1 Introduction

Appearance date of the ice run, freeze-up and ice cover break-up, the length of ice-affected period and the ice freezing seasons describe the ice regime of a river. The ice regime has important impacts on hydraulic engineering, such as the operation and maintenance of hydraulic plants, waterpower, shipping and water transfer, and has great influence on the river’s environment and ecology (Wang et al. 2008; Morin et al. 2015). River ice is sensitive to climate warming. An increasing number of studies have showed significant trends for later freeze-up and earlier break-up occurrence, and annual number of ice freezing days decrease (Loader et al. 2011; Hodgkins et al. 2005; Prowse and Beltaos 2002). Some records show that, for the twentieth century, freeze-up dates occur on average 2–25 days later, break-up dates 2–24 days earlier, and the ice freezing period also has decreased by an average of 2–38 days over the past 100 years in the Northern Hemisphere (Agafonova and Frolova 2007; Jiang et al. 2008; Klavins et al. 2009). Also, many non-climatic factors could also influence the occurrence of different river ice phenomena, including bed morphology or human activity (Starosolszky 1990; Radoane et al. 2010; Agafonova and Frolova 2007), which could exaggerate or hide natural trends in the river ice regime if they operate in either the same or opposite direction as the natural forces. Therefore, understanding the new ice regime characteristics impacted by environmental change is very important for avoiding ice flood disaster.

The Yellow River is the second-longest river in China. The total length of the main river course is 5464 km; the river basin area is 795,000 km2; and the altitude difference between the river source and the estuary is 4830 m. Compared with other large rivers, Yellow River has the characteristics of high silt content, many twists and turns for its flow, and significant terrain differences. The Ning-Meng reach of the upstream Yellow River (104°E–112°E, 38°N–41°N) is located at the northern part of the Yellow River basin in China, which is more than 1237 km long, and the total water catchment area is 144,000 km2. In this region, the time of air temperature below zero usually lasts from November to March of next year. Also, because the water flows from a low latitude to a high latitude, the freeze-up occurs from downstream to upstream in winter and break-up occurs from upstream to downstream in spring, which may lead to ice jams or ice dams due to the quick increase in the ice-melt flood (Ashton 1986; Beltaos and Burrell 2003; Yang 1996; Wu et al. 2015). The Ning-Meng reach of the Yellow River has experienced frequent ice flooding disasters, which have occurred 88 times in 60 years from 1951 to 2010. Since the 1990s, under the influence of natural factors and human activities, such as higher temperature and reservoir regulation, the ice conditions of the Ning-Meng reach have experienced significant changes. Many existing studies have examined the weather change and water resource evolution for the Yellow River (Huang and Zhang 2004; Saito et al. 2001; Fu et al. 2007; Liu et al. 2012; Zhang et al. 2011); this paper mainly investigates the ice condition changes during the ice flood season. In this paper, simple test is introduced to detect the effect of reservoir operation impacts on river ice regime by using the meteorological, hydrological and ice condition observation data of the Ning-Meng reach.

2 Study area and materials

2.1 Study area

The Ning-Meng reach of the upstream Yellow River belongs to the temperate, arid and semi-arid desert and desert steppe zone. The annual average precipitation is 155–366 mm, with more in the eastern part and less in the western part. In addition, the rainfall is very unevenly distributed during the year, with 75 % of the rainfall concentrated between July and September. Influenced by the continental monsoon climate, the Ning-Meng reach has a cold and dry winter. The days with an average daily temperature below 0 °C can last 4–5 months, and the lowest temperatures in winter can reach −35°. For the whole region, January has the lowest temperature, followed by December, February, November and March. The total freezing periods from November to March of next year can last approximately 120 days, and the total freezing length can reach 1000 km.

The study reaches have six hydrological stations: Xiaheyan, Qingtongxia, Shizuishan, Bayangaole, Sanhuhekou and Toudaoguai, as illustrated in Fig. 1 and Table 1. These stations divide the study reaches into five sections.

