Abstract

Extreme hydrological events have occurred in many climate zones in recent decades. Most importantly, the water distribution in hydrological components has changed with apparent variations in climate. The associated impact on water resources is of concern because an understanding of the hydrological response mechanism is necessary for human survival. In this study, we compare precipitation and streamflow responses to climate variations in two different climate zones. Continental-scale frigid zone (CSFZ) data were collected from Russia, while island-scale subtropical zone (ISSZ) data were collected from Taiwan. The results show that the teleconnection of the precipitation between the ISSZ and CSFZ is subtle and is linked to global atmospheric conditions. The daily maximum precipitation and the duration without precipitation increased in both the CSFZ and the ISSZ. The streamflow response became more extreme in the ISSZ and was associated with pronounced dry and wet seasons. In contrast, a rise in winter temperatures has led to more uniform streamflow and extreme hydrological situations have become less frequent. The responses of streamflow to recent climate variations in the CSFZ and ISSZ are different. Precipitation and temperature are driving forces for the change in streamflow in the CSFZ while precipitation is for the ISSZ.

INTRODUCTION

Water resources are vital for all life. The anthropogenic effect has significantly altered the hydrologic cycle at both global and regional scales, with potentially devastating impacts on water resources (Soden & Held 2006; Kundzewicz et al. 2007; Barnett et al. 2008; Milly et al. 2008). The outcome could become even worse with the projected changes in temperature and rainfall during the 21st century (Von Holle 2008). This is coupled with the likelihood that climate change will exacerbate extremes (Alley et al. 2003). Variations in the global climate and the rapid increase of water demand will lead to a critical imbalance between water supply and demand.

Although climate change has been widely recognized (IPCC 2007, 2013), the impacts of climate change on the global water cycle are not fully understood (Allen et al. 2004). The main reasons for this are that: (1) extensive historical data about the characteristics of climate change are not available; (2) driving forces that cause such changes are as yet unclear; and (3) there are complex interactions between hydrological components, which make it difficult to clarify causal connections. Variations in hydrological components have appeared in many areas of the world; however, the spatial characteristics of such variations are seldom fully understood. Streamflow is a hydrological component that is closely related to the development of human society. It controls the rise and fall of civilizations. Rivers provide resources for irrigation, domestic water, hydroelectric power, shipping, recreation, and wastewater discharge. The danger of flooding and the influence of dry streams on ecology are never too important to be neglected. However, the impacts of climate variations on streamflow are not clear for different climate zones.

Precipitation is the main source of surface water and groundwater in the island-scale subtropical zone (ISSZ) while snow and rain play a similar role in the continental-scale frigid zone (CSFZ). Taiwan is a typical area located in the ISSZ while Russia is highly representative of the CSFZ. As in many areas of the world, precipitation in Taiwan changes in terms of both spatial and temporal characteristics (Hsu & Li 2010). Droughts and strong-intensity rainfall have appeared more frequently in recent years (Liu et al. 2009). In the CSFZ, an increase in winter temperature has led to a decrease in the depth of frozen soil and an increase in soil drainage. In addition, an increase in the number and duration of winter thaws can be observed. The water yielded from the snowmelt recharges groundwater and promotes surface streamflow formation. The permafrost process slows in catchment areas contributed to by winter runoffs. Rises in temperature also lead to a decrease in the runoff that forms the ice cover and an increase in the hydraulic conductivity of watercourses that increases the low flow in the cold season (Kalyuzhny & Lavrov 2012; The Federal Service for Hydrometeorology and Environmental Monitoring of Russia (Roshydromet) 2014). Increases in the infiltration of snowmelt water results in increases in groundwater resources (Dzhamalov et al. 2015).

Streamflow in the ISSZ (Taiwan) is highly correlated with both precipitation and human activities. Most of the streamflow is intercepted by reservoirs or weirs. Watersheds with less anthropogenic effects in Taiwan are located in mountainous areas (e.g., Ailiao). Due to the steep slopes across the terrain, precipitation takes the form of overland flow, and the streamflow is very rapid. Heavy rainfalls result in flooding and landslides. During the dry season, groundwater provides the baseflow of streams for irrigation. When the stream level is too low, irrigation has to be shut down to allocate the water necessary for human and industrial use. On the other hand, streamflow in the CSFZ (Russia) has greatly changed under the impact of climate change and human activities (The Federal Service for Hydrometeorology and Environmental Monitoring of Russia (Roshydromet) 2014). The most significant changes in the European region of Russia have occurred along the border between the forest and the steppe zones (e.g., Oka Basin) (Dzhamalov et al. 2015). This region will be facing water scarcity in the future according to climate- and economic-related projections (Georgiadi et al. 2014). One of the most striking changes concerns the flow regime. It is predicted that there will be an increase in streamflow during the summer and winter and a decrease in spring flooding (Frolova et al. 2017). Due to frequent winter thaws, it has become difficult to distinguish between winter floods and the spring flooding period. The spring flooding period now starts 5 days earlier than before; however, the date of the end remains unchanged, resulting in it being longer in duration. Similar tendencies can be observed for river runoffs throughout Europe (Günter et al. 2017). A decrease of 10–15% in the spring flood flow has been observed in the Oka Basin. A change in maximum discharge of 20–40% for the Oka River was recorded from 1946 to 2010 (Frolova et al. 2017). Changes in precipitation during the warm season have an influence on the formation of floods in the Oka Basin, but this has not been studied yet.

Although changes in hydrological components have been observed in both the ISSZ and CSFZ, the characteristics of these changes are not fully understood and thus hinder the study of global water balance. Understanding this connection will help lead to an understanding of global water distribution. In this study, we investigate the impact of climate variation on both precipitation and streamflow. Data from Taiwan and Russia were analyzed using the same methodology, so the hydrological characteristics and teleconnection could be analyzed.

