). The former is typical for the East Asian winter monsoon (EAWM). Nearly 90% of the driest October months can be explained by wind directions typical for the EAWM. The upward trend in precipitation indices in January cannot be explained by changes in large-scale circulation. The analysis of the precipitation indices delivers more detailed information on observed changes than other studies in the same area. This can be attributed to the higher station density, the quality of daily data, and the focus on monthly trends in the current study.

1. Introduction

According to China’s National Assessment Report on Climate Change (Ding et al. 2007), changes in annual precipitation have been observed in China for the last century. In the last decades, an increase in national average precipitation was detected, with increasing spring precipitation but slightly decreasing autumn precipitation. The Fourth Intergovernmental Panel on Climate Change (IPCC) Assessment Report (Trenberth et al. 2007) is in line with China’s National Assessment Report on Climate Change and indicates that the frequency of heavy precipitation events will very likely increase in China. Regional changes are, however, diverse—southeast China, for instance, experienced an increase in annual precipitation by around 60 to 130 mm during the last 50 years (Ding et al. 2007). The East Asian monsoon plays an important role in regional precipitation patterns. A strong (weak) winter monsoon with northerly winds leads to decreased (increased) winter precipitation over south China (Chen et al. 2000; Zhou and Wu 2010).

It is therefore important to understand regional changes in precipitation patterns and the causes of these changes. Several studies have analyzed precipitation records for the whole of China. Feng et al. (2007) have analyzed annual maximum precipitation time series (1-, 2-, 5-, and 10-day) for 651 stations in China from 1951 to 2000 and detected negative trends in extreme events in north China. Significant positive trends were observed at stations in the Yangtze River basin and northwestern China, where extreme events with a 50-yr return period in the 1950s became more frequent 25-yr events in the 1990s. More results for China using longer time series and analyzing linear trends in rain days with different intensities are provided by Fu et al. (2008). Liu et al. (2005) used 272 stations in China and examined change rates for eight regions and different seasons from 1960 to 2000 and found a nonsignificant increase of precipitation in southeast coastal China.

Qian and Lin (2005) comprehensively analyzed regional characteristics of daily precipitation indices at 494 stations in China (1961 to 2000) and showed interdecadal differences. In coastal southeast China, a negative decadal tendency in annual and summer precipitation was detected. Likewise, Wang and Zhou (2005) used linear regressions to analyze trends in annual and seasonal mean precipitation in China during 1961–2001. The results show decreasing extreme events in summer and a higher decreasing trend in autumn. Yang and Lau (2004) focused on spring and summer precipitation and noted positive trends in spring precipitation for a gridded dataset covering 1951–98, while Yao et al. (2008) focused on the time series 1978–2002 and concluded that for summer precipitation, the amount of total precipitation and light–moderate precipitation exhibited positive trends over southeast China.

Zhai et al. (1999) detected no significant trends in annual precipitation but a significant increase in above-normal mean intensity of precipitation in east China from 1951 to 1995. Zhai et al. (2005) applied trend techniques to a time series covering 1951–2000 and delivered the currently most comprehensive analysis of annual and extreme precipitation in China. Annual total, spring, and extreme precipitation significantly increased along the southeastern coastline of China, and winter precipitation increased in the south.

More and more studies on precipitation changes focus on a river-basin scale in order to identify results for hydrologically homogeneous areas as compared to administrative boundaries. Significant increasing trends in extreme precipitation have been observed for the Zhujiang River basin, also known as the Pearl River basin, in the last decades (Luo et al. 2008; Yang et al. 2010). These studies have mostly analyzed time series of annual and seasonal precipitation in the Zhujiang River basin to define precipitation extremes and their spatial pattern (Zhang et al. 2009ac). Only few of these studies examined the temporal and spatial variations in monthly precipitation events.

Yang et al. (2010) analyzed the regional frequency of annual precipitation extremes based on consecutive 1-, 3-, 5-, and 7-day averages for 42 stations in the Zhujiang River basin from 1960 to 2005. It was identified that major precipitation events in regions of low elevation in the lower (southeastern) part of the basin occur mainly in May, June, July, and August, whereas the main precipitation periods for the mountainous region upstream are June, July, and August.

Liu et al. (2009) observed variations of seasonal precipitation in the Zhujiang River basin at 64 stations (1961–2007) and showed a decrease in autumn precipitation but increases in spring, summer, winter, and annual precipitation. An enhancement of interannual variability of annual and winter precipitation and a weakening of autumn precipitation is projected using the general circulation model (GCM) ECHAM5-Max Planck Institute Ocean Model (MPI-OM; Roeckner et al. 2003).

Based on data from 63 stations for the period 1959–2003, Luo et al. (2008) observed a downward trend in precipitation over the Beijiang River basin (northeastern tributary of Zhujiang) in the early flood period (April–June), especially in May, and an upward trend in July and in the dry season (October–March). The common precipitation belt over the Beijiang River, which is caused by frontal precipitation when the front of warm air masses from the south meets cold air masses from the north, has shifted northward in recent years. This has caused positive precipitation trends in the north and negative trends in the south of the catchment.

