The river discharges have decreased continuously during the last half century in the Yellow River, the second-largest river basin in China. In particular, a drying up of the main river along the lower reach has occurred since 1972, and the situation has become more and more serious during the 1990s. Using 50 years of meteorological data from 108 stations together with a collection of irrigation data, the long-term changes in the river discharge have been investigated with a view to identifying the reason for the drying up of the Yellow River. It was found that the annual precipitation generally decreased (−45.3 mm/50 yr) while the air temperature generally increased (+1.28°C/50 yr). From the 1960s to the 1970s the precipitation decreased by 29.6 mm/10 yr, the evaporation increased by 7 mm/10 yr (for pan evaporation), and the irrigation water usage increased by 10.5 mm/10 yr. As a consequence the drying up of the Yellow River has occurred since 1972. Irrigation was developed continuously in the 1980s, but the drying-up situation maintained at the same level as during the 1970s. The reason for this was the increase in precipitation (by 10.3 mm/10 yr) and the sharp decrease in the evaporation (by 133 mm/10 yr for pan evaporation). During the 1990s the irrigation was maintained at a level similar to that during the 1980s, but the drying-up situation was greatly aggravated. The reason for this was found to be a result of the decrease in precipitation (by 38.2 mm/10 yr) and the increase in evaporation (by 52 mm/10 yr for pan evaporation).
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Climate variability directly influences the availability of water resources. Global warming has been the most commonly discussed issue related to climate change over the past 2 decades. The influence of global warming has been identified by several researchers [Diaz, 1986; Jones et al., 1986; Yamamoto et al., 1986; Fu et al., 1999; Qian and Zhu, 2001], the two main warming periods of which occurred during the 1920s∼1940s and from the 1970s to the present. The Intergovernmental Panel on Climate Change (IPCC) has reported that the global mean surface air temperature has risen by about 0.3∼0.6°C because of human-induced greenhouse warming over the past 100 years [IPCC, 1996, 2001]. In addition to global warming, the rapid growth of the world population which started in 1950 has resulted in an estimated population of 6.1 billion in the year 2000, nearly two and a half times the population of 1950 [United Nations Population Division, 1999]. Increases in human activities, such as cultivation and deforestation, have also introduced a variability in the water cycle and subsequent changes to the amount of water available, in terms of both the spatial and the temporal distributions. The most commonly reported changes are decreases in the water resources in most regions and increases in the incidences of flooding in some parts of the world.
 The impacts of climate change on water resources have been widely discussed using the general circulation model's (GCM) projection on global warming [Arnell, 1999; Vorosmarty et al., 2000; Smith and Lazo, 2001]. However, climate fluctuations cannot be assessed because of the questionable accuracy of the GCM's projection. On the other hand, historical climate data are generally available and provide a reference for studying the impact of climate fluctuations on water resources. Trend analysis is commonly employed for this purpose [Burn and Hag Elnur, 2002; Zhang et al., 2001].
 During recent times, China has experienced an unusual climate. In particular, the continuously dry climate that was experienced in the 1990s resulted in adverse effects in the Yellow River basin, and the long-term heavy rainfall that occurred in 1998 in the Yangtze River basin caused large floods along the middle and lower reaches. The population of China has tripled during the last 50 years, reaching 1.3 billion at the present time. Under the pressure of an increasing population, the water crisis is becoming more and more serious in northern China in particular. Nearly one third of the population (about 381 million) is living under conditions of water scarcity. The Yellow River basin is one of the regions facing serious water shortages due to the dry climate and heavy water demands.
 Soil erosion and sedimentation, flooding, and water shortages are the major water-related issues associated with this basin. Because of the aggravation of the drying up in the 1990s, especially in 1997, the water shortage problem is becoming more and more serious [Wang and Fong, 2001]. Several recent studies have addressed the problem of water resources in the Yellow River from the point of view of understanding the reason for the drying up and improving the management of the river in the future [Cheng et al., 1998; Wen, 1998; Cheng et al., 1999]. He  analyzed the status of water resources and, in particular, the changes in the water resources over the past 20 years. Ren et al.  discussed the variation in the natural runoff and water utilization in three subcatchments. It was found that the natural runoff decreased, but the artificial water utilization ratio (defined as the ratio of river water intake to the naturally available river discharge) increased in the three study sub-basins during the last 50 years.