Fig. 1
figure 1

Map of Ning-Meng reach of the Yellow River

Full size image
Table 1 Hydrological station data for the Ning-Meng reach
Full size table

The Ning-Meng reach belongs to the second terrace of the Yellow River. Most areas in this zone are desert and desert steppe, with a gentle river bed and slow water flow, and there are large alluvial plains on both banks. Table 2 lists the mainstream watercourse characteristics for the upstream of the Yellow River. As we can see, the watercourse cross-sections from upstream to downstream display a pattern from relatively narrow and deep to relatively wide and shallow. Once the Yellow River flows into Inner Mongolia, its slope becomes smaller. The watercourse morphology for the Ning-Meng reach and its changes have a significant correlation with the ice flooding disasters. This reach of the river flows from Southwest to Northeast; the temperature is warm in the upstream and cold in the downstream; the watercourse is wide in the upstream and narrow in the downstream; and the slope is steep in the upstream and gradual in the downstream (Fig. 1). These features are all adverse with regard to ice discharge. Among them, the river reaches located downstream of the Shizuishan station are the reaches associated with the highest latitude in the Yellow River basin, where the watercourse is wide and shallow, and the mainstream swings and meanders. Therefore, they are the main reaches of the Yellow River to produce ice floods.

Table 2 Mainstream watercourse characteristics for the Yellow River upstream
Full size table

2.2 Data

On the Yellow River, ice monitoring was started at the end of the nineteenth century in several hydrological stations. Both the freeze-up and break-up dates, drift-ice and ice cover were observed. In parallel with ice monitoring, regular water level observation also started in the nineteenth century. In this paper, long-term records of river ice observation (water level, ice cover thickness and a variety of other factors) from 1951 to 2010 for the hydrological stations from the Hydrology Bureau of the Yellow River Conservancy Commission are used. Ice phenomena are recorded daily (i.e., day–month–year), and the definition criteria of the different year river ice phenomena are common and consistent through the entire record. Daily temperature data for the upstream Yellow River are available from the standard methodology of the China Meteorological Administration.

2.3 Overview of ice condition in the Ning-Meng reach

An ice condition is the phenomenon of a rising river water level caused by resistance to the water flow by the ice. It is the consequence of the combined effects of heat, power and watercourse morphology (Smith 2000; Choinski et al. 2010). The concrete manifestations are ice jams, ice dams and drifting ice (Fig. 2). Ice jam is the phenomenon that a large number of small ice deposits formed under the ice cover, leading to the rising of upstream water level. The ice jams formation is a part of surface ice cover progression process. The transport and deposition of the entrained ice floes on the underside of the cover can form thick accumulations commonly known as ice dams. Ice dams usually appear in the ice break-up period across all or most of the section, and then the flow capacity in the river channel is greatly reduced. During the ice-affected season, frazil ice phenomenon could cause a serious ice flood disasters such as overflow, bursting and river building damage. The ice flooding disasters can be categorized into three types: ice jam disasters, ice dam disasters and drifting ice impact disasters. The ice dam disasters can be more abrupt and violent with a higher potential of ice dam formation.

Fig. 2
figure 2

Ice jam and ice dam in the Ning-Meng reach of the Yellow River

Full size image

For the Ning-Meng reach, ice condition occurs every year. From 1951 and 2010, ice condition occurred 346 times, and the average number per year is 5.77 (Table 3). These instances included 88 disasters, accounting for 25.4 %. No disasters were generated for the other 258 incidents, accounting for 74.6 %. From the type of the ice condition, there were 23 ice jams, accounting for 6.7 %; 322 ice dams, accounting for 93 %; and 1 drifting ice, accounting for 0.3 %. Also, ice dam disasters are the main form of ice flood disasters in the Ning-Meng reach. The reaches associated with the most serious ice jam disasters and ice dam disasters are the reaches between Xiaheyan–Shizuishan and Sanhuhekou–Toudaoguai, respectively.

Table 3 Ice dam and ice jam in the Ning-Meng reach of the Yellow River in 1951–2010
Full size table

3 Factors of influencing river ice regime

3.1 Temperature

River ice occurrence and the amount of drift-ice are directly controlled by air temperatures. The relationship between river ice phenomena and monthly mean temperatures for the Ning-Meng reach was analyzed by correlation tests (Table 4). The strongest relationship was found between mean winter air temperature and the ice cover thickness of ice-affected days.