Zhang et al. (2009b) calculated a decreasing precipitation concentration index (CI) for the southwestern and northeastern parts as well as for the West River and East River basin of the Zhujiang River basin using precipitation data from 42 stations from 1960 to 2005. The study also described a significant increasing precipitation CI after 1990 in the West River basin, in the lower North River basin, and in the upper Beipan River basin.

Based on daily precipitation data for 47 stations from 1951 to 2005, Zhang et al. (2009a) did not detect significant trends in annual, summer, or winter precipitation. However, the trends in precipitation intensity, variability of precipitation, and high-intensity rainfall events increased. Zhang et al. (2009c) identified trends toward dryness in the rainy season (April–September) and an increase of wet conditions in the dry season (December–February).

Previous studies on the Zhujiang River basin have detected seasonal changes in precipitation, but most research focused on summer precipitation. Little has been written about the characteristics and causes of precipitation extremes—for example, changes in precipitation intensity on a daily basis and their statistical trends. The station density used in existing studies for the whole of China is sufficient to describe large-scale changes, but insufficient on a regional scale. In addition, the causes of precipitation trends are not entirely understood. The objective of this paper is to analyze trends in and causes of precipitation extremes in the Zhujiang River basin at the highest possible spatiotemporal resolution allowed by available data.

2. Data and methodology

a. Study area

The Zhujiang (Pearl River) basin is located in South China and falls within the provinces of Guangdong, Guangxi, Guizhou, and Yunnan. The Zhujiang River basin is one of China’s largest river basins (Zhai et al. 2010) and drains an area of 579 000 km2 (including the Leizhou Peninsula region). It has a tropical and subtropical climate. The annual mean temperature varies from 14°C in the west to 22°C in the east, and the annual average precipitation is 800 mm in the west and more than 2000 mm at the coastline (average for the basin: 1500 mm). The months from June to August are influenced by the East Asian summer monsoon. The basin is characterized by mountainous areas with peaks above 2500 m in the western part. The northern and northeastern parts are composed of lower mountain ranges and hills that surround the central and southern (southeastern) lowland areas. Following these elevation steps, the flow directions of the rivers are mainly from west and north toward the coast of the South China Sea in the southeast of the basin. The river system forms a large network delta before it enters into the sea. A comprehensive overview on the hydrological setting of the basin is given by Zhang et al. (2009b). Figure 1 provides a topographical sketch map of the Zhujiang River basin with the location of 192 weather stations and the main river system.

Fig. 1.

Topographical map of the Zhujiang River basin showing the 192 weather stations.

Fig. 1.

Topographical map of the Zhujiang River basin showing the 192 weather stations.

b. Data

Daily precipitation data covering the Zhujiang River basin were provided by the National Climate Center (NCC) of the China Meteorological Administration (CMA). For the basin, 253 stations passed the internal homogeneity check of the China National Meteorological Center (CNMC), including the moving t test (Peterson et al. 1998), standard normal homogeneity test (Alexandersson 1986), and departure accumulating method (Buishand 1982). Stations that were installed after 1961 and those with data gaps were excluded. As a result, 192 weather stations with precipitation records for 47 years (1 January 1961 to 31 December 2007) were selected. Data gaps account for less than 0.01% (less than 100 daily records out of over 3 000 000), and were reconstructed by the median precipitation from at least three neighboring stations. In most of these cases, 0-mm precipitation was interpolated as the neighboring stations did not record precipitation. In four cases, daily precipitation of less than 2 mm was interpolated.

National Centers for Environmental Prediction (NCEP) reanalysis data (geopotential height, and u- and υ-wind pattern for 1948–2010) provided by the National Oceanic and Atmospheric Administration/Oceanic and Atmospheric Research/Earth System Research Laboratory (NOAA/OAR/ESRL) Physical Sciences Division (PSD) were examined to understand the causes of observed precipitation trends (Kalnay et al. 1996).

Based on the CMA classification, a rain day is considered if at least 0.1 mm day−1 can be measured. Rainstorms are classified by minimum 50 mm day−1. An overview of the indices is given in Table 1.

Table 1.

Definition of precipitation indices.

Definition of precipitation indices.
Definition of precipitation indices.

c. Methodology

The Mann–Kendall trend test was applied to detect trends at all 192 stations (1961–2007) for annual, monthly, and daily precipitation sums; annual and monthly number of rain days and rain intensity; annual and monthly maximum precipitation; 5-day maximum precipitation, and number of rainstorms (>50 mm day−1). Trends for the indicator peak over threshold were calculated for the 90th, 95th, and 99th percentile, representing the annual number of days above the 90th, 95th, and 99th percentile for each station (1961–2007). The thresholds were calculated with daily data based on the time series 1961–2007.

The 90% confidence level was taken as the threshold to classify the significance of positive and negative trends for all indices. Trends below 90% confidence level were not considered. A comprehensive description and reference list of the Mann–Kendall trend test is provided by Gemmer et al. (2004) and Gao et al. (2010).