 On the basis of previous studies the present research attempts to investigate the water resources in a more quantitative way and, in particular, to examine the long-term changes in precipitation, air temperature, pan evaporation, and river discharge in order to identify the reason for the drying up of the Yellow River. In this research, 50 years of meteorological data and river discharge data have been collected. The spatial and seasonal characteristics of the precipitation, air temperature, pan evaporation, and river runoff have been examined. Together with a collection of irrigation data, changes in the river discharge and the reason for the drying up along the lower reaches have been analyzed.
2. Study Area
 The Yellow River, also called Huanghe in Chinese, is the second-longest (also the largest) river basin in China. It originates from the Tibetan plateau, wanders through the northern semiarid region, crosses the loess plateau, passes through the eastern plain, and finally discharges into the Bohai Gulf (see Figure 1). The Yellow River flows about 5500 km in distance in the main course and accumulates 753,000 km2 of drainage area. About 100 million people live within the catchment, and the catchment consists of 1200 million ha of farmland, of which nearly half is irrigated by the Yellow River.
 From the origins to the river mouth the Yellow River experiences three typical landforms, the Qingzhang (Tibet) high plateau with elevations from 2000 to 5000 m, the loess plateau and midstream tributaries with elevations from 500 to 2000 m, and the alluvial plain in the eastern part. The climate conditions vary from cold to temperate zones and change from arid and semiarid to semihumid regions [Cheng, 1996]. The main irrigation areas are located in the northern part, in tributaries of the midstream and on both sides along the lower reaches (see Figure 1). On the basis of the topography, climate, and water utilization, the whole of the basin can be characterized into six regions, as shown in Table 1.
Table 1. Regions With Different Characteristics of Hydrology and Water Uses
Located in the Tibetan plateau, one of the main source areas
Between the Lanzhou and Toudaoguai gauges
Located in the northern arid and semiarid regions, one of the main irrigated areas
Between the Toudaoguai and Longmen gauges
Located in the loess plateau, a semiarid area
Between the Longmen and Sanmenxia gauges
Includes several main tributaries in the middle reaches, one of the main irrigated areas
Between the Sanmenxia and Huayuankou gauges
Located in the mountainous area with heavy precipitation during the summer
Downstream of the Huayuankou gauge
Located in the alluvial plain, suspended river, one of the main water use areas, the Lijin is the final discharge gauge close to the sea
 Agricultural irrigation began more than 1000 years ago in the Yellow River basin. In ancient times, irrigation occurred mainly in the Weihe basin near Xi'an and downstream of the Qingtongxia dam (see Figure 1), but, needless to say, neither dams nor weirs existed at that time. Vast irrigation projects were developed after 1949, mainly during the period from the end of the 1950s to the beginning of the 1970s, in the upstream region of the Huayuankou gauge. In the 1970s and 1980s the irrigation areas were widely expanded to the outside of the basin along the lower reaches from the Huayuankou gauge. The area of irrigation has increased by nearly a factor of 10 during the last 50 years [Xi, 1996]. Large irrigation projects with areas of more than 20,000 ha are managed by the Irrigation Department of the Ministry of Water Resources. These large projects provide irrigation to 7.13 million ha at the present time, sharing about 72% of the total irrigation area [General Institute of Water Resources and Hydropower Planning and Design (GIWP), 2001]. The distribution of the main irrigation areas by the large projects is shown in Figure 1 and is also given in Table 1. It should be mentioned that most of the irrigation areas along the lower reaches (downstream of the Huayuankou gauge) are not located in the drainage basin because of the suspended river in the downstream area.
