Environ. Sci. Technol. 2003, 37, 859-864
Scale-Dependent Temporal Variations in Stream Water Geochemistry SONIA A. NAGORSKI,† J O H N N I E N . M O O R E , * ,† T E M P L E E . M C K I N N O N , †,§ A N D DAVID B. SMITH‡ Murdock Environmental Biogeochemistry Laboratory, Department of Geology, The University of Montana, Missoula, Montana 59812, and U.S. Geological Survey, MS 973, Denver, Colorado 80225
A year-long study of four western Montana streams (two impacted by mining and two “pristine”) evaluated surface water geochemical dynamics on various time scales (monthly, daily, and bi-hourly). Monthly changes were dominated by snowmelt and precipitation dynamics. On the daily scale, post-rain surges in some solute and particulate concentrations were similar to those of early spring runoff flushing characteristics on the monthly scale. On the bi-hourly scale, we observed diel (diurnal-nocturnal) cycling for pH, dissolved oxygen, water temperature, dissolved inorganic carbon, total suspended sediment, and some total recoverable metals at some or all sites. A comparison of the cumulative geochemical variability within each of the temporal groups reveals that for many water quality parameters there were large overlaps of concentration ranges among groups. We found that short-term (daily and bi-hourly) variations of some geochemical parameters covered large proportions of the variations found on a much longer term (monthly) time scale. These results show the importance of nesting short-term studies within long-term geochemical study designs to separate signals of environmental change from natural variability.
I. Introduction The growing demand for natural resources and the conversion of lands to cities, farms, and mines will draw increasing regulatory and scientific focus on the pollution of natural waters. Though water resources are becoming increasingly stressed in terms of both quantity and quality, scientists still have a poor understanding of the temporal variability of the trace element geochemistry of both polluted and pristine streams. The adaptation of “clean” or “ultra-clean” sampling and analytical methods in the past decade have greatly increased sensitivity of trace-metal studies of surface waters (1-5). However, there are few guidelines on the sampling frequency necessary to accurately measure stream geochemistry with these highly sensitive methods. Most major monitoring programs can rarely afford to sample for trace elements more often than quarterly; however, it is unclear whether * Corresponding author phone: (406) 243-6807; fax (406) 2434028; e-mail:
[email protected]. † The University of Montana. ‡ U.S. Geological Survey. § Current address: Texas Water Development Board, P.O. Box 13231, Austin, TX 78711-3231. 10.1021/es025983+ CCC: $25.00 Published on Web 01/29/2003
2003 American Chemical Society
shorter-term variations can significantly complicate or obscure longer-term trends needed to assess environmental degradation of streams and rivers. There are many unanswered questions regarding how seasonal geochemical dynamics compare with those that occur on shorter time scales, such as on daily and hourly scales. Studies of metal loading into streams have shown that both particulate and dissolved metals can be rapidly mobilized into stream channels during storm events in mining-contaminated basins, and that the bulk of a stream’s annual metal load can be transported during a small number of short-lived high-flow events (6-8). A wide range of processes have been linked to diel cycles in streams and lakes (9-14). Photosynthesis and respiration change pH and dissolved gases that control dissolved arsenic concentration and speciation through adsorption/desorption processes (11). Photoreduction of iron oxides can have similar effects on dissolved iron (9). Daily changes in evapotranspiration by bank vegetation effect streamflow and the relative contributions from shallow groundwater, which can also control trace element content (12). Diel fluctuations of snowmelt hydrology also can effect stream chemistry by dilution processes. However, only a few researchers have placed diel variations into the context of seasonal variations. Notably, McKnight and Bencala (15) reported that during a 48 h period, changes in Fe concentrations in the Snake River, CO, reflected as much as 47% of the total variability seen over six years at the site. Similarly, Constanz (16) found that the diel surface water temperature in two Sierra Nevada streams captured 3040% of the annual variation. In a study of pH variability in three Colorado lakes, Turk (10) showed that diel pH variations encompassed 25-40% of the seasonal variation, and that these diel variations were larger than reported inter-annual differences at the lakes. The purpose of this study is to compare the amount of variability in the dissolved ( daily bi-hourly > monthly > daily
TSS, Al(tot), Ba, TSS, Al(tot), Ba, Ca, Fe, (b) TSS at CFBM Ca(tot), Fe, K, Li, Li, Mg, Si, Sr, Zn(filt) Mg, Mn(filt), S, Si, Sr pH, DO, WT, DIC, pH, WT, DIC, K, Mn (c) pH at BFB As, Mn(tot) (d) S(tot) at CFBM DO, As DO at BH pH at CFBM
Al(tot), Ba, K, Li, Mg, P, Si, Sr, Ti monthly > daily ) bi-hourly TSS, DIC, Al(tot), WT, TSS, DIC, As, Ba, Ca, Fe, As, Ca, Cu(tot), Fe, K, Li, Mg, S, Mn, Na Si, Sr, P(tot) monthly > bi-hourly > daily pH, DO, WT, Mn S pH, DO
