Influence of summer monsoon diurnal cycle on moisture transport and precipitation over eastern China



[1] Diurnal variability of the summer monsoon over China, a key element affecting regional climate, is examined using the latest reanalysis dataset and satellite rain estimates. Diurnal variation of low-level wind is found to be pronounced over South China during active monsoon days. Mean wind speed attains a maximum (minimum) in early morning (afternoon), with a diurnal range of 2–3 m/s, double that of inactive monsoon days. The largest amplitude typically appears at 850 hPa, consistent with radiosonde observations. Such a monsoon diurnal cycle can strengthen low-level moisture transport at night by about 20% more than during the day. Nocturnal moisture fluxes converge toward Central China and lead to a meso-synoptic-scale moisture sink during the late night and morning, which plays a role in regulating the regional water budget on a diurnal time scale. Monsoon flow also helps provide substantial moisture over low-lying areas in the morning hours. Correspondingly, morning rainfall undergoes a remarkable increase and greatly contributes to the diurnal cycle of summer rainfall. The strongest response comes from meso-α-scale rain events that not only become prominent during active monsoon days but also possess a dominant morning peak. These morning events occur mainly on the basins and plains of Central China, where the monsoon diurnal cycle promotes nighttime mesoscale convection. These tend to shift northward from June to August, with the progress of the monsoon diurnal cycle, thereby producing the morning-peak summer rainband. The findings point to an efficiency of nocturnal monsoon flow influencing the warm-season weather and climate over eastern China.

1 Introduction

[2] The summer monsoon has a profound influence on the weather and climate over East Asian countries. The monsoon variability, ranging from diurnal to interdecadal time scales, is a much researched subject [e.g., Ding, 1992; Chen et al., 2004]. In recent decades, the diurnal variation in winds, convective activities, and precipitation has been recognized as being a key aspect of the warm-season climate in tropical and monsoon regions [Ohsawa et al., 2001; Wang et al., 2004; Hirose and Nakamura, 2005; Yang and Smith, 2006; Yu et al., 2007; Chen et al., 2009a; Mao and Wu, 2011]. Gaining further knowledge of the monsoon climate on a diurnal time scale has important hydrological implications for improved heavy rain prediction and water resource management and remains a topic for current and future works.

[3] The monsoon flow, on a short time scale, is characterized by remarkable diurnal variation [Ramage, 1952]. Over the Asian continent, the summer monsoon exhibits a pronounced diurnal pulsation of its large-scale divergent circulation [Krishnamurti and Kishtawal, 2000]. Over eastern China (Figure 1), the diurnal variation is clearly seen in the low-level winds on a regional scale. For example, the southwesterly of tropical monsoon is most evident at night, as illustrated by the observations made in Hong Kong [Johnson, 2006]. The low-level wind speed, recorded at mountain sites in South China, tends to reach a peak in the predawn [Yu et al., 2009]. In particular, the diurnal wind variation over eastern China attains large amplitude in the warm season after the onset of the summer monsoon [Chen et al., 2009b]. The increased velocity of the southwesterly wind during the night helps to establish the low-level jet (LLJ) over South China [Rife et al., 2010]. These observational studies reveal that nocturnal strong southwesterlies, with a meso-synoptic span of several hundred to a thousand kilometers, are fairly common occurrences over eastern China. Although such events seem to be rooted in the monsoon activities, a direct survey of the diurnal cycle of monsoon flow over eastern China has not been fully researched.

Figure 1.

Daily mean of TRMM rainfall and 850 hPa horizontal winds on (a) active and (b) inactive monsoon days. In Figure 1a, the low-level convergence over eastern China stronger than –1 × 10−6 s−1 is hatched. Eastern China implies the area of mainland China east of 105°E, including three subdomains as marked by the dashed rectangles: South China (105°–120°E, 21°–27°N), Central China (105°–120°E, 27°–35°N), and North China (105°–120°E, 35°–38°N). The long dashes mark the elevations of 1500 m and 3000 m.

[4] Diurnal variation in the low-level wind has been linked to the moisture supply and moist convection over eastern China. The low-level southerly flow is shown to enhance moisture transport from South China to Central China during the night [Li et al., 2007]. The intensity of this moist inflow to the Meiyu frontal zone is a deciding factor in the duration of the nocturnal rainband over Central China [Yamada et al., 2007]. The nocturnal southwesterlies, in addition to transporting moisture, also strengthen the low-level moisture convergence, frontogenesis, and convective instability that favor nighttime convection growth [Chen et al., 2009b; Chen et al., 2010; Yuan et al., 2010; Ueno et al., 2011]. Climatologically, the mesoscale convective systems prefer to form and grow in the environment of an enhanced, moist southerly at night [Miller and Fritsch, 1991]. Observations have been made of the significant relationship between the nocturnal LLJ over South China and the extreme heavy rainfall over Central China [Monaghan et al., 2010].

[5] Precipitation diurnal variability has also been the subject of research to illustrate the possible influence of the summer monsoon over eastern China on a short time scale. The rainband along the Meiyu front is found to attain a morning peak, while the rainfall in adjacent areas exhibits an afternoon peak [Geng and Yamada, 2007]. The morning rainfall dominates eastern China in the Meiyu season, while the afternoon rainfall is dominant in midsummer [Xu and Zipser, 2011]. The morning-peak rainfall tends to shift northward, along with the summer progress of the monsoon rainband [Yin et al., 2009; Chen et al., 2009a]. It has recently been shown that morning rainfall is pronounced over Central China during active monsoon period, while afternoon rainfall becomes more pronounced during monsoon break period [Yuan et al., 2010]. In particular, research has focused on the morning rain systems as they explain the majority of the variance in the amount of seasonal rainfall, leading to the anomalous wet and dry seasons over Central China [Chen et al., 2012a]. As indicated above, the diurnal cycle of hydrological processes in relation to the summer monsoon should be addressed as a key issue in extending our understanding of the regional climate.

[6] Previous studies have shed insight into diurnal variability, but most have been based on a summer mean or cases of short periods. Diurnal variability over eastern China, in relation to monsoon activities, requires further studies. In particular, the impact of the monsoon diurnal cycle on moisture transport and summer precipitation remains to be quantified. The objective of this study is to address such processes using the latest archived data. We provide a unique addition to previous works in that the relative importance of monsoon diurnal cycle in summer climate is estimated. This article is organized as follows: Section 2 describes the dataset used in this study and categorizes the monsoon days. Section 3 offers a brief explanation on the general features of monsoon activities and related rainfall diurnal cycle. Section 4 examines the spatial structures of the monsoon diurnal cycle. Section 5 presents the induced diurnal cycles of moisture transport and regional water budget. Section 6 illustrates the role of mesoscale rain events in producing the rainfall budget as a response to the monsoon diurnal cycle. The final section presents a discussion and summary.