The inverse distance weighting (IDW) method is used to interpolate results for each station and to project the results two-dimensionally in the Arc geographical information system (ArcGIS). This interpolation method has been used and described by Gemmer et al. (2004). The interpolation results were compared with interpolations carried out by the kriging method (Goovaerts 2000; Yang et al. 2010), and little discrepancy was found. The Kriging method produced less generalized areas and followed the steps in the digital elevation model in more detail. For the sake of the readability of the illustrations in the necessary size for this journal, maps are produced by applying the IDW method. Interpolation was carried out for the following figures: 1) Fig. 2 shows the long-term annual precipitation for the Zhujiang River basin (1961–2007) using data interpolated from the 192 stations; 2) the shading in Fig. 4 shows the trend in January precipitation, the area being interpolated from the results of 192 stations; and 3) the shading in Fig. 5 shows the trend in October precipitation, the area being interpolated from the results of 192 stations.

Fig. 2.

Annual mean precipitation in the Zhujiang River basin, 1961–2007.

Fig. 2.

Annual mean precipitation in the Zhujiang River basin, 1961–2007.

Fig. 4.

January precipitation trends (area, left legend) and maximum precipitation trends (symbols, right legend) in the Zhujiang River basin, 1961–2007.

Fig. 4.

January precipitation trends (area, left legend) and maximum precipitation trends (symbols, right legend) in the Zhujiang River basin, 1961–2007.

Fig. 5.

As in Fig. 4, but for October.

Fig. 5.

As in Fig. 4, but for October.

To distinguish between dry and wet years (Fig. 8), the annual values of five precipitation indices are standardized by subtracting the indicator’s arithmetic mean from the annual value and dividing it by the indicator’s standard deviation. For each year, the arithmetic mean of the five annual values is calculated. When the arithmetic mean of the five annual standardized values is negative (positive), we determine a dry (wet) year. The degree dryness and wetness is higher when the values are higher.

Fig. 8.

Arithmetic mean of five standardized precipitation indices (monthly precipitation, 5-day maximum precipitation, precipitation intensity, rain days, maximum precipitation) in the Zhujiang River basin for October 1961–2007, mean wind direction in the eastern part of the Zhujiang River basin October 1961–2007, linear trend (line), and highlighted severe dry (triangle) and wet (square) Octobers.

Fig. 8.

Arithmetic mean of five standardized precipitation indices (monthly precipitation, 5-day maximum precipitation, precipitation intensity, rain days, maximum precipitation) in the Zhujiang River basin for October 1961–2007, mean wind direction in the eastern part of the Zhujiang River basin October 1961–2007, linear trend (line), and highlighted severe dry (triangle) and wet (square) Octobers.

3. Results

a. Annual and monthly precipitation trends

Annual precipitation (multiannual mean from 1961 to 2007) in the Zhujiang River basin shows an increase from west to south, from less than 1000 mm to more than 2000 mm at the coastline in the east (Fig. 2). The annual average precipitation in the basin is 1500 mm. The Mann–Kendall trend test was applied to the annual precipitation of the 192 stations in the Zhujiang River basin. Hardly any significant trends in annual precipitation during 1961–2007 can be detected. Only eight out of 192 stations (4%) show significant trends above the 90% confidence level. Seven of these stations are located in the mountainous west of the Zhujiang River basin and show negative trends. The other station is located in the east and shows a positive trend (not displayed).

The monthly precipitation (multiannual mean over 1961–2007) in the Zhujiang River basin shows a distinct seasonality. Sixty-one percent of the annual precipitation falls from May to August. With an average of 266 mm (29 mm), June (December) is the month with the highest (lowest) precipitation. The Mann–Kendall trend test was applied to the monthly precipitation time series of the 192 stations. December is the only month that shows no significant trend at any station at a 90% confidence level (Fig. 3). Monthly precipitation in January and October shows the largest trends in precipitation indices. Precipitation in January shows significant upward trends at 50 stations (26%) and October precipitation shows significant downward trends at 76 stations (40%). Precipitation in April has significantly decreased at 50 stations (26%). However, the increasing and decreasing trends of the four other precipitation indices (Fig. 3) are less dominant in April than in January or October.

Fig. 3.

Observed monthly precipitation trends 1961–2007 at the 90% confidence level in the Zhujiang River basin for five indices.

Fig. 3.

Observed monthly precipitation trends 1961–2007 at the 90% confidence level in the Zhujiang River basin for five indices.

Positive precipitation trends in January are concentrated in a belt from the tributaries in the west of the basin to the northern boundary. This spatial distribution is similar to the area that experienced trends in maximum precipitation in January (Fig. 4). The negative precipitation trends for October can be found in a belt stretching from the middle reaches of the Zhujiang in the center of the basin to the boundary in the east along the coastline. This area is similar to the region for which trends in monthly maximum precipitation in October can be detected (Fig. 5).

