Influence of a dual monsoon system and two sources of groundwater recharge on Kofu basin alluvial fans, Japan

Environmental tracers such as oxygen and hydrogen stable isotopes (δ¹⁸O and δD) are commonly used in hydrological investigations, e.g., to estimate the groundwater recharge, to identify multiple recharge sources, and to investigate the interactions between the surface and groundwater, such as the flow paths and mixing in regional aquifers, advection/diffusion in groundwater aquifers, and the effects of evaporation on groundwater systems (Deshpande et al. 2003; Palmer et al. 2007). The values of δ¹⁸O and δD remain constant as long as phase changes or fractionation does not occur along the flow path (Senturk et al. 1970; Perry et al. 1980; Clark & Fritz 1997). However, stable isotope fractionation occurs when the water changes its state (e.g., through evaporation or condensation, raindrop formation, and infiltration in aquifers). Consequently, the isotopic composition varies between the different forms of water (i.e., precipitation, surface water, and groundwater). A comparative analysis of the δ¹⁸O and δD composition and spatiotemporal distribution of precipitation, river water, and groundwater provides a useful tool for evaluating recharge mechanism and defining the status of mixed groundwater recharge zones (Chen et al. 2006; Mukherjee et al. 2007; Negrel et al. 2011; Yeh et al. 2011; Liu & Yamanaka 2012; Peng et al. 2012).

To investigate groundwater recharge through stable isotope analysis, the measured δ¹⁸O is plotted against δD, producing the ‘local meteoric water line’ (LMWL), which is then compared with the ‘global meteoric water line’ (GMWL, Craig 1961; Gat 1996). The deviations and their characteristics (e.g., slope and intercept) of the LMWL from the GMWL may be used to infer the groundwater recharge mechanism. The LMWL may vary due to evaporation and condensation (e.g., temperature and humidity), terrain environment, altitude, and other reasons. The intercept is also called deuterium excess or d-excess (d = δD − 8.0 δ¹⁸O) (Dansgaard 1964). A globally representative d-excess is 10‰ (Craig 1961); however, it may vary depending on the evaporation conditions in the origin of the water source. For example, d-excess is 6.03 in North America and 3.97 in the Tropical Island area (Gat 1996). In Japan, d-excess in different seasons is reported to vary from 14.8 in April–September to 23.8 in October–March, even though the LMWL slope remains approximately the same (Liu & Yamanaka 2012). Comparing the differences in the stable isotopic signatures of groundwater, river water, and local precipitation allows the effect of different sources of groundwater recharge to be identified (Scanlon et al. 2002; Kalbus et al. 2006), whereas comparing the d-excess values of groundwater and precipitation between the different seasons allows identification of the effect of the dual monsoon (Yeh et al. 2011).

Rainwater characteristics in Japan are controlled by monsoon (summer, May–October) and pre-monsoon precipitation (winter, November–April). These two seasonal circulation systems are a part of the larger monsoon circulation in South Asia, and they are caused by seasonal temperature and pressure gradient reversal, which is associated with the wind circulation following the annual northward and southward motion of the sun. Normal precipitation d-excess values for the pre-monsoon and monsoon periods are greater than 20‰ and less than 10‰, respectively (Nakayama et al. 2000). Such seasonal variation is caused by the contribution of the dual moisture sources predominantly from the Pacific Ocean in summer with SE monsoon winds and predominantly from the Sea of Japan with NE monsoon winds in winter (Waseda & Nakai 1983; Araguás-Araguás et al. 1998). Yeh et al. (2011) estimated seasonal contributions of precipitation to groundwater recharge in the Chih-pen Creek basin of eastern Taiwan. However, very few studies that have examined the aspects have investigated both the recharge contribution from different sources and its temporal variability.

Groundwater is an important source of water for industrial, agricultural, and drinking water use in Kamanashgawa and Midaigawa alluvial fans, located in Kofu basin, because this rural area has no dams for surface water use. This basin is located in inland Japan and the distances from the Pacific Ocean and Sea of Japan are about 70 km and 160 km, respectively. The rainwater characteristics in this area are controlled by precipitation during the summer (SE) and winter (NE) monsoons. The SE monsoon operates during the months of May–October and NE pre-monsoon during the months of November–April. Therefore, in a local area such as the Kofu basin, the d-excess for the different seasons is expected to differ from the general values of Japan.