2 Dataset and Monsoon Day Categorization

2.1 Reanalysis Data, Sounding Profile and Satellite-Derived Rainfall

[7] As in situ observations are limited by time-space sampling, reanalysis data are generally employed for climatological studies on wind diurnal oscillation and other related processes [Higgins et al., 1997]. In the current study, we use the latest reanalysis product from the European Centre for Medium-Range Weather Forecasts (ECMWF). The ECMWF Re-Analysis (ERA-interim) assimilates surface observations, soundings, and a variety of satellite products [Dee et al., 2011]. The variables include the pressure-level analysis of winds, humidity, temperature, and so on. The spatial resolution is 1.5° × 1.5°, and the temporal resolution is six hourly. With an interval of 25 hPa below 750 hPa, the ERA-interim is suitable for resolving the diurnal variation of wind and moisture in the lower troposphere. It also offers a forecast for the surface variables such as precipitation and evaporation at three hourly intervals. As shown by Dee et al. [2011], the ERA-interim has a good representation of the hydrological cycle because of its improved physics scheme and data assimilation. In this study, the ERA-interim data from 1989 to 2010 are used to examine the long-term mean of the monsoon diurnal cycle and its related moist processes (sections 4 and 5).

[8] The ERA-interim dataset has been shown to represent the horizontal pattern of wind diurnal variation over many regions around the world, including China [Rife et al., 2010; Monaghan et al., 2010]. However, the question remains as to whether the data capture the nature of the monsoon diurnal cycle, particularly the vertical structure. For validation, we used the intensive observations from the South China Sea Monsoon Experiment [Lau et al., 2000]. Six hourly sounding profiles are available at a dozen sites in South China during 5–21 June 1998 when the summer monsoon is active over eastern China. These quality-control data of horizontal winds, temperature, and moisture have a good vertical resolution in the lower troposphere, and thus, they are able to depict the vertical structure of monsoon diurnal cycle. A spatial average of sounding profiles over South China gives a reference of six hourly vertical profiles, which is compared with those in reanalysis data in section 4.2.

[9] In this study, the Tropical Rainfall Measuring Mission (TRMM) 3B42 product is used for rainfall observation. The 3B42 dataset offers three hourly rain estimates with a spatial resolution of 0.25° × 0.25° [Huffman et al., 2007]. The rain rate has been calibrated by rain-gauge observations on land and thus is applicable for studying the rainfall budget. This short-interval dataset is widely used to monitor precipitation diurnal variability [Kikuchi and Wang, 2008; Mao and Wu, 2011]. Over eastern China, 3B42 resolves a large amount of morning rainfall, despite an underestimate of a few fractions [Shen et al., 2010]. In particular, it captures the difference in the rainfall diurnal cycle between the anomalous wet and dry seasons associated with the summer monsoon [Chen et al., 2012]. The 3B42, with its 0.25° mesh resolution, is also useful for studying the rainfall volume by mesoscale convection [Huffman et al., 2007; Demaria et al., 2011]. In this study, 3B42 data from 1998 to 2010 are used to record the huge number of rain events shown in section 6. The plentiful samples derived from 13 years of archives have helped us quantify the role of precipitation systems in producing the rainfall diurnal cycle as a response to monsoon variability.

[10] Local time (LT = UTC + 8 h) is applied in this study. The four synoptic times (14:00, 20:00, 02:00, and 08:00 LT) in the reanalysis data correspond to the afternoon, early evening, late night, and morning, respectively. In studying the rainfall budget, we group three hourly 3B42 rain estimates into P.M. (14:00, 17:00, 20:00, and 23:00 LT) and A.M. (02:00, 05:00, 08:00, and 11:00 LT) hours, i.e., the afternoon-evening rainfall and the late night-morning rainfall.

2.2 Categorization of Active/Inactive Monsoon Days

[11] The monsoon flow passes over South China before releasing precipitation over Central China and other East Asian regions. The flow features a surge of low-level southwesterlies in the Meiyu season or a channel of southerlies at the western flank of the subtropical high [e.g., Ding, 1992]. To classify monsoon days, certain variables such as wind, rainfall, moisture, and the pressure gradient are referred to according to different focus of studies. In this study, we specify the monsoon flow as a continuous strong southerly, similar to studies using the meridional wind as a monsoon intensity index [Zeng et al., 2012]. First, based on the daily mean wind at 850 hPa, the strong southerly is identified as either an obvious southerly (v ≥ 4 m s− 1) over more than 50% of all grid points in South China or a longitudinal band of intense southerly (v ≥ 8 m s− 1). Strong southerlies that last for at least 2 days are then registered as active monsoon days, and the remaining summer days are registered as inactive monsoon days. A few cases of southerlies due to landfall tropical cyclones are treated as inactive monsoon days. There are 951 (1073) summer days in 1989–2010, which are categorized as active (inactive) monsoon days, accounting for 47% (53%) of a total 2024 summer days. The averaged number of active monsoon days per month for June, July, and August is 16, 20, and 7, respectively. Those in June and July mainly come from the Meiyu period that lasts from mid-June to mid-July. This is consistent with the rainfall-based criterion that defines a Meiyu period of approximately 1 month over Central China [e.g., Chen et al., 2004]. Active monsoon days are less frequent in August, corresponding to weak southerlies and a short-period monsoon revival [Chen et al., 2004; Zeng et al., 2012]. Most of the active monsoon days come from continuous periods of 5 days or longer, which correspond to the persistent features of monsoon activities. Through comparing the groups of active and inactive monsoon days, rather than presenting a summer mean, we locate the monsoon diurnal cycle and explore its relative contribution to the summer climate.