S
(a) Ca(filt) at LF
No subscript ) true for total recoverable and filterable fractions; subscript “tot”) total recoverable fraction only; subscript “filt”) filterable fraction only. DO ) dissolved oxygen; DIC ) dissolved inorganic carbon; TSS ) total suspended sediment; WT ) water temperature.) Elements not listed were below detection limits. a
862
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 5, 2003
FIGURE 5. Boxplots comparing data distributions from each temporal sampling scale (bi-hourly, monthly, and daily): (a) filterable Ca at LF; (b) TSS at CFBM; (c) pH at BFB; (d) total recoverable S at CFBM; (e) DO at BH; and (f) pH at CFBM. Boxes depict the interquartile range (IQR) of data; the line within each box is the median; the whiskers span the full range of the data. Open circles depict mild outliers (1.5 × IQR), asterisks depict extreme outliers (3 × IQR). variation was at site BFB, where the daily and bi-hourly variability of filterable S encompassed 50% of the total monthly variability. For pH and DO at CFBM, the ranges of the bi-hourly values actually exceeded those on the monthly scale (Table 1, Figure 5f). Parameters for which the bi-hourly range covered 50-99% of the monthly were the following: filterable Mn at BFB; pH, water temperature, DO, total recoverable As, and filterable K at BH (Figure 5e); and pH, water temperature, and DO at LF. For all other elements the ranges of bi-hourly concentrations were less than 50% of the monthly ranges. Yet several parameters were in the higher end of this category. For example, the bi-hourly ranges of pH, water temperature, and DO at BFB were 32-46% of the monthly ranges (Figure 5c). At CFBM, the bi-hourly TSS range spanned 46% of the monthly range (Figure 5b); filterable and total recoverable As covered 33-36%; total recoverable Cu spanned 21%; and filterable and total recoverable Fe and Mn spanned 22-28% of the monthly concentration ranges.
Finally, comparison of bi-hourly to daily variability shows that bi-hourly variability was generally similar to or smaller than the daily variability (Table 1). However, for parameters with strong diel cycling, such as pH, DO, and water temperature, bi-hourly variability exceeded daily variability (Table 1). Additionally, the bi-hourly ranges of TSS values at CFBM, BH, and BFB were at least as large as the daily ranges. That is, there was as much or more variability in TSS values during two calm 24 h periods than measured daily during the 2 week long period of the September rains. We did not study enough mining-impacted and pristine sites to draw conclusions about differences in the temporal variation patterns between these two types of sites. Yet, it is worth noting that it was the large, mining-impacted site (CFBM) that showed diel cycling of total recoverable metals and that had the greatest similarities between short term and long-term concentration variations. Implications for Monitoring Water Quality. From these results it is clear that sampling at consistent times of the day VOL. 37, NO. 5, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
863
and over variable climatic and hydrologic conditions is essential to accurately evaluate temporal trends in trace metal geochemistry. Failure to do so may produce skewed and unrepresentative samples that may inaccurately portray the geochemical dynamics in streams. Although there were some distinct patterns within each time scale, concentration ranges seen on the three different scales of temporal resolution can be strikingly similar. Whereas much is left to be understood about the mechanisms involved in regulating the physical and chemical processes involved at various time scales, it is clear that the overall variability at short-term time scales needs to be acknowledged and characterized in water quality study sampling designs. Water quality samples should be taken at consistent times of day at each study site over the long-term, and antecedent climatic conditions should be noted, in order to make valid conclusions about any observed long-term temporal changes. Likewise, short-lived drastic events, such as surges in snowmelt and summer rainstorms, should also be accounted for in studies purporting to address long-term trends in water quality. Additionally, daily and bi-hourly samples may have greater ranges during different times of the year, and this variability should be defined as well. To the detriment of accurate water quality research, however, most current sampling designs do not adequately address the complicating factor of short-term variations. We propose that serious effort be made by water researchers to include short-term, yet intensive, studies into their longterm study designs. We also advocate the development of sophisticated means of automating ultra-clean water quality sampling, as the scale needed to adequately sample streams is hardly possible with human samplers on a large scale, especially in areas with difficult access and climatic conditions that hamper frequent sampling. The research presented here illustrates that much more work is needed to characterize natural geochemical variability, for only after we achieve this can we can improve on the reliability of water quality models that include temporal dimensions.