 Relative to the last 50 years of agricultural irrigation, the water was particularly badly misused by overirrigation during the 3 years from 1959 to 1961. Salinization became a serious problem in a large portion of the irrigated areas with poor drainage systems, especially in Inner Mongolia in the northernmost end of the basin and downstream of the Huayuankou gauge. Farmlands in these areas lost their productive capacity. For this reason, irrigation was even stopped in the Inner Mongolian region and in the downstream region from 1962. By improving the drainage system, irrigation in the salinized areas was gradually recovered from 1966. The main improvements in the drainage systems were made during the end of the 1970s and the 1980s [Xi, 1996]. With increasing pressure from the population, the irrigation areas continuously expanded, especially in the downstream region. Consequently, water shortages became more and more serious, and the main river along the lower reaches has dried up during the irrigation season since 1972. The duration of the drying up increased rapidly in the 1990s (see Figure 2). In the most serious situation encountered in 1997, the main river close to the sea dried up for 226 days, and the no flow distance reached 704 km from the river mouth. Regarding the flow condition in the main river along the lower reaches, three typical time periods can be identified during the last half century: the period from the 1950s to the 1960s with no drying-up events; the 1970s and the 1980s, during which time drying up was infrequent; and the 1990s, when a drying-up event occurred every year.
3. Data and Method
3.1. Data Used in the Study
 The climate data from 1951 to 2000 were obtained from the China Meteorological Administration. This data set is available at a daily temporal resolution at 108 meteorological stations inside and close to the study basin (see Figure 1). This data set contains information from all of the observation stations available at the national level. Several stations located in the Tibetan plateau began from 1959, and a few stations were shifted to nearby locations in the 1990s. The meteorological parameters include the precipitation, the mean air temperature, the minimum and maximum air temperature, wind speed, relative humidity, sunshine time, and pan evaporation.
 In order to calculate the basin average precipitation and temperature, a 10-km gridded data set covering the study area was interpolated from the data measured at the meteorological stations. A grid size of 10 km was selected for potential use for hydrological simulations in this basin. Precipitation is interpolated using an angular distance weighting method [New et al., 2000]. This method first involves selecting the eight nearest stations and calculating the distance weighting using
where x is the distance from the center of a grid to the station and x0 is the attenuation distance. Here eight stations are selected. The value of x0 is a variable with an upper limit of 300 km, and m is an adjustable parameter, taken as 4 as suggested by New et al. . This distance weighting is modified according to the relative direction among the meteorological stations. The correction coefficient is given as
where θ (k, l) is the angular separation of stations k and l and wl is the distance weighting of station l. The angular distance weighting is calculated using
Therefore the interpolated precipitation at a grid is calculated using the weighted average as
where Pint is the interpolated precipitation of the target grid and Pl,obs is the observed precipitation at station l. The temperature at a station is interpolated for each grid using an elevation-corrected angular direction weighting method. Using the temperature data from 1961 to 1990, the mean elevation factor is estimated for different regions and for each month.
 The basin average of the annual precipitation from 1951 to 2000 calculated from the interpolated 10-km gridded data is 440 mm. Using a global gridded data set of monthly precipitation at a resolution of 0.5° longitude/latitude [New et al., 1999, 2000] which has poor availability of meteorological stations and highly uneven distribution in the Yellow River basin, the mean annual precipitation from 1951 to 1998 (these data are available up to 1998) is estimated to be 419 mm. Since the same interpolation method is used, but the present study includes all measured data available in the national database, there is no doubt that the accuracy of the 10-km gridded data set generated in the present study is better than the global 0.5° data set. In addition, the Thiessen polygon method is used for comparison. It has been found that the basin average of the annual precipitation in the 50 years is 456 mm, which is slightly higher than that estimated by the angular distance weighting method. In order to obtain a spatially continuous gridded data set for hydrological modeling in future studies, considering the low density of meteorological stations over mountainous areas, in particular over the Tibetan plateau, the angular distance weighting method is adopted.