METHODOLOGY

Site description

The Ailiao River watershed is located in southern Taiwan between the latitudes of 22°47′37″N and 22°48′17″N and between the longitudes of 120°42′52″E and 120°43′E. It has an area of 829 km2. The relief of the watershed is significant, with a height difference of 2,696 m. The Ailiao watershed enjoys abundant precipitation to provide for the agriculture and the livelihood of the Pingtung Plain. The Ailiao River is a 69-km-long upstream tributary of the largest river of Taiwan, the Gaoping River. The water from the Ailiao watershed drains toward the Pingtung Plain, which is the largest alluvial fan in Taiwan. The Pingtung Plain also accommodates other shorter tributaries: the Laonong River, the Donggang River, the Linpian River, and the Qishan River. The geography of the Ailiao watershed along with the locations of precipitation and streamflow stations are shown in Figure 1. The Water Resources Agency provided the streamflow and precipitation data for the Pingtung Plain. For the data analysis, we chose stations at the foothills with less human activity. Thus, the Sandimen Flow Station and the Sandimen Precipitation Station were selected. Table 1 shows data acquisition for the Aliliao watershed. The total time span of the data is 50 years from 1965 to 2014.

Figure 1

Geographical location of the Ailiao watershed. The location of the hydrological station is also shown.

Figure 1

Geographical location of the Ailiao watershed. The location of the hydrological station is also shown.

Table 1

Data acquisition for the Ailiao watershed, Taiwan

Type Site Data period Years 
Precipitation Sandimen 1965–2014 50 
Streamflow Sandimen 1965–2014 50 
Type Site Data period Years 
Precipitation Sandimen 1965–2014 50 
Streamflow Sandimen 1965–2014 50 

The part of the Oka Basin of interest is located on the East European Plain between the latitudes of 55°30′00″N and 52°12′00″N and between the longitudes of 33°00′00″E and 39°00′00″E. It has an area of 54,900 km2. The relief of the basin is flat, with a height difference of 104 m. The basin is located mainly in the mixed forest zone, where precipitation is more volatile. The southeastern part of the basin is located in the forest-steppe zone, where precipitation does not exceed the potential evapotranspiration (Moscow-Oka Basin Water Administration of Federal Agency for Water Resources 2015). The geography of the Oka Basin and the locations of the meteor and gauge stations are shown in Figure 2. There is a cold season (November–May) when rainfall has little effect on the streamflow. There is also a warm season (June–October). This area is subject to global warming. The annual average air temperature increased by 0.35 °C/10 years from 1950 to 2012. This was almost twice the growth rate of the global average air temperature over land (0.18 °C/10 years) (Hartmann et al. 2013). There is no permafrost in the Oka Basin, but from November to March, there is a layer of frozen, impermeable soil. The average thickness of this layer significantly decreased from 1950 to 2012 (by 50–70 cm) as a result of an increase in the winter air temperature (Kalyuzhny & Lavrov 2016). Table 2 shows data on precipitation and stream flow for the Oka Basin. The total time span of the data is 50 years from 1965 to 2014. The streamflow data for Kaluga Station has missing monthly data in 1994, which are supplemented using downstream data from Murom Station.

Figure 2

The Oka Basin with its rivers, meteostations, and the streamflow gauge.

Figure 2

The Oka Basin with its rivers, meteostations, and the streamflow gauge.

Table 2

Data acquisition for the Oka Basin, Russia

Type Site Data period Years 
Precipitation Basin average 1965–2014 50 
Streamflow Kaluga 1965–2014 49 
Type Site Data period Years 
Precipitation Basin average 1965–2014 50 
Streamflow Kaluga 1965–2014 49 

The Thiessen polygons method is used to calculate the area-weighted precipitation in the Oka Basin. This method is used for two reasons: (1) there is no missing data in the precipitation during 1965–2014 and (2) the altitude differences in the Oka Basin are relatively small (about 100 m), so the influence of relief on precipitation can be neglected.

The average precipitation during the study period was 640 mm. The seasonal distribution of precipitation was uneven. Most of it (70%) fell from April to October, with a maximum in July and August. The least amount of precipitation was observed for February and March. Rain, snow, and mixed precipitation comprised 70%, 20%, and 10% of the total amount, respectively (Moscow-Oka Basin Water Administration of Federal Agency for Water Resources 2015). The Oka Basin is essentially unregulated although its drainage basin contains large cities, with a total population of two million. The annual irretrievable water consumption in the Oka Basin reached 6.8 million m3 from 1965 to 1989 (Moscow-Oka Basin Water Administration of Federal Agency for Water Resources 2015). By 2015, the irretrievable water consumption had increased to 32 million m3, which is still less than 0.4% of the total annual flow (The Federal Service for Hydrometeorology and Environmental Monitoring of Russia (Roshydromet) 2016). About half of the annual flow of the Oka River manifests during the spring flooding period. One-third of the annual flow occurs between June and November. The winter flow is consistently low; it is only 16% of the annual flow (Frolova et al. 2017).

Analysis methods

The characteristics of precipitation and streamflow for the ISSZ (Ailiao watershed, Taiwan) and the CSFZ (Oka Basin, Russia) for the same period from 1965 to 2014 were considered. In the case of the Ailiao watershed, data were analyzed for the wet season (May–October), the dry season (November–April), and for the entire year. For the Oka Basin, data for the cold season (November–May), the warm season (June–October), and for the entire year was considered. This classification for the Ailiao watershed was based on the temporal characteristics of precipitation. Typhoons, convective rain, and monsoons provide the streamflow during the wet season. For the dry season, groundwater supplies the baseflow of the streamflow. For the Oka Basin, the classification follows the genesis of runoff. For the cold season, the source of supply for the river is underground and thawed water. For the warm season, the streamflow is supplied from groundwater and rainwater.