Acknowledgments We thank the USGS for funding this research. We also thank J. Harris, B. Nichols, N. Stevens, C. Wheeler, and A. Yakos for assistance with field and laboratory work, and to the anonymous reviewers who substantially improved the manuscript.
(4) Benoit, G. Geochim. Cosmochim. Acta 1995, 59, 2677-2687. (5) Taylor, H. E.; Shiller, A. M. Environ. Sci. Technol. 1995, 29, 13131317. (6) Bradley, S. B.; Lewin, J. Environ. Pollut. 1982, 4, 257-267. (7) Leenaers, H. Hydrol. Process. 1989, 3, 325-338. (8) Sanden, P. S.; Karlsson, A.; Duker, A.; Ledin, A.; Lundman, L. J. Geochem. Explor. 1997, 58, 145-155. (9) McKnight, D. M.; Kimball, B. A.; Bencala, K. E. Science 1988, 240, 637-640. (10) Turk, J. T. Water, Air, Soil Pollut. 1988, 37, 171-176. (11) Fuller, C. C.; Davis, J. A. Nature 1989, 340, 52-54. (12) Brick, C. M.; Moore, J. N. Environ. Sci. Technol. 1996, 30, 19531960. (13) Nimick, D. A. Eos, Abstr. Progr. 2001, 82, 47. (14) Sullivan, A. B.; Drever, J. I.; McKnight, D. M. J. Geochem. Explor. 1998, 64, 141-145. (15) McKnight, D. M.; Bencala, K. E. Arctic Alp. Res. 1988, 20, 492500. (16) Constanz, J. Water Resour. Res. 1998, 34, 1609-1615. (17) Nagorski, S. A.; Moore, J. N.; Smith, D. B. Mine Water Environ. 2002, 21, 121-136. (18) Moore, J. N.; Luoma, S. N.; Peters, D. Can. J. Fish. Aquat. Sci. 1991, 48, 222-232. (19) Moore, J. N.; Luoma, S. N. Environ. Sci. Technol. 1990, 24, 12791284. (20) Axtmann, E. V.; Luoma, S. N. Appl. Geochem. 1991, 6, 75-88. (21) McKinnon, T. E. MS Thesis, University of Montana, 2001. (22) Western Regional Climate Center. http://www.wrcc.dri.edu/ wrccmssn.html. accessed 2001. (23) Rantz, S. E. (compiler). U. S. G. S. Water Supply Paper 2175; U.S. Geolgical Survey: Washington, DC, 1982. (24) Horowitz, A. J.; Lum, K. R.; Garbarino, J. R.; Hall, G. E. M.; Lemieux, C.; Demas, C. R. Environ. Sci. Technol. 1996, 30, 954963. (25) U.S. Environmental Protection Agency. EPA-600/4-91-010; U.S. Government Printing Office: Washington, DC, 1991. (26) Martin, T. D.; Brockhoff, C. A.; Creed, J. T. EPA Method 200.15; U.S. Environmental Protection Agency: Washington, DC, 1994. (27) Nagorski, S. A.; Moore, J. N.; McKinnon, T. E.; Smith, D. B. Water Resour. Res., in press. (28) Stottlemyer, R.; Toczydlowski, D. Can. J. Fish. Aquat. Sci. 1990, 290-300. (29) Williams, M. W.; Brown, A. D.; Melack, J. M. Limnol. Oceanogr. 1993, 38, 775-797. (30) Bales, R. C.; Davis, R. E.; Williams, M. W. Hydrol. Process. 1993, 7, 389-401. (31) Campbell, D. H.; Clow, D. W.; Ingersoll, G. P.; Mast, M. A.; Spahr, N. E.; Turk, J. T. Water Resour. Res. 1995, 31, 2811-2821. (32) Moog, D. B.; Whiting, P. J. Water Resour. Res. 1998, 34, 23932399. (33) Stottlemyer, R.; Troendle, C. A. J. Hydrol. 1992, 140, 179208. (34) Dunn, I. G. Limnol. Oceanogr. 1967, 12, 151-154.
Literature Cited (1) Windom, H. L.; Byrd, J. T.; Smith, R. G.; Huan F. Environ. Sci. Technol. 1991, 25, 1137-1142. (2) Benoit, G. Environ. Sci. Technol. 1994, 28, 1987-1991. (3) Horowitz, A. J.; Dems, C. R.; Fitzgerald, K. K.; Miller, T. L.; Rickert, D. A. U. S. G. S. Open-File Report 94-539; U.S. Geological Survey: Washington, DC, 1994.
864
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 5, 2003
Received for review July 19, 2002. Revised manuscript received December 29, 2002. Accepted January 8, 2003. ES025983+