Table 4 Relations between river ice phenomena and mean monthly temperatures (Unit: °C)
Full size table

Since the 1990s, influenced by global warming, warm winter phenomena have occurred along the Ning-Meng reach. The average temperatures for the Shizuishan, Bayangaole and Sanhuhekou stations during the ice freezing season have been on the rise for the past 20 years (Fig. 3). Table 5 lists the decade averages of the temperature characteristics for the past 50 years during the ice freezing season of the Shizuishan station. In the 1990s, the average monthly temperatures are higher than in the 1980s. After 2000, the temperatures increased further. In particular, the temperature rise (up to 3.6 °C) has been very significant during December–February. The higher temperature during the ice freezing season is the main reason for the ice break-up date advance and the freeze-up date delay.

Fig. 3
figure 3

20-Year temperature data from 1991 to 2010

Full size image
Table 5 Temperature statistics during ice period of the Ning-Meng reach (Unit: °C)
Full size table

3.2 River channel shrinkage

For the past 20 years, water consumption along the Yellow River basin has accelerated because of increasing water demand by a growing population and extensive agricultural development. Over the same period, climatic changes have led to a reduction in precipitation in the river basin (Fu et al. 2007; Liang et al. 2010). Also, the amount of sediment entering Yellow River has sharply increased. As a result, the main channel for the Inner Mongolia reach has shrunk year by year, and the riverbed has been consistently elevated due to siltation.

The Yellow River has experienced a series of dry years since 1986. In the period of 1987–2005, the annual average runoff and sediment load for all the major control stations along the Yellow River upstream decreased compared to the multi-year averages. Among them, the Lanzhou and Toudaoguai stations experienced reductions of 16, 41, 30 and 64 % (see Table 6). The reduction in the sediment transport capacity results in serious silting (Xu 2002; Vanacker et al. 2005). With the Sanhuhekou station as an example, the section morphology in 2005 underwent serious deformation compared with it in 1987. In particular, the main river channel width was reduced by 120 m; the average silt thickness was greater than 1.9 m; and the cross-sectional area was reduced by 794 m2 (27 %). So, the river channel had experienced serious shrinkage. The silt resulted in reduced flow capacity across the cross section. The water levels of 550 m3/s flow at the Sanhuhekou station during 2001–2005 were 1018.8, 1018.7, 1019.3, 1019.4 and 1019.4 m, respectively, and the bankfull discharge flow volume fell from 4000–5000 m3/s in the 1970s to the present 1000–2500 m3/s, indicating a significantly reduced flow capacity in the river channel.

Table 6 Flow and sediment discharge at the main stations
Full size table

3.3 Reservoir Ice control operation

The past 50 years were a period of dramatic and unprecedented change in human history, during which many aspects of human activities, including world population, water use and damming of rivers, have changed at an increasing rate, resulting in global-scale changes in the earth’s system. Dams and reservoirs are globally essential to the river fragmentation and thus having significant impacts on the water regulation and sediment retention. For the Yellow River, the construction of dams and reservoirs is the most important human activities affecting the river ice condition. Since the 1950s, more than 3147 reservoirs had been built in the Yellow River basin, with a total storage capacity of 57.4 × 109 m3 (Ta et al. 2008; Shao and Wang 2002). There are 24 reservoirs scattering widely in the river basin with storage capacities exceeding 0.1 billion m3, among which four major reservoirs (Longyangxia, Liujiaxia, Sanmenxia and Xiaolangdi) with a total storage capacity of 30.4 billion m3 and a regulation capacity of 23.5 billion m3 make the greatest contribution to water regulation. All others are runoff stations. The reservoirs closely related to the Ning-Meng reach ice flood prevention are mainly Longyangxia and Liujiaxia stations.