 Irrigation data, in particular that for the actual irrigation area, water intake, and water drainage, are usually difficult to collect. For the present study the data related to irrigation in the Yellow River basin were searched for from as wide as possible a range. The Yellow River Conservancy Commission (YRCC) made a comprehensive assessment of water resources in the Yellow River in 1985 on the basis of a 5-year (1980∼1985) study and used this as the basis for water resources planning and management during the 1980s and 1990s (YRCC , available at http://www.yrcc.gov.cn). The data related to irrigation include the actual irrigation area and river water consumption (the gross loss of the river discharge for irrigation) for two regions, consisting of the upstream and downstream areas of the Huayuankou gauge from 1949 to 1985. The river water consumption was estimated as the difference between the intake from the river and the return to the river by the drainage canal for the upper region but the same as the water intake for the lower region because the bed of the Yellow River is higher than the surface of the irrigation land in this region. Entering the 1990s, the YRCC started to publish an annual water resources bulletin for the Yellow River [YRCC, 1991–2000]. It contains the annual discharge at the six gauges (see Figure 1), the river water intakes for irrigation, and the water losses for the major irrigation areas.
 The daily discharge data before 1990 were collected from the Hydrological Year Book published by the Hydrological Bureau of the Ministry of Water Resources of China [Information Center of Water Resources (ICW), 1950–1990]. However, no yearbooks were published after 1990. The monthly discharge data after 1990 is documented in an annual report by the Hydrological Bureau and is available at the Web site of the Ministry of Water Resources of China (http://www.hydroinfo.gov.cn/zyysq/index.htm). In this study, six discharge gauges are selected (see Table 2).
Table 2. Discharge Gauges Used in This Study
Drainage area, km2
Distance to the river mouth, km
3.2. Trend Detection
 The Mann-Kendall nonparametric test has been recommended as an excellent tool for trend detection [Zhang et al., 2001; Burn and Hag Elnur, 2002]. The present study applies this test for detecting the significance of the trends in annual and seasonal meteorological and hydrological time series. The magnitude of the trend for a time series data set is estimated by Burn and Hag Elnur :
where β is the trend magnitude. A positive value of β indicates an increasing trend, and a negative value of β indicates a decreasing trend.
 One problem associated with the Mann-Kendall test is that the result is affected by serial correlations of the time series [Zhang et al., 2001]. The presence of a serial correlation can complicate the identification of trends in that a positive serial correlation can increase the expected number of false positive outcomes for the test [Burn and Hag Elnur, 2002]. To eliminate the effect of serial correlation, one common prewhitening approach [Zhang et al., 2001; Burn and Hag Elnur, 2002] is adopted to remove the serial correlation from the data set before applying the Mann-Kendall test.
 The prewhitening approach consists of the following procedure: (1) Calculate the lag 1 serial correlation r; (2) apply the Mann-Kendall test directly to the original data series if r < 0.1; otherwise, employ the prewhitening as
where is the prewhitened value at time t and xt is the original value at time t, and (3) apply the Mann-Kendall test to the prewhitened time series [Zhang et al., 2001; Burn and Hag Elnur, 2002]. This study uses a 5% significance level in the trend analysis.
3.3. Irrigation Water
 The crop water demand depends on the crop type, the planting calendar, and the climate conditions. The main crops in the Yellow River basin include winter wheat, spring wheat, and summer corn. In addition, soybean and cotton are also popular crops for planting in the early summer. Planting starts from the lower basin and then shifts to the middle and upper streams over time. The crop water demand is calculated following the Food and Agriculture Organization (FAO) guidelines [Allen et al., 1998]. In optimal agronomic conditions the daily crop water demand Dirr is calculated using
where ET0 is the reference evapotranspiration and kc is the crop coefficient which changes with species and growth of the crop. The reference evapotranspiration is estimated using the FAO Penman-Monteith equation [Allen et al., 1998] based on the daily air temperature, wind speed, humidity, and sunshine hours, and, as a result of a lack of data, no soil heat flux is assumed. The irrigation consumption (or consumption of irrigation water) is defined as the additional water loss to the atmosphere by evapotranspiration due to increasing soil moisture by irrigation. It is calculated as the difference in the crop water demand subtracting the effective precipitation.
