Abstract

Elementary hydrostatic analysis shows that water-level measurements made with mercury manometers are subject to errors of approximately 0.02 percent per degree Celsius due to uncorrected variations in the temperature of the mercury. Normal diurnal variations in air temperature if uncorrected are sufficient to cause errors of ±0.15 percent or more. Errors of this magnitude would exceed the U.S. Geological Survey's pre-1989 tolerance of ±0.05 percent of full scale for stage-measurement instrumentation if the water levels measured were above one-third of full scale. Under the current (1991) more stringent tolerance of 0.02 percent of full scale, errors of this magnitude would exceed the tolerance at almost any water level. An informal survey conducted in 1986 indicated that most Geological Survey manometers do in fact operate at less than one-third of full scale for all but a few days per year, and that temperature sensitivity is relatively unimportant in comparison to other sources of error in measurements of high-water levels at stream-gaging stations.