The Liujiaxia and Longyangxia reservoirs were constructed in the upper reaches above Lanzhou in 1968 and 1985, respectively. The Liujiaxia reservoir has a storage capacity of 5.7 km3 and a 147-m-high dam wall. It normally stores about 2.7 km3 of water and traps 0.06 Gt of sediment annually. The Longyangxia reservoir (27.6 km3 of storage capacity and 178-m-high dam wall) is a multi-year reservoir. Reservoir regulation changes the river ice regime, because the morphological conditions of the river are altered by dams and other engineering structures. In the Yellow River, after the Liujiaxia reservoir was put into operation in October 1968, the regulation effect on the Ning-Meng reach ice flood prevention was apparent. Since the joint operation of Liujiaxia and Longyangxia reservoir after 1986, the regulation functions have been enhanced.

In order to see the potential effect of the reservoir impacts, the ice regime of years with similar conditions before and after the reservoir operation impact was compared. The entire observation term of river ice has been separated into three characteristic periods:

  • The first period: 1950–1967, no reservoir regulation.

  • The second period: 1968–1986, intensive period of regulation by Liujiaxia reservoir.

  • The third period: 1987–2010, joint regulation by Liujiaxia reservoir and Longyangxia reservoir.

Increasing average winter temperature suggests that the occurrence of ice phenomena would be expected to decrease. In order to exclude the effect of rising air temperature and to test the impact of reservoir operation on ice regime, the effect of interannual variability of winter air temperature was eliminated by selecting years from each period with mean temperature between −9.4 and −2.3 °C (Table 7). Then the ice regime of years with similar mean winter air temperature for each period was compared.

Table 7 Temperature characteristics of the selected winters
Full size table

4 Reservoir operation impacts on ice regime variation

4.1 Duration of ice freezing season and length of freezing reaches reduce

In the first period, without the Longyangxia and Liujiaxia reservoirs, Ning-Meng reach was essentially all frozen: the Shizuishan, Bayangaole, Sanhuhekou and Toudaoguai stations experienced 73, 100, 107 and 93 freezing days on average, respectively. In the second period, the durations of ice freezing at Shizuishan and Bayangaole stations reduced by 10 and 4 days, respectively, while the other two stations’ extended by 8 and 15 days, respectively. In the third period, the durations of ice freezing at Shizuishan, Bayangaole and Sanhuhekou stations reduced by 33, 22 and 8 days, respectively, compared to the duration without the reservoirs. In addition, the beginning date of break-up for the main stations all moved ahead after the joint operation of the reservoirs (Table 8).

Table 8 Beginning date of ice break-up along the Ning-Meng reach
Full size table

4.2 Ice cover becomes thinner

Figure 4 lists the statistics on the ice cover thickness changes. It can be observed that in the first period, the ice cover thickness was relatively even. In the second period, the thickness for the Shizuishan and Toudaoguai station underwent 13 and 17 cm reductions, respectively, while in the third period, the ice cover thickness at the Shizuishan, Bayangaole, Sanhuhekou and Toudaoguai station was respectively 16, 22, 25 and 24 thinner than that without reservoirs.

Fig. 4
figure 4

Ice cover thickness changes at stations along the Ning-Meng reach

Full size image

4.3 Channel storage water increase significantly increases

Since the joint operation of the Longyangxia and Liujiaxia reservoirs, the runoff along the Ning-Meng reach has increased by approximately 20 %. Using Bayangaole and the Sanhuhekou station as examples, in the first period, the average discharge was 6.09 and 5.96 billion m3, respectively, during the ice freezing season. In the third period, the average discharge was 7.35 billion and 7.25 billion m3, respectively. When the runoff increases during the ice freezing season, the channel storage water significantly increases. In the first period, the average annual maximum channel storage water was 0.883 billion m3. In the second and third period, it was 0.996 and 1.28 billion m3, respectively, which increased by 12.8 and 45 % compared to the average with no reservoir regulation.

4.4 The number of ice dams reduced, and the ice jam disasters exacerbated

Table 9 compared the times of ice conditions before and after reservoir constructions for the selected years. There were 149 ice dam incidents along the Ning-Meng reach during the period 1951–1968. For the period 1969–2010, there were 75 ice dam incidents, and the average number of ice dam incidents per year was, respectively, 11.46 and 2.78, such that the number of ice dam had reduced significantly. In contrast, the ice jam disasters grew worse. During the period 1951–1968, there were only two ice jams along the Ning-Meng reach, while during the period 1969–2010, 13 ice jams occurred, and the average number of ice jam incidents per year was, respectively, 0.15 and 0.48. Also, the average number of ice flood disasters per year decreased from 1.69 to 1.41, and the number of ice dam disasters reduced from 1.61 to 1.03.