3 General Characteristics of the Summer Monsoon

3.1 A Mean of Low-Level Wind and Rainfall Amount

[12] During active monsoon days, strong low-level southwesterlies prevail over South China (Figure 1a). As the wind speed declines northward, it exhibits a convergence feature over Central China, including the Sichuan Basin (104°–109°E, 28°–32°N) and the East China Plain (112°–120°E, 27°–35°N). Accordingly, a rainband stretches zonally from the Tibetan Plateau foothills, through the Sichuan Basin, and East China Plain to the western coasts of Korea and Japan. The mean rainrate over Central China reaches 8.5 mm/day. The accumulated monsoon rainfall (365 mm) accounts for two thirds of the summer amount (576 mm), highlighting the importance of the monsoon in delivering the seasonal rainfall budget. Rainfall is also evident near the mountain ranges and coasts of South China, which may be induced by the southwesterly monsoon flow blowing against these terrains [Xie et al., 2006].

[13] During inactive monsoon days, the monsoon flow is absent over eastern China and resides in the low latitudes (Figure 1b). Rainfall is suppressed over most of eastern China. In particular, the rainband in Central China vanishes as the rainrate declines to 4.3 mm/day. Strong rainfall is seen instead along tropical island shores or over South China coasts, arising from the activities of the tropical monsoon trough and tropical cyclones [e.g., Ding, 1992] as well as the land-sea breeze [Nitta and Sekine, 1994].

3.2 Diurnal Cycle of Monsoon Rainfall

[14] To illustrate the diurnal cycle, we express the TRMM rainfall during A.M. hours as a percentage of the daily mean (Figure 2). In general, late night-morning rainfall occurs mainly over the oceans particularly offshores and in valleys and basins. In contrast, the afternoon-evening rainfall dominates over elevated plateaus, mountains, and tropical islands. Such a terrain dependency may result from convective activities that are intensely modulated by local land-sea or mountain-valley breezes [Johnson et al., 1993; Ohsawa et al., 2001]. During active monsoon days, the A.M. rainfall is observed on the east lees of the Tibetan and Yun-gui Plateaus and over the East China Plain (Figure 2a). The A.M. rainfall can account for up to 70% of the daily mean amount over the Sichuan Basin and ~50% over the East China Plain. This regional morning-hour rainfall is distinct from the afternoon-hour rainfall, which usually dominates inland during the warm season. During inactive monsoon days, the A.M. rainfall is confined to the plateau peripheries, such as the Sichuan Basin (Figure 2b). The percentage of A.M. rainfall also declines to ~40% over the East China Plain, while the amount of P.M. rainfall rises to 60% and 1.5 times the amount of AM rainfall. The spatial pattern of the diurnal cycle shows a clear localized feature in Figure 2b, in contrast to a smooth regionality in Figure 2a.

Figure 2.

The percentage of TRMM rainfall in AM hours to the daily mean on (a) active and (b) inactive monsoon days. Figure (c) denotes the percentage difference between Figures 2a and 2b. Long dashes mark the elevations of 1500 m and 3000 m.

[15] Figure 2c shows the difference in rainfall proportion between Figures 2a and 2b. The proportion of AM rainfall increases between inactive and active monsoon days, along the summer rainband over Central China. The increase ratio is 10–15% on the east lees of the Tibetan and Yun-gui Plateaus and reaches up to 10%–20% over the East China Plain. This indicates a large change in the diurnal cycle, as the increase in the proportion of AM rainfall means a corresponding decrease in the proportion of PM rainfall. For example, an increase of 10% may change the ratio of A.M. to P.M. rainfall from 50 : 50 to 60 : 40; i.e., the AM rainfall becomes 1.5 times the amount of the PM rainfall. Therefore, the morning signature of summer rainfall over Central China is attributed to the active monsoon, which brings a large amount of rainfall (Figure 1a) with an increasing proportion in the morning hours (Figure 2c). In contrast, offshore of South China, the A.M. rainfall proportion declines by 5%–10% on active monsoon days, displaying the different effect of the regional monsoon in an upstream region.

[16] Morning rainfall in the monsoon season over eastern China has been reported in recent studies of monthly analysis [Yin et al., 2009; Chen et al., 2009a; Yuan et al., 2010]. We further characterized such morning rainfall to specific active monsoon days in the above statistics, by which the importance of the active monsoon days on the summer rainfall budget has been evaluated in a quantitative manner. In the following sections, we investigate the moist processes that result to this phenomenon, with an emphasis on the monsoon diurnal cycle.

4 Summer Monsoon Diurnal Cycle Over Eastern China

4.1 Horizontal Structure of the Diurnally Oscillating Low-Level Wind

[17] We first examine the diurnal variation of the horizontal wind at low levels from the ERA-interim dataset. Figure 3 shows the diurnal veering of 850 hPa winds over eastern China. The daily mean has been removed to highlight the diurnal oscillation. Diurnal variations are found to be evident on land. The wind vectors undergo a clockwise rotation and behave like an inertial oscillation [Blackadar, 1957]. Over South China, the wind vector exhibits an intensified southerly at 02:00 LT and a southwesterly at 08:00 LT. In the mean flow of southwesterly, this represents an acceleration of wind speed during the night. The phase rotation is relatively fast to the north of 30°N, and this is probably due to a shorter inertial period of 24 h than that of ~28 h in South China. The veering leads to a deviated northerly, or northwesterly at 08:00 LT, which possibly combines with the strengthened southwesterly coming from the south to enhance the low-level convergence over Central China.

Figure 3.

Diurnal component of 850 hPa horizontal winds at four synoptic times on (a) active and (b) inactive monsoon days from the ERA-interim data. The daily mean at each grid point has been removed. The long dashes mark the elevations of 1500 m and 3000 m.

[18] Figure 3a shows that the diurnal range of low-level winds attains 2–3 m/s over South and Central China during active monsoon days, which accounts for at least one fifth of the average daily speed. The peak of wind speed may enhance the monsoon flow to form a LLJ at night. A wind streak of up to 12 m/s can be observed at 02:00/08:00 LT on 71% of active monsoon days, compared to only 47% at 14:00/20:00 LT. This appears to explain the frequent events of nocturnal LLJ over South China [Rife et al., 2010]. Figure 3b shows that the diurnal range, however, declines to 1–1.5 m/s during inactive monsoon days. It seems clear that the large diurnal wind variation is primarily integrated with the monsoon flow. As suggested by Chen et al. [2009b], such a meso-synoptic-scale phenomenon can be designated as the monsoon diurnal cycle over eastern China. It manifests a diurnal pulsation of regional-scale monsoon flow, compared to a continental-scale concept of the Asian monsoon circulation [Krishnamurti and Kishtawal, 2000].