The amount of runoff and precipitation was calculated for each time period. To characterize the distribution of the runoff and precipitation for each period, we used the Mann–Kendall trend test (MK), the mean absolute difference (MD), and the relative mean absolute difference (RMD).

The Mann–Kendall (MK) trend test (Mann 1945; Kendall 1975) is one of the most widely used non-parametric tests to detect significant trends in hydrologic data time series (Naghettini 2017). The MK test is a function of the ranks of the observations rather than their actual values. It is not affected by the actual distribution of the data and is less sensitive to outliers. The test is based on the correlation between the ranks of a time series and their time order. For a time series , the test statistic value S is calculated as:  
formula
where the test statistic depends only on the ranks of the observations, rather than their actual values. Under the assumption that time series data are independent and identically distributed, mean and variance of S are calculated as (Kendall 1975):  
formula
 
formula
where n is the length of the period. The distribution of S tends to normality as the number of time series data becomes large. The significance of trends can be tested by comparing the standardized variable Z, which is defined as:  
formula
When , this means the time series has a statistically significant trend, where is the significance level. A positive Z indicates an increasing trend in the time series, while a negative Z indicates a decreasing trend. In this study, we set as 5% and 10% which means , respectively.
The MD value is expressed as:  
formula
where yi and yj measure either the precipitation or the average daily flow for day i and day j of the given period. RMD is twice the Ginny coefficient that was used to characterize the degree of unevenness of the runoff and the precipitation (Martin-Vide 2004; Zhang et al. 2015; Voskresenskaya & Vyshkvarkova 2016). RMD is the corresponding relative value  
formula
where the designations are similar to those for MD, and is the average daily flow or precipitation over the period under consideration. MD and RMD indicate the degree of non-uniformity. The larger the MD or RMD are, the more heterogeneity there is (i.e., non-uniformity). MD shows the absolute value for the degree of non-uniformity while the RMD shows the corresponding degree of non-uniformity.

The maximum daily precipitation and the maximum duration without precipitation within the given period were used to measure extremity. For streamflow, the maximum daily average streamflow and the minimum ten-day average water consumption were used to measure extremes.

RESULTS AND DISCUSSION

Data analysis of precipitation and streamflow

The Sandimen Station data covered precipitation and streamflow from January 1965 to December 2014. The temporal frequency of the data was monthly, and the analysis followed the above-mentioned classification of the year. Table 3 shows the trend detection using the MK test for streamflow and precipitation for the Ailiao watershed. Table 4 contains the streamflow and precipitation statistics. Figures 3 and 4 show the precipitation data and streamflow data, respectively, for each month, the wet season, and for the dry season.

Table 3

Mann–Kendall test for monthly data for the Ailiao watershed, Taiwan

Data type (monthly) Period Mann–Kendall test
 
Increase/Reduce (I/R)  Z > 1.645  > 1.96 
α = 10% α = 5% 
Precipitation Whole year 0.099 
Wet season −0.484 
Dry season 0.627 
Streamflow Whole year −1.011 
Wet season −1.739 
Dry season −1.966 
Data type (monthly) Period Mann–Kendall test
 
Increase/Reduce (I/R)  Z > 1.645  > 1.96 
α = 10% α = 5% 
Precipitation Whole year 0.099 
Wet season −0.484 
Dry season 0.627 
Streamflow Whole year −1.011 
Wet season −1.739 
Dry season −1.966 
Table 4

Streamflow and precipitation statistics for the Ailiao watershed, Taiwan

  Streamflow (m3/s)
 
Precipitation (mm)
 
Year Wet season Dry season Year Wet season Dry season 
Mean 37.46 68.25 6.67 252.08 468.19 35.97 
Max 454.38 454.38 81.6 2,093 2,093 274.5 
Min 0.076 0.46 0.076 
Range 454.3 453.9 81.5 2,093 2,091 274.5 
CV 1.49 0.95 1.79 1.38 0.82 1.26 
Cs 2.54 1.84 1.43 2.07 4.15 2.06 
  Streamflow (m3/s)
 
Precipitation (mm)
 
Year Wet season Dry season Year Wet season Dry season 
Mean 37.46 68.25 6.67 252.08 468.19 35.97 
Max 454.38 454.38 81.6 2,093 2,093 274.5 
Min 0.076 0.46 0.076 
Range 454.3 453.9 81.5 2,093 2,091 274.5 
CV 1.49 0.95 1.79 1.38 0.82 1.26 
Cs 2.54 1.84 1.43 2.07 4.15 2.06 

CV, coefficient of variation; Cs, coefficient of skewness.

Figure 3

Precipitation measured at Sandimen Station for the (a) whole year, (b) wet season, and (c) dry season.

Figure 3

Precipitation measured at Sandimen Station for the (a) whole year, (b) wet season, and (c) dry season.

Figure 4

Streamflow measured at Sandimen Station for the (a) whole year, (b) wet season, and (c) dry season.

Figure 4

Streamflow measured at Sandimen Station for the (a) whole year, (b) wet season, and (c) dry season.

Table 3 shows the results of the MK test for three different periods using monthly data for the Ailiao watershed. A value of 1 means the data have a statistically significant trend, and 0 means there is not a statistically significant trend. The Z of precipitation exhibited an increasing trend over the entire year and for the dry season but showed a reducing trend in the wet season. No statistical significance was detected. The streamflow data showed decreasing trends for the entire year, and for both the wet and dry period. Statistical significance was found for the wet season and the dry season.