In this context, the study aims to identify the local oxygen and hydrogen stable isotopic characteristics and the relative contribution of the two groundwater sources (precipitation infiltration and river water recharge) and the dual monsoon system (pre-monsoon and monsoon seasons) on groundwater recharge in Kamanashi and Midai alluvial fans of the Kofu basin in the central part of Japan.

The Midaigawa and Kamanashigawa alluvial fans (area = 90 km²) were formed in the western Kofu basin (Figure 1) due to sand and gravel deposition by the Midai and Kamanashi River, transported from steep mountain areas. The central and upper parts of the Kamanashigawa alluvial fan consist of gravel, with a depth of 75–95 m; the lower part is composed of clay soil, covered with gravel, with a depth of 15–25 m. The Midaigawa alluvial fan is formed of gravel, transported by the Midai River, without any clay soil. The Midai and Kamanashi riverbeds are raised above the Midaigawa and Kamanashigawa alluvial fans, due to sediment deposition inside the riverbanks (MLIT 2001).

Kofu basin lies in an inland region; therefore, its temperature exhibits extreme variations between summer and winter. The observed metrological data of the Japan Metrological Agency (JMA) (2008) indicates that typically, summers are hot and humid, and winters are cold, with average temperatures of 26 °C and 3 °C, respectively. Annual rainfall in Kofu basin is as little as 1,135 mm in the lower areas, but reaches up to 2,500 mm in the high stands. The entire basin receives a mean annual precipitation of approximately 2,100 mm; approximately 75% of the annual precipitation occurs during the monsoon season from May to October.

Monthly precipitation samples were collected in the period August 2003–July 2004 at a station (elevation = 250 m) in the Midaigawa alluvial fan (Figure 1, P1). Additionally, to determine the LMWL in this area, event-wise precipitation samples were acquired in the period March 2007–February 2008 at a station (elevation = 308 m) in University of Yamanashi premises, located approximately 6 km east of the Kamanashigawa alluvial fan (Figure 1, P2). All the precipitation samples were collected in polyethylene plastic bottles attached to funnels with an anti-evaporation cap. River water and groundwater samples were collected between January and December 2007 for chemical and isotopic (oxygen and hydrogen) analyses. Sampling was carried out during both wet and dry periods on a bimonthly basis from the rivers (Midai and Kamanashi) and from domestic wells (depth being 20–50 m) (Figure 1).

Hydrogen and oxygen isotope ratios were expressed as δD and δ¹⁸O, respectively, where δ = [(R_sample/R_standard) −1] × 1000 (‰), and R is D/H or ¹⁸O/¹⁶O in the sampled water (R_sample) or standard mean ocean water (R_standard). The δ¹⁸O and δD were analyzed using an isotope ratio mass spectrometer (IRMS) (Sercon, ANCA20-20, UK) after establishing equilibrium with CO₂ gas for ¹⁸O and H₂ gas for D. Each sample gas was extracted using an auto-sampling system (Sercon, WES, UK). The analytical precision was 0.1‰ for δ¹⁸O and 1‰ for δD.

On the contrary, d-excess shows distinct seasonal variability. It generally ranges from 10 to 23‰. Relatively higher d-excess values (from 15 to 23‰) are detected corresponding to the pre-monsoon seasons (October–April 2003/2004), whereas they are low (10 to 11‰) for the monsoon seasons (August–September 2003 and May–July 2004) at location P1 (Figure 2(a)). The d-excess value of event-wise precipitation samples also suggests seasonal variability; pre-monsoon seasons (March 2007–April 2007 and October 2007–February 2008) are associated with a high frequency of increased d-excess values (>10‰), whereas the monsoon season (May–September 2007) exhibits a high-frequency of low d-excess values (<10‰) at location P2 (Figure 2(b)). These results are consistent with data reported by other researchers (e.g., Asano et al. 2002; Lee & Kim 2007; Yeh et al. 2011), which indicated that precipitation d-excess in Central Japan, Korea, and Eastern Taiwan has a distinct seasonal variability, high (>15‰) in the dry season and low in the wet season (<10‰).