4.2 Vertical Structure of the Wind Diurnal Oscillation

[19] After illustrating the horizontal feature, we now focus in this section on the vertical structure of the monsoon diurnal cycle. First, we take a detailed view of the profiles of the meridional wind over South China (105°–120°E, 21°–27°N) during 5–21 June 1998 (Figures 4a and 4b). It is evident that the monsoon flow is active on 13 out of the 16 days. The southerly maxima of 6–10 m/s occur regularly in the lower troposphere during the nights of active monsoon days, while the maxima are less visible on 5–6 June and on 9 June when the monsoon is absent. Figures 4a and 4b also show that both the daily and diurnal variations of the low-level wind in reanalysis data are highly similar to those in observations. The correlation coefficient between the two dataset reaches 0.93, which is above the level of statistical significance at 99.9% confidence. Such a similarity is also seen in the temperature and moisture fields (figure omitted). Although the ERA-interim dataset assimilates the soundings twice daily, it appears to be reliable in capturing the nature of the monsoon diurnal cycle over South China.

Figure 4.

Vertical profiles of the meridional wind over South China. (a) The SCSMEX six hourly radiosonde observations during 5–21 June 1998; (b) is the same as Figure 4a but for the ERA-interim dataset; (c) the composite six hourly profiles during 5–21 June 1998 from the SCSMEX observations, with the daily mean removed; (d) is the same as Figure 4c but for all of the active monsoon days in the summers of 1989–2010 by the ERA-interim dataset.

[20] The diurnal cycle is better illustrated using the composite six hourly profiles with the daily mean removed. The meridional wind profiles in Figure 4c show that the southerly is suppressed at 14:00 LT in the planetary boundary layer (PBL) at below 800 hPa and remains weak at 20:00 LT in the upper PBL near 850 hPa. The enhanced southerly appears at 02:00 LT in the PBL and lasts until 08:00 LT in the upper PBL. The wind oscillation thus exhibits a phase lag with height, and the diurnal range at 850 hPa attains 2.5 m/s. Figure 4d reveals that the largest amplitude at 850 hPa is also seen in the reanalysis profiles of active monsoon days. This maxima layer is higher than that which usually occurs at ~950 hPa in other weather regimes [Higgins et al., 1997; Parker et al., 2005]. The southerly wind tends to maximize at 02:00 LT, and the zonal wind maximum follows at 08:00 LT (figure omitted). Such a feature of the monsoon flow seems to be reflected in the predawn wind peak recorded over South China [e.g., Yu et al., 2009]. We note that the peak time in reanalysis is somewhat earlier than that of observations, as shown by the decaying profiles of 20:00 LT and 08:00 LT in Figure 4d. Other discrepancies between the two datasets are also seen in the oscillation of near-ground winds, probably arising from the different resolution of the data and the average period used. Although the wind oscillation is strongest in the PBL, it seems to extend upward into the middle–upper levels (Figures 4c and 4d); the possible mechanisms are discussed later. During inactive monsoon days, the wind profiles exhibit a similar phase as those in Figure 4d but have much smaller amplitude in the PBL (figure omitted).

[21] The vertical structure of the diurnal cycle is further illustrated in the pressure-longitude section of wind oscillation (Figure 5). At 20:00 LT, easterly anomalies dominate the mid-lower troposphere east of 100°E, while westerly anomalies appear at the upper level as return flow (Figure 5a). These indicate an establishment of mountain-plain and land-sea solenoids between the Tibetan Plateau, the lowlands of eastern China, and the ocean. Such solenoidal circulations extending deep in troposphere are commonly seen on the east lees of high terrains [Carbone and Tuttle, 2008; Bao et al., 2011; Jin et al., 2012]. With a span of thousands of kilometers, they are also recognized as continental-scale “land-sea” breezes [Dai and Deser, 1999]. They are driven by a pressure gradient force, as negative (positive) pressure anomalies occupy the Asian continent (Pacific ocean) in the mid-lower troposphere [Huang et al., 2010]. As the pressure pattern is present during both active and inactive monsoon days (figure omitted), solenoidal circulations seem to bring a summer background of wind diurnal oscillation over East Asia.

Figure 5.

Longitude-pressure section of the wind diurnal oscillations during active monsoon days in terms of (a) zonal wind at 20:00 LT and (b) meridional wind at 02:00 LT. The variables are averaged over a zone of 25.5°–27°N where monsoon flow enters Central China, with the daily mean of active monsoon days removed. The average topography of 25.5°–27°N is shaded in black.

[22] At 02:00 LT, the low-level wind then shifts to become a southerly, while the upper wind becomes a northerly as large-scale solenoids evolve (Figure 5b). The strongest southerly occurs in the PBL at a zone of 105°E–118°E and relates to a meso-synoptic-scale monsoon diurnal cycle. A regional mechanism creating this extra amplitude may involve the variation of PBL turbulent mixing that relaxes frictional drag at night and regulates wind inertial oscillation [Blackadar, 1957; Parker et al., 2005; Jiang et al., 2007]. The diurnal cycle of the PBL mixing effect is noted to have a particularly large amplitude during active monsoon days, as shown in the ERA-interim data (figure omitted), and appears to combine with the large-scale solenoids, to give a strong diurnal variation of monsoon flow. Future estimates using observations or modeling at a higher spatiotemporal resolution may offer more insights into the formation physics of monsoon diurnal cycle.

5 Impacts on Moisture Transport and Water Budget

5.1 Diurnal Cycle of the Moisture Transport

[23] The summer monsoon plays an important role in transporting moisture to eastern China [e.g., Ding, 1992; Zhou and Yu, 2005]. On a short time scale, moisture transport may intensify at night due to the diurnally varying low-level winds [Li et al., 2007; Yamada et al., 2007; Chen et al., 2009b; Chen et al., 2010; Yuan et al., 2010]. That may cause a moisture sink over eastern China and regulate the water budget on a diurnal time scale in summer [e.g., Chen, 2006]. From the angle of monsoon diurnal cycle, we propose two questions: First, “How much does the monsoon diurnal cycle explain the diurnal amplitude of moisture transport?” and second, “To what extent does it contribute to the regional water budget?”