Table 4 shows that the mean values of streamflow and precipitation during the wet season exceeded the values in the dry season, where the mean precipitation in the wet season was 13 times that of the dry season, while the mean streamflow was ten times greater. Precipitation in the wet season provided most of the fresh water. The maximum values of both hydrological components culminated during the wet season. They were almost five times and seven times the maximum values during the dry season, respectively. This tremendous amount of water was mostly brought about by typhoons and convective rain, which sometimes induce flooding. The minimum value was close to zero for precipitation and less than 0.5 m3/s for streamflow. The huge difference between the dry and wet seasons makes it difficult to maintain a stable water source. The variations in streamflow were considerably larger than those for precipitation, as reflected by the value of the coefficient of variation (CV). The probability distribution functions (PDF) of both streamflow and precipitation exhibited significant positive skewness.

As shown in Table 4, the monthly mean for annual precipitation was 252.08 mm, slightly rising by 0.04 mm/year (Figure 3(a)). From Figure 3(b), the mean for monthly precipitation during the wet season was 468.19 mm, rising slightly by 0.12 mm/year. From Figure 3(c), the mean for the monthly precipitation during the dry season was 35.97 mm, rising slightly by 0.015 mm/year. The annual amount of precipitation did not change to any significant degree. After 2000, the precipitation increased in the wet season and decreased in the dry season.

Precipitation reaches the land surface to generate either overland flow or infiltration. The overland flow and the baseflow converge toward streams to form streamflow. The time lag between precipitation and streamflow indicates how fast the watershed responds to precipitation. Streamflow serves as the major surface water source. Figure 4(a) shows the time variations for streamflow at Sandimen Station. The mean for annual streamflow was 37.46 m3/s, rising by 0.002 m3/s/year. Figure 4(b) shows the monthly streamflow during the wet season, with a mean of 68.25 m3/s, decreasing by 0.019 m3/s/year. Figure 4(c) shows the monthly streamflow during the dry season, with a mean of 6.67 m3/s, increasing by 0.023 m3/s/year. As shown in Figure 4, more erratic streamflow has occurred in recent years, indicating more uneven distribution.

A 50-year time span was used for the data at the Oka Basin for precipitation (averaged from 31 weather stations) and for streamflow (the Kaluga Gauge Station). All of the experiments were carried out from 1965 to 2014. The analysis was based on monthly data. Table 5 shows the trend detection using the MK test for streamflow and precipitation for the Oka Basin. Table 6 shows the streamflow and precipitation statistics.

Table 5

Mann–Kendall test for monthly data for the Oka Basin, Russia

Data type (monthly) Period Mann–Kendall test
 
Increase/Reduce (I/R)   > 1.645  > 1.96 
α = 10% α = 5% 
Precipitation Whole year 1.246 
Cold season 1.427 
Warm season 0.236 
Streamflow Whole year 7.770 
Cold season 4.466 
Warm season 6.545 
Data type (monthly) Period Mann–Kendall test
 
Increase/Reduce (I/R)   > 1.645  > 1.96 
α = 10% α = 5% 
Precipitation Whole year 1.246 
Cold season 1.427 
Warm season 0.236 
Streamflow Whole year 7.770 
Cold season 4.466 
Warm season 6.545 
Table 6

Streamflow and precipitation statistics for the Oka Basin, Russia

  Streamflow (m3/s)
 
Precipitation (mm)
 
Year Warm season Cold season Year Warm season Cold season 
Mean 288 149 430 636 343 293 
Max 452 256 822 832 467 414 
Min 170 78.6 225 451 195 203 
Range 282 178 598 381 272 212 
CV 0.23 0.30 0.27 0.13 0.19 0.16 
Cs 0.31 0.71 0.82 −0.19 −0.42 0.32 
  Streamflow (m3/s)
 
Precipitation (mm)
 
Year Warm season Cold season Year Warm season Cold season 
Mean 288 149 430 636 343 293 
Max 452 256 822 832 467 414 
Min 170 78.6 225 451 195 203 
Range 282 178 598 381 272 212 
CV 0.23 0.30 0.27 0.13 0.19 0.16 
Cs 0.31 0.71 0.82 −0.19 −0.42 0.32 

CV, coefficient of variation; Cs, coefficient of skewness.

Table 5 shows the MK test results for the entire period and for the cold and warm periods using monthly data of basin average precipitation and streamflow at Kaluga Station. Both precipitation and streamflow show increasing trends in all periods. Statistical significance occurs in terms of streamflow but not in precipitation.

According to Table 6, the annual precipitation had a distribution that was close to normal. Precipitation was positively skewed for the cold season and negatively skewed for the warm season. A possible cause of this difference is the strong influence of the North Atlantic Oscillation, which modifies the winter precipitation over European Russia (ER) (Kasimov & Kislov 2011).

The streamflow variations were small, both for the whole year and for individual seasons, as shown in Table 6. The variations in precipitation were considerably lower than those for the streamflow, as reflected by the CV and the coefficient of skewness. Almost three-quarters of the annual runoff occurred during the cold season, although during the cold season, only 46% of the annual precipitation occurred. This apparent contradiction is due to limited infiltration in the spring because the soil is frozen. The CV of streamflow for the warm season was higher than that for the cold season. This may be explained by the shorter duration of the warm season.

The probability distribution function (PDF) of seasonal streamflow exhibited significant positive skewness (0.71 and 0.82 for the warm and the cold seasons, respectively) while precipitation had negative skewness in the warm season and positive skewness in the cold season. The inconsistency in skewness of precipitation and streamflow indicates that precipitation is not the only factor that affects streamflow, particularly in the cold season.

Figure 5 shows the precipitation statistics for the Oka Basin. Figure 5(a) shows a relatively small and statistically insignificant growth (3.4%) in annual precipitation from 1965 to 2014. At the same time, we found evidence of increased variations in the annual precipitation. Namely, the first and third driest years were 2010 and 2014, despite the trend in the opposite direction. Figure 5(b) shows that precipitation during the cold season increased slightly by 4%. However, there was a significant change in the pattern of precipitation from 1965 to 2014, where snowfall decreased by 20%. We see from Figure 5(c) that precipitation increased during the warm season by 2.8%. However, this growth was not uniform. The growth rate was 4.1%/10 years for 1965–1996, but it decreased by 17.5%/10 years from 1997 to 2014. The probable reason for this effect is the weakening of zonal transfer in Russia that has occurred since the late 1990s.