 Irrigation uses water from the Yellow River on the basis of the ease of the water intake and the water availability at a particular time rather than on the basis of crop water requirements. Geographically easy areas for the irrigation water intake include the area downstream of the Qingtongxia dam, Inner Mongolia, the area downstream of the Weihe tributary, and the area downstream of the Huayuankou gauge of the main river. However, irrigation along the lower reaches is limited by the water availability. With respect to the use of the river discharge, the gross loss for irrigation includes the losses in canals and leakages due to groundwater in the field and can be estimated as the difference between the intake from the river and the return to the river. The return flow to the river can be from the surface drainage system and from the groundwater. The data used in the present study contain only the return flow from the surface drainage system. This might lead to an overestimation in the irrigation loss. The return flow can be used for water resources in the downstream region, but pollution from farmland introduces water quality problems to the downstream region.
 The ratio of irrigation consumption to the irrigation gross loss can be used for checking the physical reliability of the irrigation data and for evaluating the efficiency of the irrigation system. In the present study, the mean irrigation consumption in the 1990s is estimated for the upstream region of the Huayuankou gauge. Comparing this with the gross loss of river discharge estimated from the water intake and surface drainage discharge, the efficiency of the present irrigation system in the Yellow River basin is discussed in section 4.
4. Results and Discussion
4.1. Spatial Distribution and Seasonal Characteristics
 The spatial pattern of the average annual precipitation over the past half century is examined first. Figure 3a shows the spatial distribution of the annual precipitation over the upstream region of the Huayuankou gauge. The annual precipitation ranges from 154 to 764 mm and increases from north to south and from west to east. The basin average annual precipitation is 440 mm. The monthly air temperature and precipitation show similar patterns consisting of peaks in July and troughs in January or December. Regarding the precipitation, the wet season is from June to September, which share nearly 70% of the annual total. The streamflow is delayed by natural and artificial hydrological processes; therefore the wet season starts from July to October with respect to the streamflow and consists of about 62% of the annual total runoff. From the viewpoint of water utilization a year can be divided into a wet season from July to October and a dry season from November to June. Compared with the high spatial variability of the annual amount shown in Figure 3a, the concentration of precipitation in the wet season from July to October (see Figure 3b) is more spatially homogeneous. This concentration rate ranges from 60% to 73% and has an average of 65% over the whole of the basin, and it increases from south to north. The high concentration corresponds to less annual precipitation (Figure 3b).
Figure 4 plots the accumulation of the annual discharge from the upstream to the downstream region. This is in contrast with the accumulated drainage areas which show the general spatial pattern of the annual discharge. It is found that the annual discharge does not increase in proportion to the drainage area, as the basins are located in a humid region. The reason for this is the spatial variability of the climate conditions ranging from semiarid to semihumid. This is also related to artificial water uses, in particular for irrigation, as large irrigation areas can be seen in Figure 1. The annual discharge at the Lijin gauge close to the sea is slightly higher than the annual discharge at the Lanzhou gauge with only 30% of the total basin area. This means that the middle and the downstream areas between the Lanzhou and Lijin gauges generate very limited net runoff. This implies that a huge amount of artificial water use exists in this region in reference to the spatial distribution of the annual precipitation shown in Figure 3a. The regions where the annual runoff decreases have a net consumption of water resources.
Figure 5 shows the characteristics of precipitation, net runoff, and irrigation area in the regions identified in Table 1. The water depths of the annual precipitation and net runoff (net runoff equals runoff minus water use) are calculated over the drainage areas of each region. The major source areas that contribute runoff to the main river are found to be the upstream region of the Lanzhou gauge (region I) and the region between the Sanmenxia and Huayuankou gauges (region V). Two pure water consumption regions are between the Lanzhou and Toudaoguai gauges (region II), due to the dry climate and an extensive amount of irrigation, and in the downstream region of the Huayuankou gauge (region VI), due to the extensive amount of irrigation and the suspended river.