[24] To estimate the moisture transport and water budget, we employ the governing equation of vertically integrated precipitable water [Trenberth and Guillemot, 1995]:

display math(1)

in which the precipitable water w and water vapor flux Q are given by

display math(2)

and E, P, V, q, g, p, and psfc are evaporation, precipitation, horizontal velocity vector, specific humidity, gravity, pressure, and surface pressure, respectively. For consistency, all these variables are given by the ERA-interim data. It is noted that the diurnal cycle pattern of precipitation forecast from the ERA-interim (figure omitted) is similar to the observed one shown in Figure 2c. Water vapor resides primarily in the lower troposphere and is largely transported by low-level winds [Chen, 2006]. As moisture fluxes converge (∇ ⋅ Q < 0) toward the region of moisture sink (E − P < 0), a link between the monsoon flow and the regional hydrological cycle is measured.

[25] First, we examined the daily mean of the monsoon moisture fluxes and then clarified the relative magnitude of the diurnal cycle. Figure 6a shows that on active monsoon days, the column-integrated moisture fluxes onto eastern China originate mainly from South China and the South China Sea. These fluxes have a strength ~350 kg m− 1 s− 1 over South China and account for up to 80% of the northward transport of summer moisture. More than three quarters of the fluxes come from the lower-tropospheric transport below 700 hPa. The largest transport occurs at around 110°E where the monsoon flow is the strongest.

Figure 6.

Water vapor fluxes during active monsoon days in terms of (a) the daily mean and (b) the diurnal component in the lower troposphere. The contour in Figure 6a denotes the percentage of the lower-tropospheric fluxes to the total fluxes. Long dashes mark the elevations of 1500 m and 3000 m.

[26] Figure 6b shows the large diurnal variations of low-level moisture fluxes seen over eastern China. The deviated fluxes are directed northward at 02:00 LT and eastward at 08:00 LT, and present as enhanced moisture transport through the night. The diurnal range is estimated at ~70 kg m− 1 s− 1, about one fifth of the daily mean. This indicates that during active monsoon days, the moisture transport is strengthened at night by about 20% of that occurring during the day. Note that a diurnal range of specific humidity is ~5% of the daily mean, which is smaller than that of wind oscillation (~20%). Diurnal variation of the moisture fluxes thus results mostly from the oscillation of the wind, as also indicated by the similarity between Figures 3a and 6b. During inactive monsoon days, the diurnal amplitude of the moisture fluxes becomes much smaller (figure omitted), due to a weak wind variation and moisture deficit.

5.2 Diurnal Variation of the Hydrological Processes Contributing to the Water Budget

[27] To illustrate the regional effect of the monsoon diurnal cycle, we estimated the hydrological processes contributing to the water budget over Central China (Figure 7). We divided the water amount from the low-tropospheric moisture fluxes below 700 hPa by an area size of Central China; such that it was normalized to a unit of 1 m2. The regional mean of water vapor storage, evaporation, and precipitation was also estimated using equation ((1)), while the contribution of cloud water was neglected. It should be noted that a balance among these terms may not be reached at a short time interval. Although the six hourly snapshots in Figure 7 may not contain the conservation of water mass, they offer us an instantaneous view of the diurnal variation of hydrological processes in the context of the monsoon diurnal cycle.

Figure 7.

Diurnal variations of the water budget variables over Central China during active monsoon days. Moisture fluxes are normalized by the area size of Central China and plotted as the vectors at four boundaries. Their net value is labeled at the bottom right of each box. The atmospheric storage of water vapor is enclosed by a diamond. Evaporation (E) and precipitation (P) are marked in brackets. All variables are given a unit of kg/m2 (equivalently, mm/m2) per 6 h.

[28] Figure 7 shows a strong moisture inflow into the southern boundary of Central China during active monsoon days. This southerly inflow is in its suppressed phase at 14:00 LT, while both evaporation and precipitation are active over Central China (Figure 7a). The 6 h sum of evaporation and moisture flux convergence reaches 3.23 kg/m2 and exceeds the precipitation of 2.65 kg/m2. Accordingly, the storage of atmospheric moisture increases by ~0.5 kg/m2 from 14:00 LT to 20:00 LT (Figures 7a and 7b). At 20:00 LT, the fluxes at the western boundary switch from being inflow to outflow (Figure 7b), as the upslope wind blows toward the plateaus. An enhanced outflow is also observed at the northern boundary, while a reduced outflow is seen at the eastern boundary. The resulting net moisture flux convergence over Central China declines from 1.00 kg/m2 at 14:00 LT to 0.48 kg/m2 at 20:00 LT; precipitation decreases from 2.65 to 1.22 kg/m2, respectively.

[29] At 02:00 LT, the inflow at the southern boundary increases to 3.94 kg/m2 (Figure 7c), due to the speed-up of the monsoon flow. An inflow also appears at the western boundary, arising from a reversal of the mountain-plain solenoid. The moisture flux convergence increases to 1.01 kg/m2, suggesting an enhancement of moisture convergence over Central China. This leads to a growth of precipitation from 1.22 to 1.94 kg/m2 in past 6 h. The precipitation fallout and decayed evaporation decrease the moisture storage at night. At 08:00 LT, there is a slight decline of the inflows at the southern and western boundaries, while the outflow at the northern boundary declines from 0.98 to 0.59 kg/m2 (Figure 7d). As a result, the moisture flux convergence remains as strong as 1.16 kg/m2. It combines with a resumed evaporation to support the precipitation, which increases to 2.57 kg/m2 during the morning.

[30] An overview of Figure 7 shows that the moisture supply from advection is primarily a result of the southerly monsoon inflow and partly a result of the westerly inflow. The accumulated daily supply is 15.23 kg/m2 per day, 4 times that of evaporation which is 3.6 kg/m2 per day. The variation in the southerly inflow is induced by the monsoon diurnal cycle, while that of westerly one is mainly due to the mountain-plain solenoid. Both two inflows are suppressed at 20:00 LT and result in a weak convergence of moisture fluxes. They become active at 02:00/08:00 LT to generate an enhancing convergence. A transition then occurs at 14:00 LT, with a small decrease in the convergence from 08:00 LT related to the decayed southerly monsoon inflow but strong westerly inflow. On inactive monsoon days, the moisture supplied by the southerly inflow declines to 2.79 kg/m2 per day (figure omitted). In particular, the moisture flux convergence at 02:00/08:00 LT is weak at ~0.3 kg/m2, which is much smaller than an amount of ~1.0 kg/m2 during the morning hours of active monsoon days. It seems clear that monsoon diurnal cycle combines with large-scale solenoids to play a crucial role in regulating the summer water budget over Central China.