Figure 5

Precipitation measured at the Oka Basin for the (a) whole year, (b) cold season, and (c) warm season.

Figure 5

Precipitation measured at the Oka Basin for the (a) whole year, (b) cold season, and (c) warm season.

Figure 6 shows the streamflow statistics at the Kaluga Station. Figure 6(a) shows the yearly streamflow data. The rate of increase in annual streamflow was more than 3.5 times higher than that of the precipitation. One reason for this is the reduction in potential evaporation and a slight increase in precipitation in the Oka Basin (Speranskaya 2016). Figure 6(b) shows the streamflow data for the cold season. Despite the increase in precipitation, the change in the streamflow was close to 0 (1.4%). Figure 6(c) shows the streamflow data for the warm season. The rate of increase in the streamflow exceeded that of precipitation by more than 15 times (by 43%). The reason for this change is an increase in the area of unfrozen soil during snowmelt, which increases meltwater infiltration (Kalyuzhny & Lavrov 2016). The intensified infiltration of meltwater during the cold season is responsible for the increased groundwater level at the beginning of the warm season. The groundwater level in the ER has risen 50–130 cm since 1950 (Shiklomanov 2008).

Figure 6

Streamflow measured at Kaluga Station for the (a) whole year, (b) cold season, and (c) warm season.

Figure 6

Streamflow measured at Kaluga Station for the (a) whole year, (b) cold season, and (c) warm season.

Teleconnection between precipitation in the ISSZ and CSFZ

There is a possible connection between the precipitation over an area and the atmospheric conditions in other places. Figure 7 shows a contour map for the correlation coefficients between the local monthly precipitation from the Sandimen Station and the world-wide grid-based sea level pressure (monthly mean). Figure 8 is similar and provides the average monthly precipitation for the Oka Basin. The grid-based sea level pressure data were obtained from the National Center for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) reanalysis data with a spatial resolution of 2.5° by 2.5°. The time period of the collected data extends from 1965 to 2014. As shown in Figure 7, there was a highly negatively correlated area near Taiwan (i.e., a correlation coefficient of −0.7). This indicates that the low sea level pressure around Taiwan was accompanied by a high amount of local monthly precipitation. In terms of atmospheric teleconnection, the Persian Gulf area and Siberia were also highly negatively correlated, and Euro-Russia was mildly negatively correlated with the local monthly precipitation of Taiwan. Figure 8 provides the results for the Oka Watershed. There was a mildly negatively correlated area near Euro-Russia (i.e., a correlation coefficient of −0.5). The Euro-Asia continent (including Taiwan) was also found to be mildly negatively correlated with the local monthly precipitation in the Oka Basin. The atmospheric teleconnection was subtle for Taiwan and Russia even though these areas are separated by a great distance. Figure 9 show the correlation functions for the precipitation of the Sandimen and Oka watersheds. The streamflow results are also shown. There was almost no time lag in the precipitation for the Sandimen and Oka watersheds, but there was a nine-month lag between the Aiolia River and the Oka River, which indicates that high regulation exists in the Oka River.

Figure 7

Contour map for the correlation coefficients of the Sandimen Station.

Figure 7

Contour map for the correlation coefficients of the Sandimen Station.

Figure 8

Contour map for the correlation coefficients of the Oka Basin.

Figure 8

Contour map for the correlation coefficients of the Oka Basin.

Figure 9

Correlation coefficient versus the lag time for the precipitation and streamflow.

Figure 9

Correlation coefficient versus the lag time for the precipitation and streamflow.

Characteristics of inhomogeneity and non-stationarity

Due to the impact of climate change, precipitation and streamflow vary not only in terms of their amount, but also in terms of their time-wise uniformity. We used MD and RMD to characterize the streamflow and precipitation distribution for the wet and dry seasons as well as for the whole year.

Figure 10 shows the results of the MD analysis. In Figure 10(a) and 10(b), the streamflow and precipitation curves exhibit similar patterns. It should be noted from the data from the last decade that the non-uniformity in the streamflow increased, while that of the precipitation decreased. In Figure 10(c), the MD streamflow and precipitation values are much smaller than those in Figure 10(a) and 10(b) because of the low precipitation and streamflow in the dry season. It should also be noted that the MD of the streamflow varied significantly in the last decade compared to that of precipitation.

Figure 10

Results of the MD analysis at Sandimen Station for the (a) whole year, (b) wet season, and (c) dry season.

Figure 10

Results of the MD analysis at Sandimen Station for the (a) whole year, (b) wet season, and (c) dry season.

Figure 11 shows the results of the RMD analysis. The precipitation looks stable and does not exhibit significant changes during the analyzed periods. In contrast, the streamflow for the dry season varies significantly, as shown in Figure 10(c).

Figure 11

Results of the RMD analysis at Sandimen Station for the (a) whole year, (b) wet season, and (c) dry season.

Figure 11

Results of the RMD analysis at Sandimen Station for the (a) whole year, (b) wet season, and (c) dry season.

To assess the dynamics of climate variations for the Ailiao watershed, we considered the time series for maximum daily precipitation, maximum daily streamflow, and maximum duration without precipitation for a given period. From Figure 12, it can be seen that the maximum daily precipitation increased by 3.94 mm/year. The maximum daily precipitation of over 800 mm occurred in recent years (2007, 2009, and 2010). The peak in 2009 was caused by Typhoon Morakot, setting the highest record for Taiwan in its recorded history. The typhoon caused flooding at lower altitudes while causing debris flow and landslide in the piedmonts. Figure 13 shows the maximum daily streamflow, where the trend shows a slight decrease. Erratic variations occurred after 1990. The peak in 2009, caused by Typhoon Morakot, induced serious flooding in southern Taiwan, with more than 600 dead.