Significant negative trends in monthly precipitation can only be detected from April to June and from August to November. Positive trends occurred from January to September only. Both positive (19 stations or 10%) and negative precipitation trends (17 stations or 9%) can be detected in August.

b. Trends in annual extremes

Annual maximum precipitation (the highest rainfall per day at one station per year) can be detected in areas that are similar to the spatial distribution of annual mean precipitation (Fig. 2). Annual maximum precipitation is below 80 mm in the west of the catchment, between 120 and 160 mm along the coastline, and more than 160 mm in spots along the coast. The Mann–Kendall test detects trends in annual maximum precipitation at a 90% confidence level, but this precipitation index decreased at 6 stations (3%) and increased at 12 stations (6%) only. The results are not displayed as no large-scale trend can be found for annual maximum precipitation.

The annual 5-day maximum precipitation (long-term annual average) is less than 140 mm in the western part and more than 260 mm in the south and southeast coast of the research area. The Mann–Kendall test shows negative trends at a 90% confidence level at 5 stations (3%) and positive trends at 8 stations (4%) only and is not displayed.

The number of annual rain days shows a significant downward trend at a 90% confidence level at 94 stations (49%). These trends were mainly experienced in the west of the Zhujiang River basin. The trend test shows no clear signal for the annual number of rainstorms days (>50 mm day−1). Positive trends can be detected at 12 stations (6%) and negative trends occurred at 2 stations (1%) only.

The results for the Mann–Kendall test on the peak over threshold at 90th, 95th, and 99th percentile detects trends at few stations only. For the 90th percentile, seven positive (4%) and three (2%) negative trends were detected. Eleven positive (6%) and two (1%) negative trends occurred for the 95th percentile. Fourteen positive (7%) and four (2%) negative trends are detected for the 99th percentile.

c. Trends in monthly extremes

The results of the Mann–Kendall trend test for indices on a monthly basis are more diverse and concise than those for annual data. Figure 3 displays the results of the trend test for five indices on a monthly basis, indicating the number of stations and percentage of stations that have experienced significant trends. The results for trends in monthly precipitation have been discussed in the previous subsection. For the five indices, significant trends in either a negative or positive direction appear in certain months. The only exception is August, where monthly precipitation and monthly 5-day maximum precipitation show a balanced number of positive and negative trends.

The number of monthly rain days in October experienced the most dominant trend of all indices. Rain days are significantly decreasing in October at 121 stations (63%), followed in the ranking by November (76 stations or 35%) and September (62 stations or 32%). No increasing trend for rain days can be found at any station in these three months. A decreasing trend in rain days occurred exclusively in the months between April and December. In general, hardly any positive trends in the number of monthly rain days can be found in the Zhujiang River basin. Only May shows a slight increase in rain days at 10 stations (5%).

Precipitation intensity shows a scattered appearance over the months. The highest increase in precipitation intensity can be observed in January (64 stations or 33%), followed by July (29 stations or 15%) and December (25 stations or 13%). No station experienced a decreasing trend in precipitation intensity in these months. Negative trends in precipitation intensity mainly occurred in October (39 stations or 20%) and April (38 stations). The magnitude of these decreasing trends is much lower compared to the 64 stations (33%) with increasing trends in precipitation intensity in January.

Maximum precipitation experienced the highest increase in January (73 stations or 38%) and the most severe decrease in October (65 stations or 34%).

Monthly 5-day maximum precipitation experienced a significant increase in March (45 stations or 23%) and January (33 stations or 17%). October experienced a negative trend in 5-day maximum precipitation amounts over the period from 1961 to 2007. This decreasing trend was detected at 77 stations (40%).

Figures 4 and 5 illustrate trends in monthly precipitation and maximum precipitation in January and October, respectively. January and October represent the most dominant increase and decrease in the precipitation indices. The figures show that the northwest of the Zhujiang River basin experienced higher precipitation and increased maximum precipitation in January. A band stretching along the middle and lower reaches of the Zhujiang River experienced a decrease in precipitation and lower maximum precipitation in October. It is noteworthy that the trends in the indices, whether positive or negative, occurred in the same regions and stations for a given month. With the exception of one station for one index, no significant negative trends can be detected for any of the indices between January and March. With the exception of six stations, the months October and November experienced negative trends for the indices and no positive trends.

Considering all five indices, January (October) experienced the most pronounced increasing (decreasing) trend in precipitation and the analyzed precipitation extremes.

4. Interpretation: Relation to large-scale circulation

Yang et al. (2010) described an approach to link flood seasons in the Zhujiang River basin with large-scale circulation and identified monthly mean moisture transports from the southwest Pacific Ocean and Indian Ocean as being the driver of wet and dry seasons.

In another example from the Yangtze River catchment, reasons for positive precipitation trends in July were found in variations in the meridional wind pattern at the 850-hPa level, which account for an increased transport of warm moist air to the Yangtze River catchment during the summer months (Becker et al. 2006). Hartmann et al. (2009) found phases and anti-phases in high precipitation at some stations in the Yangtze River catchment that are linked with SST in the Bay of Bengal. The detected directions of 850-hPa winds on a seasonal basis are in line with the seasonal water vapor flux in the Zhujiang River basin described by Zhang et al. (2009c).