5.3 Spatial Pattern of the Moisture Flux Convergence and Moisture Content

[31] Section 5.2 emphasized the importance of the moisture flux convergence affecting regional water budget. To further clarify the spatial pattern of convergence, we split the lower-tropospheric moisture fluxes into two parts: stream function and velocity potential. We focused on the velocity potential component, because it relates closely to the moisture flux divergence and convergence. The summer daily mean at each grid point was removed to show the diurnal oscillation on active monsoon days.

[32] Figure 8a shows that at 14:00 LT, moisture fluxes converge toward the coasts of South China and tropical islands, where convection develops in the early afternoon [Ohsawa et al., 2001]. At 20:00 LT, moisture fluxes diverge over the western Pacific and China coast, while they converge toward the Tibetan and Yun-gui Plateaus (Figure 8b). This suggests that the large-scale sea breezes not only transport moisture to terrain as reported by previous studies [Huang et al., 2010; Yuan et al., 2012] but also play a key role in regulating the moisture divergence and convergence. Such patterns appear to collocate with the convective activities dominating the plateau slope in early evening but which are suppressed over the ocean or even dissipating over the eastern China coast [Asai et al., 1998; Hirose and Nakamura, 2005; Chen et al., 2009a].

Figure 8.

Six hourly anomalies of the velocity potential of low-level moisture fluxes during active monsoon days. The daily mean of summer days has been removed at each grid point. The contour interval is 0.4 × 107 g∙kg−1m2s−1.

[33] At 02:00 LT, the moisture flux divergence is located over the tropical regions (Figure 8c). The convergence occurs on the plateau foothills, the Central China, and the Korean coast, which corresponds with the nocturnal rainfall that forms over low-lying areas (Figure 2a). Figures 8b and 8c also show that the moisture divergence and convergence exhibits as an east-west dipole pattern at 20 LT and rotates clockwise to a south-north pattern at 02:00 LT. This is thought to result from the diurnal veering of the low-level wind, as shown in Figures 3a and 6b. Similar to studies on a continental-scale mode [Krishnamurti and Kishtawal, 2000; Chakraborty and Krishnamurti, 2008], this depicts a poleward moisture discharge at night driven by the diurnal pulsation of the monsoon flow on a regional scale.

[34] At 08:00 LT, the moisture fluxes turn to converge toward Central China, particularly over the East China Plain (Figure 8d). This convergence feature is associated with the deviated southwesterly from South China and the northwesterly from North China. The intensity of the fluxes attains ~20 kg m− 1 s− 1 at 02:00/08:00 LT, amounting to about 50% of the anomalous moisture fluxes (cf. Figures 8c, 8d, and 6b). Such a meso-synoptic-scale convergence is shown to persist through the late night and morning. This corresponds to the largest increase of morning rainfall over the East China Plain during active monsoon days (Figure 2c). During inactive monsoon days, however, the morning flux convergence is suppressed over Central China and retreats to the South China coast (figure omitted). We can therefore clearly see that the primary contribution to the strong moisture sink on summer mornings over Central China comes from the monsoon diurnal cycle.

[35] The importance of the monsoon diurnal cycle compared to local breezes is further estimated in terms of the horizontal pattern of moisture content. We assume that the local breezes are shallow as indicated in Figure 4 and can be plotted by the near-ground winds at 975 hPa. We focus on mesoscale pattern, as the ERA-interim data are less able to resolve the very localized pattern than the in situ records (cf. Figures 4c and 4d) or the high-resolution data [Rife et al., 2010]. Figure 9a shows that the moisture content is concentrated at the foothills and offshores, where the local breezes are convergent, at 08:00 LT on inactive monsoon days. Such a terrain dependency implies a dominant effect of the local mountain-valley or land-sea breezes [Johnson et al., 1993]. Figure 9b illustrates that local breezes are still visible on active monsoon days; morning moisture however has changed notably from Figure 9a. Figure 9c shows the moisture difference between Figures 9a and 9b, highlighting the effect of the monsoon diurnal cycle. The moisture content is clearly enhanced over Central China but suppressed on the South China coast, due to nocturnal monsoon transport. Such a spatial pattern corresponds well with the change in morning rainfall (Figure 2c). Moreover, Figure 9c shows that the moisture anomaly induced by the monsoon diurnal cycle has a magnitude of O(1) kg/m2, which is comparable to that of the local breezes in Figure 9a. This anomaly looks strong enough to modify the regional diurnal cycle, and it even reverses the phase over the East China Plain and off the shores of South China.

Figure 9.

Diurnal components of lower-tropospheric water vapor content (shaded) and 975 hPa wind (vector) at 08:00 LT during (a) inactive and (b) active monsoon days. (c) The difference of water vapor content at 08 LT between active and inactive monsoon days. Long dashes mark the elevations of 1500 m and 3000 m.

6 Impacts on Precipitation Diurnal Variability

6.1 Rainfall Budget by the Various-Size Rain Events

[36] In this section, we examine the role of the population of rain events in the context of the rainfall diurnal cycle as a response to the monsoon diurnal cycle. In this respect, we use the 3B42 data to categorize the rain events according to the size of their area. We employ a concept of the contiguous rain area enclosed within a specified isohyet [Ebert and McBride, 2000]. An isohyet threshold of 1 mm/h is applied to mask out drizzle and so that isolated (organized) systems can be recorded as small (large) events. The rainfall volume of mesoscale events derived from this method is found to correlate well with the ground observations [Demaria et al., 2011]. The location of an event is indicated by the rain-weighted center of a rainy area. Using a 13 year archive of TRMM rain estimates enabled us to document the size, location, and rainfall volume of a huge population of summer rain events.

[37] Figures 10a shows the diurnal cycle of the rainfall budget over Central China on inactive monsoon days. Meso-α-scale rain events which have an area size of at least 2 × 104 km2 are capable of producing 143 mm of rainfall in one summer. These rain events explain 77% of the total rainfall amount of the moderate to intense rainrate (≥1 mm/h) on inactive monsoon days. The largest contribution is from rain events with an area size of ~105 km2, in relation to mesoscale organized convection. The resulting rainfall diurnal cycle exhibits a major peak at 17:00 LT and a secondary peak at 08:00 LT. On the other hand, the small–medium events with an area less than 2 × 104 km2 yield a smaller rainfall amount of 43 mm and display a single peak in the afternoon.

Figure 10.