Figure 12

Year-wise maximum daily precipitation at Sandimen Station.

Figure 12

Year-wise maximum daily precipitation at Sandimen Station.

Figure 13

Year-wise maximum daily streamflow at Sandimen Station.

Figure 13

Year-wise maximum daily streamflow at Sandimen Station.

Figure 14 displays the maximum duration without precipitation at the Sandimen Station, where there was an increasing trend. The longest maximum was 65 days, more than two months, which occurred in 2009, the same year as Typhoon Morakot. Oscillations in precipitation, streamflow, and the dry periods exacerbated, causing both water-related disasters and shortages in water resources.

Figure 14

Year-wise maximum duration without precipitation at Sandimen Station.

Figure 14

Year-wise maximum duration without precipitation at Sandimen Station.

Figure 15(a)15(c) show the MD statistics for the Oka Basin in Russia, where precipitation and the streamflow exhibit different trends. The MD for precipitation grew by 18%, which is significant at a p-value = 0.12%, while the MD for streamflow declined by 42%, with a change in the main phase occurring in the 1970s. Figure 15(b) shows the MD for precipitation and streamflow, where it can be seen that there was a rising trend for precipitation of 24% in the cold season and a declining trend in streamflow of 55%. This clearly shows that in the CSFZ, global warming has led to a significant decrease in the unevenness of streamflow during both the entire year and in the cold season. Figure 15(c) shows the MD for both precipitation and streamflow during the warm season in the Oka Basin. The MD of streamflow during the warm season rose by 23%. Since groundwater levels were higher at the beginning of the low water period, the groundwater recess rate increased, leading to an increase in MD. The MD of precipitation during all periods increased by 15% with a p-value = 1.4%, which was not very significant.

Figure 15

Results of the MD analysis at Oka Basin for the (a) whole year, (b) cold season, and (c) warm season.

Figure 15

Results of the MD analysis at Oka Basin for the (a) whole year, (b) cold season, and (c) warm season.

Figure 16(a)16(c) display the RMD statistics for the Oka Basin. From Figure 16(a), it can be seen that the variations in the RMD for the whole year are similar to those for the MD, but are smaller. As a result, a relatively small change in the RMD of precipitation (a growth of 8.5%) is highly significant, with a p-value = 1.4 × 10−5%. This result can be attributed to the decrease in continuous precipitation and the increase in convective precipitation. The reduction in the streamflow RMD was 50% since the spring flood flow decreased, and the groundwater flow increased. Figure 16(b) shows the RMD for the cold season in the Oka Basin. The streamflow RMD declined by 55%. Thus, it became difficult to separate the winter low-flow period and the spring flood period. The precipitation RMD increased by 8% for the cold season, as well as for the whole year. From Figure 16(c), it can be seen that the precipitation RMD for the warm season in the Oka basin grew by 11%. This may be explained by three factors: (1) independently of warming, convective precipitation formed during the warm season; (2) according to the Clausius–Clapyeron relation, the water vapor content of air increased noticeably during the warm season (convective precipitation was formed mainly from local water vapor); and (3) the growth of soil moisture during the warm season also contributed to an increase in convective precipitation. The streamflow RMD declined by 22% during the warm season. This was due to the increased proportion of groundwater recharge to rain-feeding.

Figure 16

Results of the RMD analysis at the Oka Basin for the (a) whole year, (b) cold season, and (c) warm season.

Figure 16

Results of the RMD analysis at the Oka Basin for the (a) whole year, (b) cold season, and (c) warm season.

Figure 17 shows the time variations in the maximum daily precipitation for the Oka Basin. The maximum daily precipitation grew by 19.5% from 1965 to 2014 (p-value = 2.4%). However, for the intensive warming period (1976–2014), there was no obvious trend. Despite the increase in the average precipitation, the maximum daily precipitation did not change. This may be explained by the significant decrease in the number of days with precipitation in the Oka Basin, especially since the late 1990s.

Figure 17

Year-wise maximum daily precipitation at Oka Basin averaging all stations.

Figure 17

Year-wise maximum daily precipitation at Oka Basin averaging all stations.

Figure 18 shows the maximum daily streamflow at Kaluga Station. There was a significant decreasing trend. The maximum daily streamflow reduced four-fold from 1965 to 2014. At the same time, the maximum snow water equivalence only reduced by 10%. The soil moisture at the beginning of the spring flooding period increased only slightly. Therefore, the main factor contributing to the reduction in the maximum streamflow was the increase in meltwater infiltration.

Figure 18

Year-wise maximum daily streamflow at Kaluga Station.

Figure 18

Year-wise maximum daily streamflow at Kaluga Station.

From Figure 19, it can be seen that the maximum duration of the dry periods increased by 48%. This was associated with more blocking anticyclones of higher intensity, especially during the summer. The maximum duration of the dry periods was 12–15 days, which was still not enough to create a water deficit.

Figure 19

Year-wise maximum duration without precipitation at Oka Basin averaging all stations.

Figure 19

Year-wise maximum duration without precipitation at Oka Basin averaging all stations.

Result comparison of Ailiao watershed and Oka Basin

The Ailiao watershed represents the ISSZ, which is with a small area with a complex and dissected relief while the Oka Basin is a typical continental watershed with a large area and flat relief. The area of Oka Basin is 54,900 km2, which is about 66 times the area of Ailiao watershed. The size of an area certainly affects the amount of precipitation and streamflow, but the dynamic responses to climate variations are representative for the associated climate zone.