We have therefore investigated geopotential heights and winds at 850 hPa for January and October for the entire time series covering the period 1961–2007 in order to explain reasons for the observed monthly (extreme) precipitation trends in the Zhujiang River basin. Interesting conclusions can be drawn. Figure 6 summarizes the findings and displays geopotential heights and winds at 850 hPa. The mean geopotential heights and winds for the time series (upper panel) are shown for January (Fig. 6a) and October (Fig. 6b).

Fig. 6.

Characteristics of 850-hPa geopotential height and winds: (a) mean of all Januarys 1961–2007, (b) mean of all Octobers 1961–2007, (c) composite of Januarys with high precipitation, (d) composite of Octobers with high precipitation, (e) composite of Januarys with low precipitation, and (f) composite of Octobers with low precipitation (the Zhujiang River basin is outlined in black; winds at 850 hPa are theoretical for the Qinghai–Tibet Plateau or any higher mountain range).

Fig. 6.

Characteristics of 850-hPa geopotential height and winds: (a) mean of all Januarys 1961–2007, (b) mean of all Octobers 1961–2007, (c) composite of Januarys with high precipitation, (d) composite of Octobers with high precipitation, (e) composite of Januarys with low precipitation, and (f) composite of Octobers with low precipitation (the Zhujiang River basin is outlined in black; winds at 850 hPa are theoretical for the Qinghai–Tibet Plateau or any higher mountain range).

a. January precipitation

When investigating the observed data, the positive precipitation trends in January cannot be explained by a shift in precipitation from December or February to January, as neither December nor February shows negative trends. In an attempt to explain the changes in January and October, focus is placed on the East Asian winter monsoon (EAWM).

According to the mean geopotential height at 850-hPa level, and as can be seen from Fig. 6a, January precipitation in the Zhujiang River basin is influenced by an interaction between three high pressure areas. Here, the most distinct area is over west China; others are a large area over southeast China crossing Taiwan toward the west Pacific, and one over India. Prevailing winds in the Zhujiang River basin circulate over southeast China, the west Pacific, and the South China Sea, and the wind direction is south–southwest.

Figure 6c shows a composite of the monthly circulation pattern for years with high January precipitation (1969, 1983, 1992, 1997, and 2003). This circulation pattern is similar in each of the years with “wet” January. The circulation is influenced by a distinct high over west China and a band of high pressure that covers the entire east and southeast Asian continent. Circulation is similar to the multiannual mean (Fig. 6a), but the circulation over the Zhujiang River basin is less strong. A composite of the large-scale system for the years 1963, 1976, 1986, 1994, and 2006 with low January precipitation (dry January) is shown in Fig. 6e. It is quite similar to the mean; however, the high over India is less distinct. At the same time, the western extension of the anticyclone over southeast China hardly covers the coastal area. The speed of the winds entering the Zhujiang River basin is below average.

When examining the monthly time series of winds at 850 hPa since 1961, it can be concluded that average and high precipitation in January occurs when a high pressure field north of the Himalayas forms, the intensity and size of which is stronger than that of the high over India. Reasons for positive precipitation trends in January, however, cannot entirely be explained by changes in large-scale circulation. The atmospheric conditions are, to a certain extent, stable in January with no distinct changes in the mean wind directions. The positive trends might be related to local climatic conditions in the topographically diverse basin of the Zhujiang River; they might also be driven by the different wind speed that can be observed for years with precipitation means above and below January.

b. October precipitation

The average large-scale circulation in October is shown in Fig. 6b. In the multiannual mean in October, two anticyclones form over west and east China, respectively. Strong easterly winds are supported by a low pressure system over the Philippines and are blocked from entering the Zhujiang River basin. In October, the long-term wind direction in the basin is easterly and the winds have crossed the East China Sea. Strong winds from the east transport water vapor to the basin if the high over east China has not developed. The system is further supported by the belt of lower pressure south of China.

In Fig. 6d, the circulation in October has been composited for five years with high October precipitation (1965, 1976, 1995, 2000, and 2002). Winds from the east-southeast transport water vapor to the Zhujiang River basin. Precipitation, number of rain days, and other precipitation indices in October are higher in the basin if the high over west China is small in size.

Figure 6f shows a composite of the large-scale circulation in October for the years 1967, 1979, 1992, 2004, and 2007, which show the most significant negative precipitation indices (see Fig. 7). It represents the October circulation during dry conditions. Low October precipitation indices are detected if a high pressure system forms over east China and if this is bordered by a trough to the east. The mean wind direction is north-northeast. The water vapor is potentially lower as it is transported directly from north China and the Bohai Sea.

Fig. 7.

Arithmetic mean of five standardized precipitation indices (monthly precipitation, 5-day max precipitation, precipitation intensity, rain days, maximum precipitation) for October from 1961 to 2007 in the Zhujiang River basin.

Fig. 7.

Arithmetic mean of five standardized precipitation indices (monthly precipitation, 5-day max precipitation, precipitation intensity, rain days, maximum precipitation) for October from 1961 to 2007 in the Zhujiang River basin.

