Relationship between catchment scale and the spatial variability of stream discharge and chemistry in a catchment with multiple geologies

Transcription

1 Hydrological Research Letters 7(2), (2013) Published online in J-STAGE ( doi: /HRL.7.12 Relationship between catchment scale and the spatial variability of stream discharge and chemistry in a catchment with multiple geologies Tomohiro Egusa, Nobuhito Ohte, Tomoki Oda and Masakazu Suzuki Graduate School of Agricultural and Life Sciences, The University of Tokyo, Japan Abstract: This study investigated whether the representative elementary area (REA) concept can be adopted in catchments with multiple geologies. We observed stream discharge at 65 points and water chemistry at 157 points in a 55 km 2 catchment that included multiple geologies. At observation points with uniform geology, stream chemistry became constant beyond about 1 km 2 in granodiorite and volcanic rocks. At observation points with multiple geologies, spatial variability remained large beyond a few square kilometers. SiO 2 and Mg 2+ concentrations became constant above 10 km 2, but Ca 2+ and electrical conductivity did not become constant until 55 km 2. Our calculations revealed that for areas of 1 17 km 2, almost all of the observed variables were explained by mixing based on geological percentages. However, at greater than 17 km 2, the observed values were higher than the calculated values. Therefore, in regions with multiple geologies, the range of the REA with singleparameter geology was confirmed. In our catchments, the REA concept was applicable to areas of 1 17 km 2, but areas larger than 17 km 2 was outside the range. KEYWORDS spatial variability; stream water; catchment scale; water chemistry; bedrock geology; REA concept INTRODUCTION Correspondence to: Tomohiro Egusa, Graduate School of Agricultural and Life Sciences, The University of Tokyo, Yayoi, Bunkyo-ku, Tokyo , Japan. egusa@fr.a.u-tokyo.ac.jp 2013, Japan Society of Hydrology and Water Resources. One of the grand challenges of catchment hydrology is predicting stream discharge and chemistry in ungauged or poorly gauged catchments (Sivapalan, 2003a). For this purpose, we must have sufficient knowledge of the hydrological and biogeochemical processes in entire largescale catchments. However, even within small-scale catchments, hydrological properties often have large spatial variability, produced by heterogeneities of topography, soils, geology, land use, and so forth (McDonnell et al., 2007). It is difficult to conduct process studies in every small catchment forming a target large catchment. Therefore, we need to develop concepts and methods for understanding the hydrological processes of large-scale catchments based on limited information from observations of small catchments (Blöschl, 2001; Sivapalan, 2003b). In the last three decades, many hydrologists have tried to address this problem by understanding the relationships between catchment area and hydrological processes, and some have focused on the process by which repeated stream confluences result in progressively reduced spatial variability as catchment area increases. In a theoretical study, Wood et al. (1988) presented the representative elementary area (REA) concept. They suggested that there exists a certain range of catchment area in which the spatial variability among small catchments becomes small and can be ignored, and they called this range of catchment area the REA. Subsequently, empirical research conducted in several catchments confirmed the existence of the REA. However, most of these studies targeted the lower bound of the REA and did not clarify the upper bound. Hereafter, we use REA values to refer to the lower bound of the REA and range of the REA to indicate the range. In addition, REA values have been found to vary among catchments. For example, Woods et al. (1995) observed stream discharge at two catchments and confirmed convergence above 0.5 km 2 and 2km 2, respectively. Wolock et al. (1997) showed that convergence occurred above 3 km 2 based on six variables of water chemistry. Likewise, Temnerud and Bishop (2005) confirmed that convergence was found for areas larger than 15 km 2 at two catchments from four variables of water chemistry. Asano et al. (2009), on the basis of discharge and eight variables of chemistry, reported convergence at sizes ranging from km 2. The