Three hourly 3B42 rainfall amount averaged over Central China, accumulated from the rain events with respect to area-size categories during (a) inactive and (b) active monsoon days for a summer mean between 1998 and 2010. The triangle marks a size threshold of 2 × 104 km2 that separates small–medium and meso-α-scale rain events. This threshold approximates a horizontal scale of ~150 km, which is slightly smaller than the widely accepted 200 km. The time duration of rain events is not considered.

[38] Figure 10b shows that the rainfall produced by small–medium events on active monsoon days remains as weak as on inactive monsoon days. The rainfall from meso-α-scale events, however, increases remarkably on active monsoon days. The accumulated rainfall amount of meso-α-scale events in one summer attains 254 mm, with 134 mm recorded in A.M. hours and 120 mm in P.M. hours. It is particularly interesting to note that the rainfall peak at 05:00–08:00 LT becomes dominant and exceeds the peak of 17:00 LT. Such a plenty of morning-peak rainrate delivers a major source of the warm-season morning rainfall. The largest amount comes from the rain events with an area of ~2 × 105 km2 (i.e., the size is twice of those of inactive monsoon days).

[39] Nighttime rain events over eastern China have been linked to the propagating organized convections that often reach maturity in the late night and early morning [e.g., Asai et al., 1998; Wang et al., 2004; Chen et al., 2012a]. In many previous studies, these rain events are observed as active alongside the large diurnal variation of low-level winds. Their distinct activities during active monsoon days, as revealed in this statistics, highlight a strong connection between the mesoscale convection and the monsoon diurnal cycle.

6.2 Spatial Distribution of Meso-α-Scale Rain Events

[40] As meso-α-scale rain events respond to the monsoon, their spatial distribution is examined to clarify the regionality of the rainfall diurnal cycle. Figures 11a and 11b show that on inactive monsoon days, there are much fewer rain events over Central China than those over the coasts of South China and tropical islands, and there is a greater occurrence in the P.M. hours than in the A.M. hours over most of Central China, except over the plateau foothills. This explains the weak rainfall over Central China, which has a major afternoon peak on inactive monsoon days. The suppressed rain events in the A.M. hours may result from a moisture deficit, and a small diurnal variability that is less supportive of nocturnal mesoscale convection.

Figure 11.

Meso-α-scale rain events in (a) P.M. and (b) A.M. hours of inactive monsoon days, and (c) P.M. and (d) A.M. hours of active monsoon days. The dots denote the location of rain events. The density (shaded) is estimated as an occurrence within 150 km to any grid point and adjusted to one summer. Long dashes mark elevations of 1500 m and 3000 m.

[41] Figures 11c and 11d show that rain events over eastern China become more frequent on active monsoon days. In the P.M. hours, rain events mainly occur over the southeastern part of the Tibetan Plateau, South China, and the mountain ranges at 110°E (Figure 11c). In the A.M. hours, rain events usually occur to the east lees of the plateaus, the East China Plain, and the East China Sea (Figure 11d). Such a shifting between P.M. and A.M. hours suggests a strong dependency of the diurnal occurrences on major terrains, when they yield the double peaks of rainfall seen in Figure 10b. In particular, the active A.M. occurrence gives the maxima of morning rainfall along the summer rainband over the low-lying areas (Figures 11d and 2c). Meso-α-scale rain events are thus responsible for the regionality of the rainfall diurnal cycle on active monsoon days. On the other hand, these rain events possess a relatively large area and are likely to be of long duration. As rain events migrate, they may therefore cast a rain footprint over a wide region and result in a relatively smooth signature of the diurnal cycle (Figure 2a), which is distinct from the localized one during inactive monsoon days (Figure 2b).

[42] At the east lees of the Tibetan Plateau, the nocturnal rain events usually originate from the convective systems that form on the plateau slope near midnight and then move eastward [Asai et al., 1998; Wang et al., 2004; Sugimoto and Ueno, 2012], which is a similar phenomenon to those seen in North America [Carbone et al., 2002; Carbone and Tuttle, 2008]. These rain events are greatly supported by such features as the low-level convergence of the southwesterly monsoon [Chen et al., 2010; Ueno et al., 2011; Chen et al., 2012a], the rising motion of mountain-plain solenoid reversal [Bao et al., 2011; Jin et al., 2012], and the convective instability by moisture transport [Chen et al., 2009b; Yuan et al., 2012]. The propagation of the system is also facilitated by an enhancement of the vertical wind shear due to the strong low-level southwesterly [Wang et al., 2012]. Because of its large amplitude, the monsoon diurnal cycle appears to play an active role in the above mechanisms supporting convective initiation, growth, and migration at night. It not only increases moisture transport at night (Figures 6b and 9c) but also strengthens low-level convergence (Figures 8c and 8d) and vertical wind shear (Figure 4). In dosing so, the monsoon flow acts to promote the frequent nocturnal occurrence of self-sustaining convective systems which inherently reach maturity in the late night-early morning and result in an increasing amount of morning rainfall on active monsoon days.

[43] Over the East China Plain, the southwesterly monsoon flow may impinge on the Meiyu front to help produce a zone of moisture convergence, convective instability, and mesoscale lifting, in which nocturnal convection is promoted [Yamada et al., 2007; Chen et al., 2012a]. The local mountain-plain solenoid also helps to initiate convection, which can be enhanced by the nocturnal LLJ over the plain [Bao et al., 2011; Sun and Zhang, 2012]. As the East China Plain is located downstream of the wind diurnal veering, the strong moisture supply and the low-level convergence may persist through the late night and morning (Figures 6b, 8c, and 8d). In section 5, it was stressed that these persistent conditions are most evident in the presence of the monsoon diurnal cycle. As such conditions favor the nighttime growth of long-lived rainfall systems [Chen et al., 2009b; Chen et al., 2010; Yuan et al., 2010], they are expected to yield the largest increase of morning rainfall over the East China Plain during active monsoon days.

[44] It is well known that the summer monsoon undergoes a northward march from June to August. The monsoon-rainfall link on a diurnal time scale needs to be further evaluated with consideration of the summer progress. Figure 12 shows the monthly distribution of wind oscillation and morning rain events on the active monsoon days from June to August. The nocturnal enhanced southwesterly tends to extend northward, suggesting the progress of the monsoon diurnal cycle. The northward progress is most evident over the East China Plain. By August, the monsoon diurnal cycle can reach the northern part of Central China or the southern boundary of North China (~35°N). Correspondingly, the location of morning rain events shift northward over Central China. The morning events occur mainly over the low-lying areas within the monthly rainband, toward which the nocturnal monsoon flow is converging. An increasing occurrence is also observed over North China in July and August, which can be supported by the nocturnal southwesterly [He and Zhang, 2010; Chen et al., 2012b]. It seems clear that the response of morning rain events to the monsoon diurnal cycle is established during the summer progress. This is thought to explain the morning-peak rainband that moves northward during the summer season [Yin et al., 2009; Chen et al., 2009a].