The negative trend in precipitation indices in October is influenced by a large-scale high over west China and a deep pressure system in the western North Pacific that forms a trough, changing the mean wind direction in the Zhujiang River basin from east to northeast or east-northeast. This is the so-called East Asian Trough, which is typical for the EAWM (Wang et al. 2009). Wang and Ding (1997) described the onset of the EAWM as being in October. The EAWM is delivering dry, cold air from northeast China to the Zhujiang River basin.

Figure 7 shows the annual arithmetic mean (see section 2) of five standardized precipitation indices (monthly precipitation, 5-day maximum precipitation, precipitation intensity, rain days, and maximum precipitation) in the Zhujiang River basin in October from 1961 to 2007. As can be seen, some years show pronounced negative means and indicate dry October months (2007, 2005, 2004, 2003, 1992, 1979, 1985, and 1967). Figure 8 shows the arithmetic mean of the same five standardized precipitation indices for October during 1961 to 2007 and the according mean wind direction in the eastern part of the Zhujiang River basin for each index (e.g., wind direction in dry and wet October months). The linear trend of the wind direction 1961–2007 is also displayed. It becomes clear that years with extremely negative means of standardized precipitation indices (dry October months) occurred when the large-scale winds at 850 hPa entered the Zhujiang River basin from the northeast. With the exception of 1985, the nine October months with a mean of −1 or below (i.e., the nine most severe dry October months, indicated by the triangle in Fig. 8) can be explained by general circulation typical for the EAWM (89%). Wet October months can be observed in years when the mean wind direction is east or southeast (indicated by the square in the right part of Fig. 8) with the exception of 1965. The dry (wet) October in the exceptional year 1985 (1965) might be explained by local circumstances—for example, occurrence of a tropical cyclone or heat wave instead of by large-scale wind patterns. The linear trend displayed in Fig. 8 illustrates that the more wind comes from northeast (southeast) in October, the drier (wetter) the weather conditions are. The observed trends toward lower precipitation in October might be related to an early onset of the EAWM, but is definitely caused by the change of the mean wind directions. Over the time series 1961–2007, the mean wind direction in the Zhujiang River basin has experienced a change by 50° from east-southeast to east-northeast (Fig. 9). Checking the occurrence of El Niño and La Niña years leads to the assumption that low precipitation and its indices in October are associated with northerly winds in the transition phase between two El Niño events. This is consistent with the findings of Zhou et al. (2007), who found many strong EAWM in the years before the developing year of an El Niño or during the decaying La Niña years.

Fig. 9.

Mean wind direction in the eastern part of the Zhujiang River basin for October 1961–2007 and its linear trend.

Fig. 9.

Mean wind direction in the eastern part of the Zhujiang River basin for October 1961–2007 and its linear trend.

5. Discussion and conclusions

The paper has made use of some basic methods to analyze data on precipitation trends in the Zhujiang River basin that have not been made available in the current form elsewhere. The approach is very helpful to detect positive and negative trends in precipitation and indices related to precipitation extremes on a monthly and daily basis. Few stations show significant trends above a 90% confidence level for annual indices. The only index that shows a somewhat negative direction is the number of annual rain days. This is consistent with the findings of Zhang et al. (2009a). Annual maximum precipitation shows trends at 18 of 192 stations. This is the highest number of detected trends on an annual basis and indicates that less than 9% of all stations in the Zhujiang River basin show significant trends in annual extreme precipitation events (maximum precipitation, rainstorm days, intensive rain days, extremely intensive rain days for 90/95/99 percentile rain events). No spatial pattern can be detected for the stations with significant trends. Zhang et al. (2009a) also report that no trends can be detected for annual precipitation in the Zhujiang River basin. We can therefore conclude that no distinct regions in the Zhujiang River basin have experienced trends for annual indices.

Our results for the monthly precipitation trends as well as for the monthly indices of rain days, intensity, maximum precipitation, and 5-day maximum precipitation are somewhat in line with the seasonal findings by Zhang et al. (2009a), who estimate increasing trends in total precipitation, precipitation intensity, and rain days in winter [(December–February (DJF)]. Our findings are also supported on a seasonal basis by Liu et al. (2009), who detected that autumn precipitation decreased, but spring, summer, and winter precipitation increased during the period covered by the same time series.

Each of the previous studies delivered findings that are correct for the time series of data and the number of stations used. Comparing this paper’s results on monthly and daily basis with previous studies on seasonal basis shows that much more interpretation is possible on monthly basis. If monthly results were merged for seasons (e.g., September, October, and November merged to “autumn”), the information on the significant decrease of precipitation intensity in October would have been lost. Additionally, monthly changes of wind direction that could explain trends in October precipitation could not be assessed.

In this study, 192 stations that passed a homogeneity and quality check have been used, compared to other studies that used less than one quarter of these stations without official checks. This, in combination with the use of the maximum possible number and quality of stations, delivers new information on the quality and quantity of observed trends.

The observed positive precipitation trends in January cannot be explained by changes in large-scale wind directions. Reasons might be found in local climatic conditions or in the generally low monthly precipitation. Generally, the wind speed is different in years with observed high, mean, and low precipitation in January, which might be a factor causing positive trends. This will be examined in future studies.