soils and climates in these studies differed among studied catchments and multiple soil types sometimes existed in a catchment. It was therefore difficult to extract the influence of each individual factor, such as climate, geology, or soil, from previous studies reporting values of REA. Differences in bedrock geology affect rainfall-runoff processes owing to differences in geophysical characteristics (Hattanji and Onda, 2004). It is therefore important to clarify the influence of geology on the spatial variability of stream water discharge and chemistry and REA values in low-flow conditions. In addition, most previous studies were conducted in catchments with uniform geology (Woods et al., 1995; Wolock et al., 1997; Temnerud and Bishop, 2005; Asano et al., 2009). Particularly in North America, where most previous studies were conducted, the geology is often homogeneous over a wide area (Blöschl and Sivapalan, 1995). By contrast, in Japan, the geological settings are complex and the geological boundaries are not always the same as watershed boundaries. Consequently multiple geologies often exist within meso-scale (about km 2 ) catchments. In these catchments, the REA concept has not been confirmed. In the range of the REA, only minimum knowledge of the heterogeneities such as topography, soils, geology, and land use is needed for adding or dividing subcatchments (Wood et al., 1988). To then verify the applicability of the REA concept in meso-scale catchments Received 27 November, 2012 Accepted 28 March,

2 SPATIAL VARIABILITY OF STREAM WATER with multiple geologies but uniform soil and land use, we need to determine whether the spatial variability of discharge and chemistry can be explained by mixing based on geology percentages. We observed stream discharge and chemistry in a 55 km 2 catchment with multiple geologies but uniform soil and land use, and examined the relationships between catchment scale and stream discharge and chemical parameters. Our objectives were to better understand the influence of a difference in geology on the REA values, and to confirm the REA concept in catchments with multiple geologies. METHODS Site description The Yozukugawa catchment ( 'N, ' 139 2'E) is located in the Tanzawa Mountains in central Japan. The 55.6 km 2 Yozukugawa catchment is underlain by ten different geologies (Figure 1). The Tanzawa Mountains consist of a central plutonic body surrounded by marine-derived volcanic tuffs and clastic rocks including conglomerates. The plutonic body is thought to be a magmatic intrusion. The volcanic tuffs are thought to have formed in an accretionary prism, and the clastic rocks are thought to be trough-fill sediments (Amano et al., 2007). The northern part of the catchment is underlain by plutonic rocks: granodiorite (42.1% of the catchment), gabbro (5.5%), Figure 1. Geological map and observation points in the Yozukugawa catchment. Blue solid circles, black solid circles, green solid triangles, and red solid squares denote category 1 (granodiorite), category 2 (volcanic rocks), category 3 (felsic plutonic rocks, mafic plutonic rocks, and non-marine sedimentary rocks), and category 4 (multiple geologies), respectively. and granitoid plutonic rocks (5.8%). In the southern part of the catchment, basaltic and andesitic volcanic rocks (Early to Middle Miocene non-alkaline mafic; 20.0%, Middle to Late Miocene non-alkaline mafic; 7.7%, non-alkaline felsic; 3.9%, and non-alkaline felsic; 3.1%) occur. In addition, tephra layers occur in some mountain ridges. The boundaries of the geology differ from the watershed boundary. Forests cover the entire catchment and consist mainly of coniferous plantations of Japanese cedar (Cryptomeria japonica) and cypress (Chamaecyparis obtusa). Japanese cypress accounts for 80% of conifers. A natural broadleaf forest of Fagaceae (Fagus crenata) also exists at high altitude. A typical brown forest soil occupies 90% of the catchment, with Andosol occupying the remaining 10%. The altitude ranges from 340 to 1375 m above sea level, and the mean slope gradient is The mean annual precipitation for the period was 2281 mm at Tanzawako, which is 3 km east of Yozukugawa, and the mean annual temperature was 9.0 C at Yamanaka, 8 km west of Yozukugawa. The precipitation and temperature were measured at the respective stations by the Automated Meteorological Data Acquisition System (AMeDAS) of the Japan Meteorological Agency. Sampling and measurements We conducted snapshot sampling of stream water on nine low-flow days in 2008: August 10 12, September 25 26, and October We measured discharge at 65 points and water chemistry at 157 points; 25 points were observed repeatedly. We measured electrical conductivity (EC) in situ. The concentrations of Ca 2+ and Mg 2+ were analyzed in the laboratory of the University of Tokyo using ion chromatography (Shimadzu LC10-A), and the concentration of SiO 2 was analyzed using the molybdenum yellow method. In addition, the accuracy of the analysis was confirmed by comparing observed and calculated values of EC from major ion concentrations (HCO 3 was calculated using the ion balance). Regarding sampling points with repeated observations, the differences among observations were small (second or third EC = first EC , R 2 = 0.48, n = 32). We measured water velocity with an electromagnetic velocity meter (KENEK, VE-20) and calculated the discharge by multiplying the cross-sectional area and water velocity. In addition, specific discharge was calculated by dividing the discharge by the catchment area. The size, soil, and geology of the catchment area were determined using a 10-m digital elevation model (DEM; Geographical Survey Institute), a soil map at a scale of 1 : 200,000 (Geographical Survey Institute), a geological map at a scale of 1 : 200,000 (Geological Survey of Japan, AIST), and ArcGIS (ESRI). We defined streams from the average catchment area of all springs (0.013 km 2, n = 15) and defined stream order using the method of Strahler (1952). We classified our observation points into four categories. Category 1 consisted of granodiorite observation points (granodiorite occupied more than 80% of the sub-catchment area of these observation points; n = 19). Category 2 was volcanic rock observation points (non-alkaline mafic volcanic rocks of the Early to Middle Miocene occupied more than 80% of the area; n = 41). These two groups of observation points had uniform geology. In category 3, a geology other than one of the two geologies mentioned 13

3 T. EGUSA ET AL. above occupied more than 80% of the area (felsic plutonic rocks, mafic plutonic rocks, and non-marine sedimentary rocks; n = 15). Category 4 included observation points with multiple geologies (no single geological setting occupied more than 80% of the area; n = 82). We defined the REA values of catchments with a single geology as the catchment area, above which all observed values were ±10% of the averaged value. For catchments with multiple geologies, we confirmed whether the spatial variability of discharge and chemistry can be explained by mixing based on geologic percentages. We targeted observation points in category 4, for which five geologies (granodiorite, non-alkaline mafic volcanic rocks of the Early to Middle Miocene, felsic plutonic rocks, mafic plutonic rocks, and non-marine sedimentary rocks) constituted more than 80% of the total area (n = 59). We used the observed values in the largest catchment areas as the convergent values of each geology. The largest catchment areas of granodiorite, non-alkaline mafic volcanic rocks of the Early to Middle Miocene, felsic plutonic rocks, mafic plutonic rocks, and non-marine sedimentary rocks were 8.3, 1.2, 0.9, 0.6, and 0.2 km 2, respectively. We multiplied the geologic percentage and convergent values and combined all values. RESULTS Figure 2 shows the relationship between catchment area and observed variables (EC, SiO 2, Ca 2+, Mg 2+, and specific discharge) for the case of uniform geology (granodiorite or volcanic rocks). Similar to the results of other studies, the concentrations and discharges showed large variability among small catchments (<0.1 km 2 ). This variability decreased with increasing catchment area. Using our definition, the REA values are 0.6 km 2 for EC, 1.5 km 2 for Mg 2+, and 1.5 km 2 for Ca 2+ in granodiorite, and 0.1 km 2 for EC, 0.8 km 2 for SiO 2, and 0.1 km 2 for Mg 2+ in volcanic rocks. We did not have enough data regarding the specific discharge of each of these geologies. SiO 2 in granodiorite and Ca 2+ in volcanic rocks had higher than average values for catchments above 1 km 2. However, those two geologies were also constant above km 2. In addition, the solute Figure 2. The relationship between catchment scale and observed variables for the case of uniform geology. Open circles indicate granodiorite (category 1) and solid circles indicate volcanic rocks (category 2). Figure 3. The relationship between catchment scale and observed variables for the case of heterogeneous geology (category 4). Vertical bars with solid and broken lines on the left of vertical axes show the range of observation values for volcanic rocks and for granodiorite. 14