Table 3 summarizes the regional and seasonal characteristics of precipitation and runoff. The annual runoff discharging into the sea (viewed as the annual discharge at the Lijin gauge) is 34.6 billion m3 (46 mm), and most of it is concentrated in the wet season. Regarding the seasonal characteristics of runoff, the upstream area of the Lanzhou gauge (region I, the major source area) has a better even distribution of runoff over a year. But the net runoffs generated from other regions are mainly concentrated in the wet season from July to October. About 70% of the annual precipitation in region V (between the Sanmenxia and Huayuankou gauges) falls during the wet season from July to October, which generates frequent floods. Regarding the runoff consumption, region VI (downstream of the Huayuankou gauge, one of the major water consumption areas) intakes river water mainly during the dry season, but region II (see Table 1 and Figure 1) consumes nearly an equal amount of water during the two seasons because of its dry climate.
Table 3. Regional and Seasonal Characteristics of Precipitation and Runoff
Regions are the same as those identified in Table 1.
Net runoff after water use is calculated as the difference in the discharges at the two ends of a region. For the upstream area of the Lanzhou gauge (region I) and the whole of the basin it is the same as the discharge at the Lanzhou gauge and the Lijin gauge, respectively.
Value in parentheses is the water depth averaged over the drainage area, in mm.
Precipitation is calculated over the upstream area of the Huayuankou gauge on the main river.
4.2. Trends in Climate and Runoff During the Last 50 Years
Figures 6a and 6b show the basin averages of annual precipitation and annual mean temperature over the past 50 years, respectively. The annual precipitation shows a decreasing trend, −45.3 mm/50 yr. The annual mean temperature shows a clear increasing trend, +1.28°C/50 yr, especially during the 1990s. Figure 6c shows the annual pan evaporation that is calculated as the average of the 81 meteorological stations within the Yellow River basin. The annual pan evaporation has no single trend during the last 50 years. It decreased from the beginning of the 1970s to the middle of the 1980s but increased during the 1990s. Figure 6d shows the annual runoff observed at six gauges during the last half century. A common decreasing trend can be found in the range of −28 mm/50 yr ∼ 61.5 mm/50 yr, and this decreasing trend becomes stronger from the upstream to the downstream area.
 The trends in monthly precipitation and air temperature for the 108 meteorological stations are detected (see Figure 1). The precipitations show decreasing trends in most stations and for most months, but most are not significant. The air temperatures have increasing trends in most stations and for most months. There are also nonsignificant trends in many cases. However, it is found that air temperature during the winter, especially in December, has a significant increasing trend in many of the stations.
4.3. Decadal Variations in Climate and River Discharge
 In order to understand the long-term variability in climate and river discharge, decadal averages of the annual precipitation, annual mean temperature, annual pan evaporation, and river discharges are examined. The upper basin (upstream of the Lanzhou gauge) and the whole of the basin are considered. For both regions, as shown in Table 4, the annual precipitation and annual runoff have the same general decreasing trend and follow the same pattern of a dry-wet-dry-wet-dry cycle from the 1950s to the 1990s; however, the annual mean temperature increases continuously. The precipitation and discharge of the upper basin show less change compared with the whole of the basin. During the 5 decades, for both regions the precipitation, temperature, and runoff show their smallest changes between the 1970s and the 1980s but their biggest changes from the 1980s to the 1990s. The pan evaporation shows different decadal changes over the upper and the whole of the basin. It decreases from the 1950s to the 1980s but increases during the 1990s over the upper basin. In a similar pattern the ratio of the annual pan evaporation to the annual precipitation in the upper basin ranges from 2.77 to 3.58 (see Table 4). For the whole of the basin the annual pan evaporation increases continuously from the 1950s to the 1970s but decreases dramatically in the 1980s and increases again in the 1990s. The ratio of the annual pan evaporation to the annual precipitation ranges from 3.65 to 4.12 (see Table 4), which follows the same decadal variations as the annual precipitation and runoff.