The change of the mean wind direction in years with below-average precipitation and rain days in October is an interesting finding that has not been described before. Chou (2004) described a weak anomalous cyclone over the Philippine Sea and the South China Sea in the developing summer of an El Niño event. October, marking the transition period between the East Asian summer monsoon (EASM) and the EAWM, has not received much attention in climate studies yet, but it appears to be sensitive with regards to areal precipitation. The negative trends in the precipitation indices in October are consistent with the trend in the mean wind direction at 850 hPa. Therefore, low precipitation records in October can be explained by pressure systems typical for the EAWM and winds from northeast and east-northeast in the eastern part of the Zhujiang River.

In forthcoming research, the authors will further examine the return periods of extreme events on a monthly basis for below- (above-) average October (January) precipitation and indices and investigate links with the statistical onset of the EAWM.

Acknowledgments

This study was supported by the National Basic Research Program of China (973 Program) 2010CB428401 and the Non-Profit Industry Fund of the Ministry of Water Resources (200701005). The positions of Marco Gemmer and Thomas Fischer at the National Climate Center are supported by German Development Cooperation through the Center for International Migration and Development (http://www.cimonline.de). Cordial thanks are extended to the editor Dr. Nathan Gillett and the anonymous reviewers for their professional comments and suggestions, which greatly improved the quality of this manuscript, and to Dr. Andreas Wilkes of the World Agroforestry Center (ICRAF) China Programme, who helped improve the English of this manuscript.