4 SPATIAL VARIABILITY OF STREAM WATER Figure 4. Comparison between calculated and observed values. Open circles show catchments under 1 km 2, solid circles show catchments of 1 10 km 2, and open triangles show catchments >10 km 2. The solid line is the 1 : 1 line, and the broken line indicates a 10% variation. concentrations of the volcanic rocks were higher than those of the granodiorite. Figure 3 shows the relationships between observed variables (EC, SiO 2, Ca 2+, and Mg 2+ ) and catchment area for the case of multiple geologies. SiO 2 and Mg 2+ were similar to those for uniform geology, and the variability among the small catchments decreased as the catchment area increased. Variation was still found after a few square kilometers, but values became constant beyond 10 km 2. However, Ca 2+ concentrations and EC increased beyond 10 km 2, and clear convergence did not occur until 55 km 2. In addition, the concentrations of the 55 km 2 catchment (Ca 2+ = 290 µmol L 1 ; EC = 114 µs cm 1 ) were much higher than the average values of all of our observation points (Ca 2+ = 190 µmol L 1 ; EC = 77 µs cm 1 ). Figure 4 shows the relationships between calculated values based on geology percentages and observed values. In small catchments (< 1 km 2 ), the observed values tended to depart from the 1 : 1 line, except for EC. The observed values for 1 10 km 2 showed narrower ranges of variability than did smaller catchments. EC, Mg 2+, and SiO 2 almost fell within ±10%. For Ca 2+, only several points departed markedly from the 1 : 1 line, and many points tended to approach the line. For the observed values for areas more than 10 km 2, only SiO 2 almost fell within ±10%. For other observed variables, two points (11 and 17 km 2 ) were almost on the line, whereas another three points (19, 28, and 30 km 2 ) departed markedly from the line. In addition, for these three points, the observed values were higher than the calculated values. DISCUSSION A comparison of REA values among different geologies In the case of uniform geology, stream chemistry became constant beyond km 2 in granodiorite (except for SiO 2 ) and km 2 in volcanic rocks (except for Ca 2+ ). The REA values for granodiorite were slightly higher than the values for volcanic rocks (Figure 2). However, the difference between the two geologies was not as large as the differences among the observed variables. Considering the definition of REA, we can speculate that the size of the REA is closely related to the degree of variability among small catchments and the numbers of confluences. This means that smaller variability among small catchments and more confluences result in a smaller REA. In the two geologies, the coefficients of variance (CVs) of first-order streams were EC = 23 and 13, SiO 2 = 20 and 17, Mg 2+ = 21 and 16, and Ca 2+ = 24 and 12 for granodiorite and volcanic rocks, respectively. All values fell within The CVs of granodiorite were slightly higher than those of volcanic rocks. The average catchment areas of first-order streams were 2.0 km 2 in granodiorite (n = 11) and 1.6 km 2 in volcanic rocks (n = 26) and were not significantly different when t-tests were performed. From Horton s first law, the number of stream confluences increases with stream order at a constant rate (Horton, 1945). Generally, a smaller catchment area of zero or first-order streams leads to a higher drainage density and to more confluences because the drainage network is well developed. Our results indicated that the two geologies differ little in their CVs and have similar drainage densities. As a result, the REA values of the two geologies displayed only a slight difference. Asano et al. (2009) undertook a similar study in a catchment with a brown forest soil in a warm humid climate (similar to conditions in our study). The geology of their study site was weathered granite, and they confirmed convergence above km 2. They used a moving coefficient of variance based on 10 values and showed