Table 4. Decadal Averages of River Discharge, Precipitation, and Temperaturea
Q is annual river discharge, P is annual precipitation, T is annual mean temperature, and Epan is annual pan evaporation estimated from 81 stations located inside the Yellow River basin.
Precipitation is averaged over the upstream area of the Lanzhou gauge, and the discharge is at the Lanzhou gauge.
Precipitation is averaged over the upstream area of the Huayuankou gauge, and the discharge is at the Huayuankou gauge.
Value in parentheses is the ratio of the annual pan evaporation to the annual precipitation, which shows the dryness of the climate.
 Regarding the artificial water uses, the decadal averages of the gross loss of the river discharge are estimated (see Table 5). It is known that irrigation water uses increased significantly during the 1970s and 1980s, and no increase was found during the 1990s. The irrigation water usage has nearly doubled from the 1950s to the 1980s upstream of the Huayuankou gauge, and the irrigation water usage in the downstream region has increased by nearly a factor of 5 over the last 30 years since the 1960s. The ratio of the irrigation water use (defined as the ratio of the annual gross loss for irrigation to the annual natural runoff) increased continuously from 21% to 68% during the last 50 years. This tells us that the water shortage in this basin is closely related to the irrigation development. During the 1990s the irrigation area upstream of the Huayuankou gauge is known to have been 3.55 million ha; the gross loss of river discharge for irrigation is estimated from the irrigation data to be about 5340 m3/ha (534 mm), while the theoretically estimated irrigation water consumption is about 2560 m3/ha (256 mm). The ratio of the irrigation consumption to the irrigation gross loss is estimated to be about 0.48. This value is close to the irrigation efficiency in the upstream area at the current time [GIWP, 2001]. This low irrigation efficiency implies that there is a high potential for reducing the irrigation loss by improving the irrigation system in the Yellow River basin.
Table 5. Decadal Averages of the Gross Loss of River Discharge for Irrigation
Value in parentheses is the water depth averaged over the drainage area instead of the irrigation area because the irrigation area data of some years are not available. There is a very small drainage area in the downstream of Huayuankou gauge, and the irrigation areas are located mainly outside of the basin. Therefore the gross water loss for irrigation in this region is given in cubic meters only.
Natural runoff is estimated as the sum of the measured discharge and the loss for irrigation.
 Dams, major tools for water resources management, affect the monthly pattern of river discharge. There are six large dams on the main river (see Figure 1). As shown in Figure 7a, the river discharge at the Lanzhou gauge in the dry season (from December to May) increased particularly in the 1970s and 1980s because of the construction of the Liujiaxia and Longyangxia dams. The increased discharge at the Lanzhou gauge by reservoir regulations is used up along the midstream. The low-flow condition at the Huayuankou gauge was not improved much, as shown in Figure 7b. The reservoirs can play an important role in improving the drying up of the main river along the lower reaches while the water intake along the middle reaches is controlled. The reallocation of water resources from the upstream to the downstream region is a key issue for improving the water resources management in this basin.
4.4. Discussion on the Reason for the Drying Up of the Yellow River
 On the basis of the decadal averages given in section 4.3, the interdecadal changes in the climate and the gross loss for irrigation are analyzed and compared with the variations in the river discharge and the drying-up situation. As given in Table 6, it can be seen that the reason for the increase in the river discharge during the 1960s was the increase in precipitation of 22 mm/10 yr. The drying up that occurred in the 1970s was caused by a decrease in precipitation (29.6 mm/10 yr) and an increase in evaporation (7 mm/10 yr for pan evaporation) as well as an increase in irrigation (10.5 mm/10 yr). From the 1970s to the 1980s, irrigation water losses maintained a similar increase (6.9 mm/10 yr); however, the drying-up situation was not aggravated in the 1980s because of an increase in the precipitation (10.3 mm/10 yr) and a sharp decrease in the evaporation (133 mm/10 yr for pan evaporation). Entering the 1990s, it was found that the precipitation sharply decreased by 38.2 mm/10 yr and that the evaporation increased by 52 mm/10 yr (pan evaporation). In order to maintain the same degree of irrigation, most of the available river discharges were used for irrigation. Therefore the drying up of the main river along the lower reaches increased rapidly. The changes in the river discharges at the Huayuankou and Lijin gauges are consistent with the variations in the precipitation, evaporation, and irrigation water loss.