Figure 12.

Meso-α-scale rain events in the A.M. hours and 850 hPa deviated winds at 02:00 LT on active monsoon days of (a) June, (b) July, and (c) August. The dots denote the location of rain events with monthly density shaded. Wind vectors greater than 1 m/s are plotted. Long dashes mark the elevations of 1500 m and 3000 m.

7 Summary and Discussion

[45] Using the latest reanalysis data and satellite rain estimates, we estimated the hydrological impacts of the monsoon diurnal cycle over eastern China. We highlighted that the summer monsoon flow becomes more efficient at night in transporting moisture and promoting morning rainfall and that this greatly affects the warm-season weather and climate. The findings are summarized as follows:

  1. [46] The diurnal variation of low-level wind is found to be evident over South China during active monsoon days. The largest diurnal amplitude typically occurs at 850 hPa where the monsoon flow prevails, as consistent with radiosonde observations. The wind speed attains a maximum in the early morning and a minimum in the afternoon. The diurnal range attains a magnitude of 2–3 m/s, which accounts for one fifth of the daily mean wind speed and which may strengthen the southwesterly monsoon to form the frequent events of the nocturnal LLJ over South China. This can be viewed as the meso-synoptic-scale diurnal pulsation of the summer monsoon that develops in the lower troposphere over eastern China. In contrast, the diurnal range of low-level wind declines to 1–1.5 m/s during inactive monsoon days.

  2. [47] Moist air is advected poleward by the summer monsoon over eastern China. Due to the monsoon diurnal cycle, the moisture transport over South China is strengthened by about 20% at night. Nocturnal moisture fluxes exhibit a strong convergence at their northern terminus, which leads to a meso-synoptic-scale moisture sink over Central China through the late night and morning. Such features express the diurnal pulsation of the summer monsoon, which drives the moisture discharge from South China to Central China and thereby regulates the regional water budget on a diurnal time scale. The nocturnal monsoon flow also combines with the local breezes to strengthen moisture over the basins and plains of Central China in the morning, while reducing moisture over the South China coast. Such impacts seem to explain the evident morning rainfall over Central China and the suppressed rainfall over the South China coast during active monsoon days.

  3. [48] The mean rainrate over Central China reaches 8.5 mm/day on active monsoon days, twice that on inactive monsoon days. What is of great interest is that the proportion of morning rainfall increases by 10% or more on active monsoon days. The abundant morning rainfall greatly contributes to the summer diurnal cycle. Statistical analysis reveals that the majority of summer rainfall comes from the meso-α-scale rain events relating to mesoscale organized convection. These rain events not only become more pronounced on active monsoon days but also exhibit a dominant morning peak that exceeds the afternoon peak, in a contrast to those on inactive monsoon days. Such morning rain events generally occur on the low-lying areas of Central China along the summer rainband, where the monsoon diurnal cycle and related moist processes promote an increased nocturnal occurrence of mesoscale organized convections that generally reach maturity during the late night-early morning. The increased occurrences of morning events gradually shift northward from June to August, in response to the progress of the summer monsoon and its corresponding diurnal cycle.

[49] More effort is needed to explore the physical link between the summer monsoon and the precipitation diurnal variability. What is the importance of the formation dynamics underlying the monsoon diurnal cycle? An oscillation of pressure has been noted to induce the diurnal variation of low-level wind over the summer continent [Holton, 1967; Dai and Deser, 1999; Huang et al., 2010]. Other PBL processes such as diurnally varying friction may also regulate the wind inertial oscillation [Blackadar, 1957]. Jiang et al. [2007] suggested that diurnal oscillations of the pressure gradient force and that of vertical diffusion should be combined, in order to gain a realistic modeling of nocturnal LLJ over the Great Plains. Parker et al. [2005] found that the wind diurnal cycle is well established in the West African monsoon layer, where both the pressure gradient and the convective boundary layer are evident. It is still unclear how these physical processes are incorporated in driving the diurnal variation of low-level winds in the strong monsoon environment over eastern China.

[50] In our statistics, nocturnal mesoscale convection has been recognized as the most responsive system to monsoon diurnal cycle. The underlying mesoscale processes embedded in regional-scale monsoon flow should be regarded as a key issue. Because in situ observations and reanalysis data are coarse, it may be necessary to apply high-resolution modeling to resolve the detailed timing, intensity, and duration of mesoscale systems. To achieve this, numerical experiments need therefore to capture not only the LLJ activities but also regional forcings such as the Meiyu front, the daytime heating/moistening [Yamada et al., 2007], and the mountain-valley solenoid [Sun and Zhang, 2012]. As indicated by the current study, a reliable reproduction of the wind diurnal oscillation on both regional and local scales will be particularly meaningful for further investigation. Our ongoing works involves estimating the relative importance of the monsoon diurnal cycle in the genesis and development of nocturnal mesoscale convection, compared with other known forcings.

[51] The feedback of diurnal variability to variations on longer timescales also remains to be estimated over eastern China. Morning rainfall explains a key part of the seasonal budget [Yu et al., 2007; Yin et al., 2009] and a large proportion of the rainfall variance [Yuan et al., 2010; Chen et al., 2012a]. Morning rainfall has undergone a decreasing trend over the last four decades of weakening monsoon, contributing to a long-term decrease of annual rainfall in the northern region of Central China [Yin et al., 2011]. The long-term rainfall variations can be closely connected with those of the low-level wind gradient [Zeng et al., 2012]. Here we see that the wind gradient is most evident during the night of active monsoon days. This implies that a long-term trend of the monsoon diurnal cycle could function as an important climate forcing over eastern China. More studies will be able to shed further light on this interesting relationship that has valuable implications for the changes in the regional hydrological cycle, the energetics of summer monsoon, and the evaluation of climate model performance [Chakraborty and Krishnamurti, 2008].


[52] The authors are grateful to three anonymous reviewers for their constructive comments. Thanks also go to ECMWF for providing the reanalysis dataset and to GSFC and TSDIS of NASA for providing the satellite rainfall dataset.