The bimonthly river water δ¹⁸O and δD from both alluvial fans does not present significant temporal variations relative to the values measured in the precipitation samples. The δ¹⁸O (δD) is essentially constant in the Midai River at a value of −11.3 ± 0.12‰ (−77 ± 1.1‰) upstream (Table 2, R1) and at −10.9 ± 0.14‰ (−75 ± 1.6‰) downstream (Table 2, R2). In the case of the Kamanashi River, the values are −11.0 ± 0.13‰ (−76 ± 0.9‰) upstream (Table 2, R3), −10.9 ± 0.07‰ (−76 ± 0.7‰) middlestream (Table 2, R4), and −11.0 ± 0.14‰ (−76 ± 0.7‰) downstream (Table 2, R5). Spatial variations of δ¹⁸O and δD are insignificant for both rivers, probably due to lack of inflow. Additionally, δ¹⁸O and δD are less than the mean values in the precipitation samples (δ¹⁸O = −9.5‰ and −10.1‰, δD = −62.5‰ and −7.2‰). The fact that the alluvial fans are located in a range of altitudes (200 to 500 m), lower than those of the average catchment, suggests that altitude may affect δ¹⁸O and δD values. The river water d-excess value (11–13‰, Table 2) is higher than the annual precipitation d-excess values (9 and 10‰).

The results of the present study illustrate that precipitation values δ¹⁸O and δD exhibit a wide temporal variability (Figure 2), unlike river water and groundwater δ¹⁸O and δD values (Table 2). The isotopic fractionation by the rainfall recharge significantly affects the groundwater isotopic values. However, the amplitude of the seasonal fraction for the isotopic values in groundwater is expected to decrease with increasing residence time (Maloszewski & Zuber 1993; Asano et al. 2002). Asano et al. (2002) reported the relationship between the estimated residence time of the spring water and amplitude of the isotopic fractionation. It shows, in the case of groundwater with more than 1 year residence time, the isotopic amplification pattern will not have a significant regression. In addition, Mizutani et al. (2001) hypothesized that if the groundwater system is predominantly recharged through infiltration of precipitation, the groundwater isotopic composition will directly reflect the isotopic values of precipitation.

The two-river water (Midai and Kamanashi River) is represented by depleted isotopic composition rather than the weighted average of isotopic composition of precipitation at plain areas. The maximum altitudes of the Midai and Kamanashi River watersheds are approximately 2,400 m and 2,500 m, respectively. Therefore, it is possible that the different isotopic values between river water and precipitation result from this altitude difference. This hypothesis is further corroborated by previously reported evidence such as the depletion in surface water annual mean δ¹⁸O and δD values compared with the precipitation weighted mean in alluvial fans; and the decreasing of surface water ¹⁸O and D content with increasing altitude (Dansgaard 1964; Siegenthaler & Oeschger 1980; Poage & Chamberlain 2001; Yeh et al. 2011; Liu & Yamanaka 2012). Waseda & Nakai (1983) also indicated the altitude effect as an important factor determining the surface and meteoric waters’ isotopic compositions in central and northeast Japan, and they reported an average altitude effect on surface water δ¹⁸O and δD values in those areas as −0.25‰ and −2.0‰ per 100 m, respectively.

In the plain area, river water d-excess exhibits a higher value (11–13‰) than the annual precipitation (9 and 10‰), suggesting the evaporation effects on different elevations or two air-masses contributed to the precipitation event, one with a monsoon precipitation (low d-excess) and one with a pre-monsoon precipitation (high d-excess).

The d-excess is defined as the excess deuterium that cannot be accounted by equilibrium fractionation between water and vapor. Since condensation is most often an equilibrium process, d-excess is an indicator of kinetic fractionation during evaporation, governed by molecular diffusivity of isotopic molecular species (Dansgaard 1964; Clark & Fritz 1997). Although temperature and wind speed can influence the kinetic fractionation, relative humidity is the most important factor. The d-excess of the GMWL has a value of ∼10 representing evaporation at ∼85% relative humidity. However, the d-excess value in the regional precipitation can be greater than ten if the evaporation in the source region takes place under lower humidity (Gat & Carmi 1970).