REFERENCES

REFERENCES
Alexandersson
,
H.
,
1986
:
A homogeneity test applied to precipitation data.
J. Climatol.
,
6
,
661
675
.
Becker
,
S.
,
M.
Gemmer
, and
T.
Jiang
,
2006
:
Spatiotemporal analysis of precipitation trends in the Yangtze River catchment.
Stochastic Environ. Res. Risk Assess.
,
20
,
435
444
.
doi:10.1007/s00477-006-0036-7
.
Buishand
,
T. A.
,
1982
:
Some methods for testing the homogeneity of rainfall records.
J. Hydrol.
,
58
,
11
27
.
Chen
,
W.
,
H-F.
Graf
, and
R.
Huang
,
2000
:
The interannual variability of East Asian Winter Monsoon and its relation to summer monsoon.
Adv. Atmos. Sci.
,
17
,
48
60
.
Chou
,
C.
,
2004
:
Establishment of the low-level wind anomalies over the western North Pacific during ENSO development.
J. Climate
,
17
,
2195
2212
.
Ding
,
Y.
, and
Coauthors
,
2007
:
China’s National Assessment Report on Climate Change (I): Climate change in China and the future trend.
Adv. Climate Change Res.
,
3
, (
Suppl.
).
1
5
.
Feng
,
S.
,
S.
Nadarajah
, and
Q.
Hu
,
2007
:
Modeling annual extreme precipitation in China using the generalized extreme value distribution.
J. Meteor. Soc. Japan
,
85
,
599
613
.
Fu
,
J.
,
W.
Qian
,
X.
Lin
, and
D.
Chen
,
2008
:
Trends in graded precipitation in China from 1961 to 2000.
Adv. Atmos. Sci.
,
25
,
267
278
.
Gao
,
C.
,
M.
Gemmer
,
X.
Zeng
,
B.
Liu
,
B.
Su
, and
Y.
Wen
,
2010
:
Projected streamflow in the Huaihe River basin (2010–2100) using artificial neural network.
Stochastic Environ. Res. Risk Assess.
,
24
,
685
697
.
Gemmer
,
M.
,
S.
Becker
, and
T.
Jiang
,
2004
:
Observed monthly precipitation trends in China 1951–2002.
Theor. Appl. Climatol.
,
77
,
39
45
.
Goovaerts
,
P.
,
2000
:
Geostatistical approaches for incorporating elevation into the spatial interpolation of rainfall.
J. Hydrol.
,
228
,
113
129
.
Hartmann
,
H.
,
L.
King
,
T.
Jiang
, and
S.
Becker
,
2009
:
Quasi-cycles in Chinese precipitation time series and in their potential influencing factors.
Quat. Int.
,
208
,
28
37
.
Kalnay
,
E.
, and
Coauthors
,
1996
:
The NCEP/NCAR 40-Year Reanalysis Project.
Bull. Amer. Meteor. Soc.
,
77
,
437
471
.
Liu
,
B.
,
M.
Xu
,
M.
Henderson
, and
Y.
Qi
,
2005
:
Observed trends of precipitation amount, frequency, and intensity in China, 1960–2000.
J. Geophys. Res.
,
110
,
D08103
.
doi:10.1029/2004JD004864
.
Liu
,
L.
,
T.
Jiang
, and
F.
Yuan
,
2009
:
Observed (1961–2007) and projected (2011–2060) climate change in the Pearl River basin (in Chinese).
Adv. Climate Change Res.
,
5
,
209
214
.
Luo
,
Y.
,
S.
Liu
,
S.
Fu
,
J.
Liu
,
G.
Wang
, and
G.
Zhou
,
2008
:
Trends of precipitation in Beijiang River basin, Guangdong Province, China.
Hydrol. Processes
,
22
,
2377
2386
.
Peterson
,
T. C.
, and
Coauthors
,
1998
:
Homogeneity adjustments of in situ atmospheric climate data: A review.
Int. J. Climatol.
,
18
,
1493
1517
.
Qian
,
W.
, and
X.
Lin
,
2005
:
Regional trends in recent precipitation indices in China.
Meteor. Atmos. Phys.
,
90
,
193
207
.
Roeckner
,
E.
, and
Coauthors
,
2003
:
The atmospheric general circulation model ECHAM5.
Part I: Model description. Max-Planck-Institute for Meteorology Rep. 349, 127 pp
.
Trenberth
,
K. E.
, and
Coauthors
,
2007
:
Observations: Surface and atmospheric climate change.
Climate Change 2007: The Physical Science Basis, S. Solomon et al., Eds., Cambridge University Press, 235–336
.
Wang
,
L.
,
R.
Huang
,
L.
Gu
,
W.
Chen
, and
L.
Kang
,
2009
:
Interdecadal variations of the East Asian winter monsoon and their association with quasi-stationary planetary wave activity.
J. Climate
,
22
,
4860
4872
.
Wang
,
Q.
, and
Y.
Ding
,
1997
:
Climatological characteristics of evolution of East Asian Winter Monsoon.
Quart. J. Appl. Meteor.
,
8
,
186
196
.
Wang
,
Y.
, and
L.
Zhou
,
2005
:
Observed trends in extreme precipitation events in China during 1961–2001 and the associated changes in large-scale circulation.
Geophys. Res. Lett.
,
32
,
L09707
.
doi:10.1029/2005GL022574
.
Yang
,
F.
, and
K-M.
Lau
,
2004
:
Trend and variability of China precipitation in spring and summer: Linkage to sea surface temperatures.
Int. J. Climatol.
,
24
,
1625
1644
.
Yang
,
T.
,
Q.
Shao
,
Z-C.
Hao
,
X.
Chen
,
Z.
Zhang
,
C-Y.
Xu
, and
L.
Sun
,
2010
:
Regional frequency analysis and spatio-temporal pattern characterization of rainfall extremes in the Pearl River basin, China.
J. Hydrol.
,
380
,
386
405
.
Yao
,
C.
,
S.
Yang
,
W.
Qian
,
Z.
Lin
, and
M.
Wen
,
2008
:
Regional summer precipitation events in Asia and their changes in the past decades.
J. Geophys. Res.
,
113
,
D17107
.
doi:10.1029/2007JD009603
.
Zhai
,
J.
,
B.
Su
,
V.
Krysanova
,
T.
Vetter
,
C.
Gao
, and
T.
Jiang
,
2010
:
Spatial variation and trends in PDSI and SPI indices and their relation to streamflow in 10 large regions of China.
J. Climate
,
23
,
649
663
.
Zhai
,
P.
,
A.
Sun
,
F.
Ren
,
X.
Liu
,
B.
Gao
, and
Q.
Zhang
,
1999
:
Changes of climate extremes in China.
Climatic Change
,
42
,
203
218
.
Zhai
,
P.
,
X.
Zhang
,
H.
Wan
, and
X.
Pan
,
2005
:
Trends in total precipitation and frequency of daily precipitation extremes over China.
J. Climate
,
18
,
1096
1108
.
Zhang
,
Q.
,
C-Y.
Xu
,
S.
Becker
,
Z. X.
Zhang
,
Y. D.
Chen
, and
M.
Coulibaly
,
2009a
:
Trends and abrupt changes of precipitation maxima in the Pearl River basin, China.
Atmos. Sci. Lett.
,
10
,
132
144
.
Zhang
,
Q.
,
C-Y.
Xu
,
M.
Gemmer
,
Y. D.
Chen
, and
C.
Liu
,
2009b
:
Changing properties of precipitation concentration in the Pearl River basin, China.
Stochastic Environ. Res. Risk Assess.
,
23
,
377
385
.
Zhang
,
Q.
,
C-Y.
Xu
, and
Z.
Zhang
,
2009c
:
Observed changes of drought/wetness episodes in the Pearl River basin, China, using the standardized precipitation index and aridity index.
Theor. Appl. Climatol.
,
98
,
89
99
.
Zhou
,
L-T.
, and
R.
Wu
,
2010
:
Respective impacts of the East Asian winter monsoon and ENSO on winter rainfall in China.
J. Geophys. Res.
,
115
,
D02107
.
doi:10.1029/2009JD012502
.
Zhou
,
W.
,
X.
Wang
,
T. J.
Zhou
,
C.
Li
, and
J. C. L.
Chan
,
2007
:
Interdecadal variability of the relationship between the East Asian winter monsoon and ENSO.
Meteor. Atmos. Phys.
,
98
,
283
293
.

Footnotes

Corresponding author address: Dr. Marco Gemmer, National Climate Center, 46, Zhongguancun Nandajie, Haidian 100081, Beijing, China. Email: marco@gemmeronline.de