Table 6. Interdecadal Changes in the Drying-Up Situation, River Discharge, Irrigation, and Climatea
Total Days of Drying Up
Discharge, mm/10 yr
Loss, mm/10 yr
P, mm/10 yr
T, °C/10 yr
Epan, mm/10 yr
P is annual precipitation over the upstream area of Huayuankou (HYK), T is annual mean temperature over the upstream area of HYK, Epan is annual pan evaporation estimated from 81 stations in the Yellow River basin (mainly over the upstream area of HYK), “Loss” is the annual gross loss for irrigation in mm over the drainage area, “Lijin” is the Lijing gauge, “Whole Basin” is the whole of the basin up to the Lijing gauge, and “Above HYK” is the upstream area of the Huayuankou gauge.
 Using a 50-year data set of 108 meteorological stations and six river discharge gauges, the spatial and seasonal variability in the climate and water resources in the Yellow River were analyzed. The gross loss of river discharge for irrigation was estimated using a collection of irrigation data, and the irrigation water consumption was estimated using the FAO-recommended method. The long-term changes in the river discharge and drying-up situation were examined together with changes in precipitation, air temperature, pan evaporation, and irrigation water uses for exploring the reason for the drying up of the Yellow River.
 High spatial and seasonal variabilities in climate result in an uneven distribution of water resources from the upstream to the downstream region. On average over the last 50 years the annual precipitation has been estimated to be 440 mm. Assuming that all of the discharge at the Lijin gauge, that is, the last gauge on the Yellow River, flowed into the sea, the water discharging into the sea is 46 mm/yr and has decreased continuously since the 1960s. The gross loss of river discharge for irrigation was estimated to be about 30 mm/yr, taking about 40% of the natural runoff. The ratio of the annual irrigation water use (gross loss) to the natural annual runoff increased from 21% to 68% during the past 50 years since the 1950s. It was found that the irrigation efficiency was very low, which implies that there is a high potential for water saving in this basin in the future.
 Regarding the changes in the precipitation, air temperature, and river discharge over the last 50 years, by applying the Mann-Kendall test, it was found that the annual precipitation showed a nonsignificant decreasing trend of −45.3 mm/50 yr; the annual mean temperature had an increasing trend of 1.28°C/50 yr; the river discharges had significant decreasing trends ranging from −28 mm/50 yr to −61.5 mm/50 yr from the upper to the lower reaches. The decrease in the river discharge corresponded to a decrease in precipitation, an increase in evaporation, and an increase in irrigation water usage. The main reason for the drying up of the Yellow River was the increase in irrigation water uses. However, climate fluctuations could have greatly alleviated or aggravated the drying-up situation.
 The drying up of the Yellow River along the lower reach is becoming more and more sensitive to increases in irrigation and/or climate fluctuations. Together with the problems associated with soil erosion, sedimentation, and flood, a wise comprehensive management of the water resources is required. With the long history of flood control and irrigation in this basin, there are many successful and also failed examples of water resources management over the past 50 years. However, it is essential for there to be no more incidences of large mismanagement of the Yellow River in the future because of the present critical conditions.
 This research was partially supported by the Core Research for Evolutional Science and Technology (CREST) program of the Japan Science and Technology Agency (JST) and was partially supported by the National 973 Project of China (G19990436). The authors would like to express their appreciation for the aid of these grants in this research.