The d-excess in precipitation was generally higher in winter (d > 10) and lower in summer in Japan. Dansgaard (1964) and Waseda & Nakai (1983) have stated that the extremely high d-excess that appeared in Japanese precipitation during the winter season resulted from rapid evaporation induced by dry, continental air masses brought from the Sea of Japan. In the case of the meteoric condition, the winter season dominates NE pre-monsoon winds. The resultant precipitation will have high d-excess in Japan. Katsuyama et al. (2015) have studied a seasonal variation of d-excess values of precipitation at three locations crossing a main island of Japan; Sea of Japan side (Tottori prefecture), Sea of Japan side (Shiga prefecture), and Pacific Ocean side (Nara prefecture). The climate conditions are clearly different among these stations. In the Sea of Japan side (Tottori prefecture), much snow falls from December to March with low air temperatures. In Shiga, there is less snowfall but much more rainfall during summer. Summer rainfall is more plentiful in the Pacific Ocean side (Nara prefecture). However, similar sinusoidal d-excess variations in precipitation were repeated at these three stations, i.e., higher during winter and lower during summer. The sinusoidal pattern is caused by the contribution of the dual moisture sources predominantly from the Pacific Ocean in summer and predominantly from the Sea of Japan in winter (Waseda & Nakai 1983; Araguás-Araguás et al. 1998). Tase et al. (1997) also reported that this seasonal pattern was commonly observed at six stations in Kanto, Shikoku, and Kyushu regions in Japan. Hence, it can be inferred that dominant moisture sources are a significant influence on the seasonal variation of d-excess values in precipitation in our study area as well.

Various recharge sources for alluvial aquifer systems are identified by groundwater and recharge isotopic and chemical characteristics (Vanderzalm et al. 2011). Plotting δ¹⁸O versus δD (Figure 5) shows that groundwater samples from the Midaigawa and Kamanashigawa alluvial fans fall between river water and weighted average precipitation, indicating a possibility of recharge from both sources and their mixing in the aquifer.

The greater negativity of the δ¹⁸O in the shallow groundwater samples toward the Midai River side in the Kamanashigawa alluvial fan (Figure 4) and the virtually identical δ¹⁸O in the groundwater and river water toward the riverside of the Midai and Kamanashi Rivers in the Midaigawa alluvial fan, both suggest groundwater recharge from the rivers. On the other hand, samples most enriched in δ¹⁸O and δD, having values of −9.9‰ and −69‰, respectively, are found in the groundwater (G34), well located at an altitude (490 m) higher than the riverbed perimeter (460 m). These values are close to the annual mean δ¹⁸O and δD values of precipitation in the Midaigawa alluvial fan (−9.5‰ and −66‰), suggesting that precipitation is the primary source of groundwater at G34. Such high δ¹⁸O values of groundwater were also detected in the eastern part of the Kamanashigawa alluvial fan and southern part of the Midaigawa alluvial fan. This isotopic distribution suggests that the groundwater in this area might be recharged through precipitation infiltration.

The present study reveals the influence of the presence of two sources and seasonal variability on groundwater recharge in the alluvial fans of the Kofu basin, Japan. The alluvial fans are recharged from two sources: the river water drained by mountainous catchments and direct infiltration of precipitation in the plain area. The isotopic data indicate that river waters are isotopically lighter, whereas altitude effects become evident in the precipitation isotope values. In the Kamanashigawa alluvial fan, river water contributes 38–100% of the recharge, while precipitation contributes 29–62% in the central part and increases its contribution with distance from the Kamanashi River. In the case of Midaigawa, river water contributes 77–99% in the northern part, while in the southern side 30–93% of contribution comes from precipitation. Comparing the d-excess with δ¹⁸O values reveals mixing of the river water with precipitation in the alluvial fan aquifer. Variation in precipitation and river water d-excess values, in the plains of the study area, suggests that pre-monsoon precipitation in the mountain areas (e.g., in the form of snow) might contribute more in recharge. The mass-balance analysis of the d-excess in groundwater and precipitation of different seasons (i.e., pre-monsoon and monsoon) indicates pre-monsoon precipitation contributes 46–68% and 39–65% to the groundwater recharge of Kamanashigawa and Midaigawa alluvial fans, respectively.

Notably, the pre-monsoon precipitation contribution to the groundwater recharge is more than 39%, while the total precipitation on the plain area in the same season is only 25% of the annual precipitation. This requirement could be met by pre-monsoon precipitation in the mountain area, which is carried by the river and recharged in the form of river water. Therefore, even if the annual precipitation amount in the plain area of the alluvial fans is quite small, the pre-monsoon precipitation could play an important role in recharging the groundwater, especially in the case of mountainous watersheds.

We are grateful to all members of the Water Quality Research Group of Interdisciplinary Center For River Basin Environment, University of Yamanashi. This study was supported by the Global Centres of Excellence (GCOE) program and Grant-in-Aids for Scientific Research (C) (No. 25630226) of the Japan Society for Promotion of Science (JSPS).