Table 9 Comparison of the number of ice flooding incidents and disasters before and after reservoir operation for the selected years
Full size table

5 Discussion

Various factors determine the ice regime of a river, including air temperature, bed morphology, water level, runoff velocity and other technical objects (Beltaos and Prowse 2009). Although the date of ice appearance and the intensity of ice-drifting are determined mainly by the temperature under natural conditions, morphological factors and anthropogenic effects play an important role in the date of ice cover formation (Agafonova and Frolova 2007). Reservoir usage changes the river ice regime, because the morphological conditions of the river are altered by dams and other engineering structures. Also, reservoir operation and the altered morphological parameters change water temperature conditions (Meilutyte-Barauskiene et al. 2005; Takács et al. 2013). Consequently, downstream of the dam, flow velocity and turbulence grow, so ice occurrence and freeze-up become less frequent (Starosolszky 1990).

For the upstream Yellow River, reservoir operation is a complex decision-making process that yields maximum benefits with respect to multiple objectives such as water demands, ecosystem requirements hydropower generation and ice control. In the ice flood season from November to March of next year, the ice control is the most important objective. From 1968–1986, Liujiaxia reservoir was in operation by itself. The mode of ice control operation of Liujiaxia reservoir is as follows: (1) in order to increase the ice conveyance capacity and avoid the ice jam disaster before freeze-up period, the reservoir discharge should increase appropriately from 600 to 700 m3/s. (2) During freeze-up period, the Liujiaxia reservoir discharge is controlled as about 600 m3/s and decreased progressively with the thawing of the river to create good conditions for ice break-up. (3) In break-up period, reservoir discharge should be little than 500 m3/s to make the downstream water level of Shizuishan and Bayangaole reach decrease steadily. Table 10 is the average monthly inflow and outflow of Liujiaxia reservoir during ice flood season from 1972 to 1986. With the operation of Liujiaxia reservoir, the ice threat of Ningxia–Inner Mongolia reach reduces dramatically by controlling the reservoir discharge every month. After the joint operation of Longyangxia and Liujiaxia reservoir after 1986, the discharge limitation of Liujiaxia reservoir is the same with that before 1986. However because of the good regulation performance of the Longyangxia reservoir, the Liujiaxia reservoir has become a re-regulating reservoir to Longyangxia reservoir, and the regulation ability to comply with the ice control flow limitation is increased. Table 11 is the average monthly inflow and outflow of Liujiaxia reservoir with the joint operation of Longyangxia and Liujiaxia reservoirs.

Table 10 Average monthly inflow and outflow of Liujiaxia reservoir with the single operation (m3/s)
Full size table
Table 11 Average monthly inflow and outflow of Liujiaxia reservoir with the joint operation (m3/s)
Full size table

After the reservoir construction, the seasonal distribution of the runoff changed. The average monthly discharge at Shizuishan, Bayangaole, Sanhuhekou and Toudaoguai stations during the ice freezing season from December to March all displayed increasing trends (Fig. 5). In particular, the discharge increased during the river freeze-up season, leading to postponed dates of ice run and freeze-up at all stations, but the further downstream the station was, the less was the impact.

Fig. 5
figure 5

Average inflow and outflow of Liujiaxia reservoir in the ice-affected season a single operation, b joint operation

Full size image

In the second period, while the Liujiaxia reservoir was operated alone, the beginning date of ice run across the Shizuishan and Bayangaole station was postponed by 5 and 6 days, respectively, and the river freeze-up date was postponed for 11 and 5 days, respectively. In the third period, after the Longyangxia and Liujiaxia reservoirs were put into operation, the monthly average discharge at Lanzhou station increased by 100–150 m3/s compared to its value before the reservoir operation and the water temperature increased by more than 2°–4°. As a result, the Lanzhou reach becomes a perennial smoothly flowing reach. Tens of kilometers of reaches downstream of Qingtongxia station do not freeze anymore. Under the joint impacts of hydro and thermal effects, the ice run dates for the Shizuishan and Bayangaole stations were postponed by 15 and 12 days, respectively, and the freeze-up dates were postponed by 18 and 17 days, respectively, compared with the dates before the reservoirs operation. The ice run date for the Sanhuhekou station did not change, but under the influence of the delayed ice run from the upstream reaches, its freeze-up date postponed for 7 days (Table 12; Fig. 6).