that the maximum moving CVs of SiO 2, Ca 2+, Mg 2+, K +, and SO 4 2 ranged from 28 to 45, and the CVs of Ca 2+ and Mg 2+ had a higher variance (above 100). The average catchment area of zero-order hollows was km 2 and was smaller than that in our observations (0.013 km 2 ). A higher CV should result in higher REA values. At their site, however, the drainage density was higher and the number of confluences was larger. Therefore, these two characteristics might cancel each other. As a result, their site, which had higher CVs and a higher drainage density than our study site, might generate almost similar REA values. We speculated that one of the major factors determining the CV values and drainage density in small catchments is sedimentary age. In our study site, the granodiorite and volcanic rocks formed in the Miocene (7 15 and15 22 Ma, respectively). The weathered granite that Asano et al. (2009) targeted formed in the Cretaceous ( Ma). Bedrock is affected by weathering and erosion over time. Generally, more weathering and erosion of bedrock lead to an expanding weathered layer and deeper infiltration of groundwater. In small catchments with uniform geology and soil, Asano et al. (2009) showed that the spatial variabilities in stream water discharge and chemistry did not have a significant relationship with the catchment area, slope gradient, or topographic index defined by Beven and Kirkby (1979). The spatial variability of stream water chemistry could be explained by the mixing of soil water and bedrock groundwater (Uchida and Asano, 2010). Therefore, the 15

5 T. EGUSA ET AL. increase in bedrock groundwater infiltration with increase in sedimentary age might result in larger spatial variability in small catchments because the gap in bedrock groundwater discharge widens. At the same time, the older sedimentary age might cause greater erosion and generate smaller zeroorder hollows and a higher drainage density (Jefferson et al., 2010). As a result, the weathered granite that Asano et al. (2009) targeted might have higher CVs and drainage density than our site. Furthermore, the two geologies at our site might have little difference in their CVs and drainage density. Confirmation of the REA concept in catchments with multiple geologies In regions with multiple geologies, the ratio of each geology changes as catchment area increases, and this change in the ratio might lead to increases or decreases in discharge and chemistry. Among the catchments with uniform geology, each observed variable showed large variability in small catchments, but became constant at an area greater than about 1 km 2, and the values of the solute concentrations were specific for each geology (Figure 2). Conversely, at observation points with multiple geologies, the spatial variabilities of discharge and chemistry did not converge at less than a few square kilometers, and at least 10 km 2 was needed for convergence (Figure 3). At 1 17 km 2, almost all of the observed variables were explained by mixing based on geological percentages (Figure 4). The result suggested that in regions with multiple geologies, the adoption of the REA concept with singleparameter geology was confirmed. The range of REA was 1 17 km 2 ; areas larger than 17 km 2 were thought to be outside the range. In addition, observed values were higher than calculated ones. At greater than 17 km 2, groundwater flow might have an influence. Shaman et al. (2004) showed that the bedrock groundwater discharge increased with catchment area in a Devonian age sedimentary rock catchment. In our study site, specific discharge was not observed above 10 km 2, and thus direct evidence of an increase in groundwater was not obtained. More observations are required to identify the cause of our results. In our calculation, we did not consider the difference in specific discharge among geologies. Observed results showed constant values of about 4 5 mm day 1 in 1 8 km 2 catchments for multiple geologies (Figure 3). Musiake et al. (1981) compared discharge-duration curves among geologies. Their results showed that the differences between granites and Tertiary volcanic rocks were smaller than the differences between Quaternary volcanic rocks and Mesozoic Paleozoic formations. If these geologies are targeted, we should consider differences in specific discharge. CONCLUSION At observation points with uniform geology, of