One of the most basic measurements for understanding watershed processes and aquatic ecosystem dynamics is streamflow variation. In the McMurdo Dry Valleys, streamflow measurement is particularly important because of indications that lake levels have been rising rapidly. Lake-level rise has been attributed to increasing temperatures resulting in greater meltwater generation from the glaciers (Chinn 1993, pp. 1-51). Because of the high variability in streamflow on hourly time scales, intermittent measurements using flow meters (such as a Pygmy meter or an AA meter) provide an inadequate description of flow regimes and continuous measurements are required (Green et al, 1989, pp.129-148). As part of the basic data collection for the McMurdo Dry Valley Long-Term Ecological Research (LTER), we have established a stream gaging network for the three major lake basins in Taylor Valley. These data are critical for determining nutrient budgets for the lake ecosystems and for understanding physical factors controlling microbial mats in the streams.

Obtaining accurate continuous streamflow measurements for dry valley streams is a challenge. In addition to taking the highly variable flow into account, researchers find that the very low flow rates, which occur during cold spells in the austral summer, are difficult to measure accurately. An important consideration in designing a gaging station is minimizing contact of the streamwater with materials that might leach solutes and affect the chemistry of the streamwater. The streams in the dry valleys have very low concentrations of solutes because they are fed by glacial meltwater and because geochemical weathering processes are constrained to the streambed and adjacent areas. Finally, the unstable nature of the stream channels make measurements difficult. The streams flow through unconsolidated alluvium barren of vegetation and establishing a gaging site on exposed bedrock is not possible. Although this is a challenge, it minimizes the long-term alteration caused by activities in a stream channel at a local gaging station. Obtaining stream-stage data and defining the relation of stream stage and streamflow (rating curve) are very difficult in these conditions. The rating curve is developed from flow measurements at a range of low and high flow conditions.

Three installations (methods) were used to obtain accurate data. One method used was to select a site in the stream that had an adequate channel or section control for the development of a rating curve. Another method used was to install a weir (V-notched or broad-crested) and continuously measure water level (Chinn 1979). For low streamflows, the theoretical rating of the weir is used; otherwise, when the theoretical rating is exceeded, the rating must be defined using streamflow measurements. This method creates an impoundment even at low flows. Finally, a flume, such as a Parshall flume, was installed for measurements at high flow. At high flow, when water flows through the weir, a rating curve is used for calculation of discharge. When the flow is contained in the flume, discharge can be calculated based on the flume geometry. With this method, no impoundment occurs at low flow, and accurate measurements can be obtained because of the geometry of the flume.

In the dry valleys, all three methods described above have been used. The simplest method is the use of a channel or section control, but this method results in much less accurate data, especially at low flows. Also, the stream channels are unstable, and rating curves must be regularly updated each year. The use of weirs has been generally successful, but problems with washout at high flow and formation of an ice-cover in the impoundment at low flow have been encountered. The weirs have the disadvantage of altering the interaction with the lake, because the sediment carried in the stream is retained behind the weir at all flow regimes. This effect is not sustained, because once the wier is removed the retained sediment would be transported to the lake in a few years.

We have primarily used combined flume and weir installations and channel or section control stations (table). At streams that flow for a shorter period and have low streamflow, periodic streamflow measurements were made. We will use correlations between the different streams to infer continuous measurements for streams where periodic measurements are made. The flumes are made of black fiberglass, which is relatively inert. Because of continuous sunlight, the black color minimizes problems with ice formation in the flume. The advantages of the use of Parshall flumes are that high-quality data can be obtained at low flow and retention of waterborne sediment during low and moderate flow is minimal. For the flume/weir stations, we chose sites to minimize the length of the cutoff walls, and we built the cutoff walls using polyester cloth sandbags filled with alluvium from near the stream channel. Because the sandbags are filled with channel material, there is little possibility of directly altering stream chemistry. Stream-stage records were collected using techniques described by Rantz and others (1982a, b). Stream stage was collected at 15-minute intervals using a pressure sensor system connected to a data logger. The equipment is housed in a plywood crate that is secured at a location above the channel using ropes and buried weights.

The figure presents streamflow data for the 1993-94 austral summer for three streams in Taylor Valley, located in each of three lake basins. The data show that the large diel variation is a common feature of the flow regimes of dry valley streams. The major factors controlling the diel variation are probably sun angle and the aspect of the glaciers. The other feature illustrated in the figure is that the timing of initiation of flow between streams varies greatly, ranging from early November to late December. Again, this timing is a characteristic of the source glacier more than the lake basin itself. For example, Canada Stream and Green Creek, which are fed by the Canada Glacier and flow into Lake Fryxell, also began flowing in mid-November.

The results indicate that several years of streamflow record will be necessary for development of reliable correlations between streams and for predicting streamflow as a function of climate. By establishing the stream-gaging network at the onset of the McMurdo Dry Valley LTER, it will be possible to achieve some critical watershed modeling goals during the initial 6 years of the project. This work is supported by National Science Foundation grant OPP 92-11773.

REFERENCES.

Chinn, T.H. 1979. Hydrologic Research Report, Dry Valleys, Antarctica, 1972-73. Wellington: New Zealand Ministry of Works and Development.

Chinn, T.H. 1993. Physical hydrology of the dry valley lakes. In W.J. Green and E.I. Friedmann (Eds.), Physical and biogeochemical processes in Antarctic lakes (Antarctic Research Series, Vol. 59). Washington, DC.: American Geophysical Union.

Green, W.J., T.J. Gardner, T.G. Ferdelman, M.P. Angle, L.C. Varner, and P. Nixon. 1989. Geochemical processes in the Lake Fryxell Basin (Victorialand, Antarctica). In W.I. Vincent and J.C. Ellis-Evans (Eds.), Hydrobiologia 172. Belgium: Kluwer.

Rantz, S.E., and others. 1982a. Measurement and computation of streamflow: Volume 1, Measurement of stage and discharge (U.S. Geological Survey Water-Supply Paper 2175). Washington, DC.: U.S. Government Printing Office.

Rantz, S.E., and others. 1982b. Measurement and computation of streamflow: Volume 2, Computation of discharge (U.S. Geological Survey Water-Supply Paper 2175). Washington, D.C.: U.S. Government Printing Office.