Table 12 Freeze-up date at stations along the Ning-Meng reach
Full size table
Fig. 6
figure 6

Average discharge during the ice-affected season

Full size image

The evolution of ice condition in the Yellow River is complex and is difficult to predict. Therefore, prevention and control of ice flood is particularly important. Here are two suggestions to reduce the ice flood disasters. (1) Reservoir real-time operation combined with weather forecast is the key to ice flood prevention in the Ning-Meng reach of the Yellow River. Controlled by the Siberia monsoon climate, the weather is dry and cold in winter of the Ning-Meng reach, and sometimes the temperature could drop because of the influence of strong cold air, leading to the river freeze-up after 1 or 2 days of ice run. Also, the Liujiaxia reservoir is 1267 km from the Ning-Meng reach, and the water transmission period is only about 16.5 days during ice flood season. So, reservoir operation scheme should be made based on the weather and hydrological forecast. However, the current weather and hydrological forecast accuracy and period could not fully meet the requirements of ice flood prevention, and hence, reservoir operation scheme should be fixed according to the short-term forecast results. (2) Construction of ice flood diversion project is another effective measure. Some flood detention zones in the Ning-Meng reach have been constructed, such as the Wuliangsuhai detention zone. Optimization of the ice flood diversion process could alleviate the pressure of Liujiaxia reservoir for ice flood control and ensure the safety of the Ning-Meng reach in the ice flood season.

6 Conclusion

For the Ning-Meng reach in the Yellow river, due to its weather conditions, channel morphology and hydrology, ice conditions occur every year. There were 23 ice jam incidents and 322 ice dam incidents during 1951–2010. Among them, 88 became disasters, and the number of ice dams is far greater than the number of ice jams. Since the 1990s, the characteristics of ice regime along the Ning-Meng reach have shown significant changes affected by climate change, river erosion and deposition, reservoir operation and river engineering.

The warmer temperature is the one of the main reasons for the earlier ice break-up date, postponed river freeze-up date and thinner ice cover. Compared to the 1980s, the average temperature during the ice freezing season increased by 1.4° in the 1990s. After 2000, it increased by 2.0°. Another natural factor is the main channel of the Inner Mongolia reach has shrunk year by year, and the riverbed has been consistently elevated due to siltation. For the Sanhuhekou station, the discharge reduced from 4000–5000 m3/s during the 1970s to 1000–2500 m3/s at present, leading to poor ice discharge during the ice-affected season, increasing the probability of ice jams.

The use of reservoirs had significant impacts on the ice regime along the Ning-Meng reach under the influences of hydrodynamic and thermal effects. This paper investigated reservoir operation effect on river ice regime by eliminating the impact of climatic conditions, and the test method could be applied to other rivers, where similar anthropogenic impacts can be suspected to affect the river ice regime. Compared to the statistics when there were no reservoirs, the ice dam disasters reduced significantly, but the ice jam disasters exacerbated during the river freezing period. The dates of freeze-up for the Shizuishan, Bayangaole and Sanhuhekou stations delayed by 18, 17 and 7 days, respectively. The duration of ice freezing days reduced 33, 22 and 8 days, respectively. The Shizuishan station’s ice break-up date moved ahead 14 days, and the Bayangaole and Toudaoguai station’s ice break-up date moved ahead by 7 days. At the same time, after the reservoir began operation, the annual channel maximum storage water reached 1.28 billion m3, which is 45 % compared to the same statistics without reservoirs. The ice regime on the downstream reach of the Ning-Meng reach along the east–west direction is primarily affected by the regional weather environment. The upstream reach along the south–north direction is affected not only by regional weather environment change but also by reservoir projects. In addition, with decreasing distance from the upstream reservoir, its impact becomes more obvious.