The MK tests show that both Ailiao watershed and Oka Basin exhibited an increasing trend in precipitation while the trends of streamflow in the two areas were opposite. MD and RMD of streamflow also show that non-uniformity increased in the Ailiao watershed but decreased in the Oka Basin. This indicates that different mechanisms are driving the streamflow response in CSFZ and ISSZ. However, similar characteristics were also found in Ailiao watershed and Oka Basin. Both showed increased maximum daily precipitation, decreased maximum daily streamflow, and increased maximum duration of the period without precipitation for the whole year.

Climate variations for Ailiao and Oka basins are similar – increases in air temperature, precipitation, especially extreme precipitation. However, the leading streamflow formation factors in the two basins differ significantly. The peculiarities of the precipitation (pronounced wet and dry seasons, heavy rainfall) and geomorphological structure (large catchment slope) are related to the fact that the Ailiao streamflow depends strongly on precipitation. Thus, almost 50% of the annual flow dispersion is due to annual precipitation dispersion (the Pearson correlation coefficient between them is 0.69), while for Oka, this value is only 23%. As already mentioned, annual precipitation and river flow have increased for Ailiao and Oka basins over the study period. However, in the Oka Basin, runoff growth is ahead of precipitation growth (2.45%/10 years against 0.675%/10 years), while for Ailiao Basin, the situation is reversed (0.053%/10 years against 0.159%/10 years). If water balance is taken into account for a long-term period, the changes in evaporation and anthropogenic activity are less profound than the increase in precipitation for the Oka Basin while this is reversed for Ailiao watershed.

CONCLUSIONS

The impacts of variations in climate on precipitation and streamflow were investigated in the CSFZ and ISSZ. Russia and Taiwan were taken as examples of these two zones, respectively. Our research led to the following conclusions.

In Taiwan, precipitation and streamflow data in the Ailiao watershed, where precipitation is the only source of fresh water, were utilized to explore the impact of climate variation. Essentially, the amount of precipitation remained stable for the entire year over the study period and in both wet and dry seasons. However, the maximum daily precipitation and duration of dry periods increased. Erratic streamflow has occurred in recent years in both the wet and dry seasons. Record maximum precipitation and streamflow and record dry period durations all occurred only in 2009. The streamflow appeared to be more heterogeneous than the precipitation. Both precipitation and streamflow have significantly fluctuated, especially in the last decade. Climate variation significantly changes the temporal distribution of precipitation and streamflow, which in turn, induces challenges related to the prevention of water-related disasters and the maintenance of sufficient water resources.

In Russia, there was a significant increase in precipitation in both the cold and warm seasons in the Oka Basin from 1965 to 2014. However, since the late 1990s, the trend in the precipitation in the cold season has remained positive, but precipitation in the warm season has shown a tendency to decrease. Because the results indicate that the precipitation slightly increased and that the number of days with precipitation decreased, this resulted in increases in unevenness in terms of precipitation frequency. The increase in the maximum daily precipitation in the last two decades has been replaced by their stabilization. In the streamflow results, due to the climatic changes that have occurred in recent decades, the annual distribution of the streamflow in the Oka Basin has become more regular, with rises in soil temperature and humidity, increased winter thaws, and improved conditions for snowmelt infiltration. This explains the increases in the underground recharge of rivers and the increases in the minimal runoff that have resulted. Simultaneously, increased losses in infiltration during high water has markedly reduced the risk of floods. An increase in the minimum streamflow in a densely populated region contributes to a reduction in the degree of pollution by diluting wastewater.

As the result of the comparison of hydrological data from both Taiwan and Russia, it was found that the changes in precipitation in Taiwan and Russia are similar. The correlations in global atmospheric conditions between Taiwan and Russia were explored. The teleconnection of the precipitation between Taiwan and Russia is subtle and is linked to global atmospheric conditions. There is no time lag in precipitation between Taiwan and Russia, but there is a nine-month time lag in streamflow.

The maximum daily precipitation and the maximum duration without precipitation were shown to increase in both the frigid and subtropical climate zones. However, the streamflow regime of the temperate latitudes of Russia and Taiwan were quite different. The maximum daily precipitation and the maximum duration without precipitation were shown to increase in both Taiwan and Russia. The streamflow in Taiwan became more extreme, where the minimum flow was reduced, and the maximum flow increased, followed by an increase in the duration of the period without precipitation and the growth of the maximum daily precipitation. The streamflow in Taiwan, which has pronounced dry and wet seasons, has become more uneven. On the other hand, due to the significant regulating capacity of the watershed and its large area, the flow regime of the Oka River depends little on extreme manifestations of precipitation (periods without precipitation or extreme rainfall), but rather is formed in the process of snow accumulation and melting. The increase in winter temperatures has led to a reduction in the maximum water flow and an increase in the minimum by tens of percentages. Thus, the regime of the runoff of the rivers in the central strip of Russia, which is distinguished by uniformity, has become even more uniform. Extreme hydrological situations (floods, low water levels) have become rarer. Thus, the responses of streamflow to recent climate variations in the frigid zone (Russia) and subtropical zone (Taiwan) are very different due to different driving forces.

ACKNOWLEDGEMENTS

The study of the Oka Basin was supported by an RFBR grant (project No. 16-55-52008 MNT_a). The study of the Ailiao watershed was supported by a grant from the Ministry of Science and Technology, Taiwan (MOST) (project No. 105-2923-M-006-004-MY3). The authors gratefully acknowledge the constructive comments of the anonymous reviewers and handling editor.