either granodiorite or volcanic rocks, stream water chemistry became constant above about 1 km 2. At observation points with multiple geologies, spatial variability was still large beyond a few square kilometers. In regions with multiple geologies, the adoption of the REA concept with singleparameter geology was confirmed at scales of 1 17 km 2. Scales greater than 17 km 2 were thought to be outside the range. Therefore, in meso-scale catchments with multiple geologies, the spatial pattern of geologies plays a crucial role in determining the spatial variability of stream discharge and chemistry. As catchment area increases, we need to consider other processes, such as large-scale groundwater flow through deep bedrock. ACKNOWLEDGMENT This article was based on the research supported by grants ( ) from the Ministry of Education, Culture, Sports, Science and Technology, Japan, a grant for Global Environment Research (D-1102) from the Ministry of Environment, Japan, and a grant for environmental research (Kondo Jiro Grant) from the Asahi Glass Foundation. REFERENCES Amano K, Matsubara N, Tagiri M The basement of Mt. Fuji: The Tanzawa mountain collided and accreted Paleooceanic island arc. In Fuji Volcano, The Volcanological Society of Japan (ed). Yamanashi Institute of Environmental Sciences: Fujiyoshida; (in Japanese with English abstract). Asano Y, Uchida T, Mimasu Y, Ohte N Spatial patterns of stream solute concentrations in a steep mountainous catchment with a homogeneous landscape. Water Resources Research 45: 1 9. doi: /2008WR Beven KJ, Kirkby MJ A physically based variable contributing area model of basin hydrology. Hydrological Science Bulletin 24: doi: / Blöschl G, Sivapalan M Scale issues in hydrological modeling: A review. Hydrological Processes 9: doi: /hyp Blöschl G Scaling in hydrology. Hydrological Processes 15: doi: /hyp.432. Hattanji T, Onda Y Coupling of runoff processes and sediment transport in mountainous watersheds underlain by different sedimentary rocks. Hydrological Processes 18: doi: /hyp Horton RE Erosional development of streams and their drainage basins; Hydrophysical approach to quantitative morphology. Geological Society of America Bulletin 56: doi: / (1945)56[275:EDOSAT] 2.0.CO;2. Jefferson A, Grant GE, Lewis SL, Lancaster ST Coevolution of hydrology and topography on a basalt landscape in the Oregon Cascade Range, USA. Earth Surface Processes and Landforms 35: doi: /esp McDonnell JJ, Sivapalan M, Vaché K, Dunn S, Grant G, Haggerty R, Hinz C, Hooper R, Kirchner J, Roderick ML, Selker J, Weiler M Moving beyond heterogeneity and process complexity: A new vision for watershed hydrology. Water Resources Research 43: 1 6. doi: /2006WR Musiake K, Takahasi Y, Ando Y Effects of basin geology on river-flow regime in mountainous areas of Japan. Proceedings of the Japan Society of Civil Engineers 309: doi: /jscej _51. Shaman J, Stieglitz M, Burns D Are big basins just the sum of small catchments? Hydrological Processes 18: doi: /hyp Sivapalan M. 2003a. Prediction in ungauged basins: a grand 16

6 SPATIAL VARIABILITY OF STREAM WATER challenge for theoretical hydrology. Hydrological Processes 17: doi: /hyp Sivapalan M. 2003b. Process complexity at hillslope scale, process simplicity at the watershed scale: is there a connection? Hydrological Processes 17: doi: /hyp Strahler AN Hypsometric (area-altitude) analysis of erosional topology. Geological Society of America Bulletin 63: doi: / (1952)63[1117: HAAOET]2.0.CO;2. Temnerud J, Bishop K Spatial variation of streamwater chemistry in two Swedish boreal catchments: Implications for environmental assessment. Environmental Science & Technology 39: doi: /es040045q. Uchida T, Asano Y Spatial variability in the flowpath of hillslope runoff and streamflow in a meso-scale catchment. Hydrological Processes 24: doi: / hyp Wolock DM, Fan J, Lawrence GB Effects of basin size on low-flow stream chemistry and subsurface contact time in the Neversink river watershed, New York. Hydrological Processes 11: doi: /(SICI) (199707)11:9<1273::AID-HYP557>3.0.CO;2-S. Wood EF, Sivapalan M, Beven K, Band L Effects of spatial variability and scale with implications to hydrologic modeling. Journal of Hydrology 102: doi: / (88)90090-X. Woods R, Sivapalan M, Duncan M Investigating the representative elementary area concept: An approach based on field data. Hydrological Processes 9: doi: /hyp