REFERENCES

REFERENCES
Alley
R. B.
,
Marotzke
J.
,
Nordhaus
W. D.
,
Overoect
J. T.
,
Peteet
D. M.
,
Pielke
R. A.
Jr.
,
Pierrehumbert
R. T.
,
Rhines
P. B.
,
Stocker
T. F.
,
Talley
L. D.
&
Wallace
J. M.
2003
Abrupt climate changes
.
Science
299
(
5615
),
2005
2010
.
Barnett
J.
,
Lambert
S.
&
Fry
I.
2008
The hazards of indicators: insights from the Environmental Vulnerability Index
.
Annals of the Association of American Geographers
98
,
102
119
.
Dzhamalov
R. G.
,
Frolova
N. L.
,
Kireyeva
M. B.
,
Rets
E. P.
,
Safronova
T. I.
,
Bugrov
A. A.
,
Telegina
A. A.
&
Telegina
E. A.
2015
Modern Resources of Underground and Surface Waters of the European Russia: Formation, Distribution, use
.
GEOS
,
Moscow
.
Frolova
N. L.
,
Agafonova
S. A.
,
Kireeva
M. B.
,
Povalishnikova
E. S.
&
Pakhomova
O. M.
2017
Recent changes of annual flow distribution of the Volga basin rivers
.
Geography, Environment, Sustainability
10
,
28
39
.
Georgiadi
A. G.
,
Koronkevich
N. I.
,
Milyukova
I. P.
,
Kashutin
E. A.
&
Barabanov
E. A.
2014
Modern and Scenario Changes of River Runoff Within Russian Largest Rivers Basins. Part 2. Watersheds of the Volga and the Don Rivers
.
MAKS Press
,
Moscow
.
Günter
B.
,
Julia
H.
&
Juraj
P.
,
& 40 coauthors
.
2017
Changing climate shifts timing of european floods
.
Science
357
,
588
590
.
Hartmann
D. L.
,
Klein Tank
A. M. G.
,
Rusticucci
M.
,
Alexander
L. V.
,
Brönnimann
S.
,
Charabi
Y.
,
Dentener
F. J.
,
Dlugokencky
E. J.
,
Easterling
D. R.
,
Kaplan
A.
,
Soden
B. J.
,
Thorne
P. W.
,
Wild
M.
&
Zhai
P.
2013
Observations: Atmosphere and Surface
. In:
Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change
.
Cambridge University Press
,
Cambridge
.
IPCC
2007
Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change (IPCC)
.
Cambridge University Press
,
Cambridge and New York
.
IPCC
2013
Climate Change 2013: Fifth Assessment Report of the Intergovernmental Panel on Climate Change
.
Cambridge University Press
,
Cambridge and New York
.
Kalyuzhny
I. L.
&
Lavrov
S. A.
2012
Hydrophysical Processes in the Catchment Area: Experimental Research and Modeling
.
Nestor-Istoriya
,
St. Petersburg
.
Kalyuzhny
I. L.
&
Lavrov
S. A.
2016
Effect of climate changes on the soil freezing depth in the Volga River basin
.
Ice and Snow
56
,
207
220
.
Kasimov
N. S.
&
Kislov
A. V.
2011
Ecological and Geographical Consequences of Global Warming in the 21st Century on the East European Plain and in Western Siberia
.
MAX Press
,
Moscow
.
Kendall
M. G.
1975
Rank Correlation Methods
.
Griffin
,
London
.
Kundzewicz
Z. W.
,
Mata
L. J.
,
Arnell
N.
,
Döll
P.
,
Kabat
P.
,
Jiménez
B.
,
Miller
K.
,
Oki
T.
,
Şen
Z.
,
Shiklomanov
I.
2007
Freshwater resources and their management
. In:
Climate Change 2007: Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change
(
Parry
M. L.
,
Canziani
O. F.
,
Palutikof
J. P.
,
van der Linden
P. J.
&
Hanson
C. E.
, eds).
Cambridge University Press
,
Cambridge
, pp.
173
210
.
Liu
S. C.
,
Fu
C.
,
Shi
C. J.
,
Chen
J. P.
&
Wu
F.
2009
Temperature dependence of global precipitation extremes
.
Geophysical Research Letters
36
,
L17702
.
Mann
H. B.
1945
Nonparametric tests against trend
.
Econometrica
13
,
245
259
.
Milly
P. C. D.
,
Betancourt
J.
,
Falkenmark
M.
,
Hirsch
R. M.
,
Kundzewicz
Z. W.
,
Lettenmaier
D. P.
&
Stouffer
R. J.
2008
Stationarity is dead: whither water management
.
Science
319
,
573
574
.
Moscow-Oka Basin Water Administration of Federal Agency for Water Resources
2015
Scheme of Complex use and Protection of Water Objects
.
Moscow
(in Russian)
.
Naghettini
M.
2017
Fundamentals of Statistical Hydrology
.
Springer International Publishing
,
Switzerland
.
Shiklomanov
I. A.
2008
Water Resources of Russia and Their use
.
State Hydrological Institute
,
St. Petersburg
.
Soden
B. J.
&
Held
I. M.
2006
An assessment of climate feedbacks in coupled atmosphere–ocean models
.
Journal of Climate
19
,
3354
3360
.
The Federal Service for Hydrometeorology and Environmental Monitoring of Russia (Roshydromet)
2014
Second Roshydromet Assessment Report on Climate Change and its Consequences in the Russian Federation
.
Moscow
(in Russian)
.
The Federal Service for Hydrometeorology and Environmental Monitoring of Russia (Roshydromet)
2016
Resources of Surface and Underground Waters, Their Use and Quality
.
Moscow
(in Russian)
.
Voskresenskaya
E.
&
Vyshkvarkova
E.
2016
Extreme precipitation over the Crimean Peninsula
.
Quaternary International
409
,
75
80
.
Zhang
Q.
,
Gu
X.
,
Singh
V. P.
,
Xu
C. Y.
,
Kong
D.
,
Xiao
M.
&
Chen
X.
2015
Homogenization of precipitation and flow regimes across China: changing properties, causes and implications
.
Journal of Hydrology
530
,
462
475
.