Temperature Coeficiente
Temperature Coeficiente
Temperature Coeficiente
CONDUCTIVITY MEASUREMENTS
Stanford H. Smith
U. S. Bureau of Commercial Fisheries, Ann Arbor, Michigan
ABSTRACT
Electrical conductivity has been widely used in freshwater rcscarch but usual methods
employed by limnologists for converting measurements to conductance at a given tempera-
ture have not given uniformly accurate results. The temperature coefficient used to adjust
conductivity of natural waters to a given temperature varies depending on the kinds and
concentrations of electrolytes, the temperature at the time of measurement, and the tempera-
ture to which measurcmcnts are being adjusted. The tempcraturc coefficient was found to
differ for various lake and stream waters, and showed seasonal changes. High precision can
be obtained only by determining temperature coefficients for each water studied. Mean
temperature coefficients are given for various temperature ranges that may be used where
less precision is required.
moved from the lake, pressure is not a sig- Equation (2), of course, has no solution
nificant source of error, but it must be for t = 18. The value c at this temperature
taken into consideration if in situ measure- is estimated by the relationship:
ments are made below the surface (Har-
mon 1958 ) .
R17 - RIQ
(3)
Because the conductivity of water in- cl’ = 2 RI8
creases with temperature, the conductance
of different samples cannot be compared The temnerature coefficient of conduc-
tance is, hiwever, like conductance itself,
unless the measurements are made at the
not constant and the factors governing its
same temperature or adjusted to the con-
ductance at a single temperature. Mcasure- variation arc not simple. The coefficient
ments can be made at one temperature in differs with various electrolytes, and with
different concentrations and mixtures of
the laboratory, but more frequently they
are made at various temperatures in the electrolytes, and with temperature. The
field and adjusted to conductance at either temperature to which conductivity meas-
18” or 25°C. The temperature of 18°C is urements are adjusted also influences the
the most commonly used in freshwater value of the temperature coefficient, so the
search as this value is usually close to coefficients for correction of conductivity
temperature at which readings are taken, to 18°C cannot be used for correction to
and adjustments are subject to less error 25°C.
when the temperature diffcrcnces are The temperature coefficients of weak so-
small. lutions of salts and alkalies vary little with
The correction of conductivity mcasure- concentration, but change more widely
ments at various temperatures to conduc- with acids ( Tower 1905). The coefficient
tivities at a standard temperature is accom- also depends on the characteristics of the
plished by applying a temperature cocffi- various-solutes in the water. In most nat-
cient, which reflects the rate of change of ural waters, however, acids are absent or
conductance per degree of temperature weakly represented, and the number of
change. This coefficient has been stated other solutes as well as their concentrations
frequently as approximately 2% per degree do not vary greatly within a single b’ody.
Centigrade (c = 0.02) for dilute aqueous Under such conditions the influence of
solutions. This figure and, more recently, thcsc factors on the temperature coefficient
2.5% (c = 0.025) have been commonly used for a given water may be slight, particularly
by aquatic biologists to make temperature in the dilute concentration of unpolluted
adjustments with the formula: waters.
The influence of temperature of measure-
1,000,000 ments on the temperature coefficient can,
&8 = (1) however, have a strong effect on the accu-
R$ [l+c(t-18)]
racy of the adjustment of results to a stand-
Kj8 = conductance in micromhos at ard temperature, particularly where meas-
18°C. urements are obtained over a wide temper-
Rb= resistance in ohms at tempera- ature range and conductivities vary little.
ture t. These difficulties were encountered in re-
c= temperature coefficient. cent limnological surveys of the Great Lakes
t= temperature at which the mcas- initiated in the early 1950’s by the U. S.
urement was made. Bureau of Commercial Fisheries. Measure-
If K1s of formula ( 1) is expressed in mcnts made in the field over a wide range
terms of resistance, we obtain the following of temperatures, ‘Zn d even measurements
solution for the temperature coefficient c: in the laboratory over a narrower tempera-
ture range exhibited conspicuous discrep-
R 18- Rt ancies despite adjustments by commonly
c= (2)
Rt (t- 18) accepted methods. Thus, refinements of
19395590, 1962, 3, Downloaded from https://aslopubs.onlinelibrary.wiley.com/doi/10.4319/lo.1962.7.3.0330 by IFSC - Instituto Federal de Santa Catarina, Wiley Online Library on [05/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
332 STANFORD H. SMITH
TABLE 1. Conductivity, total alkalinity and tem- which was completely immersed. Tempcr-
perature coefficient at 18°C for selected natural
waters for which temperature coefficient studies atures were taken at each resistance read-
were made ing with a thermometer that could be read
to O.l”C.
Total A quadratic regression has been found
suitable to describe the relationship be-
Stream or lake
-- tween temperature and resis&ance for dilute
Fleming Creek1 - 442 240 0.0250 solutions at lower temperatures, but a
Huron River, fall1 413 0.0250 third-degree equation should be used for
spring 236 1G 0.0257
Whitmore Lake1 264 90
temperature ranges much in excess of 30°C
0.0251
Lake Michigan 221 113 0.0253 ( Bremner and Thompson 1937). A parab-
North Lake2 127 66 0.0258 ola
Wcber
-~- Lake3 46 19 0.0279
l Washtenaw County, Michigan.
Ri = a + bt -I- et2
2 Van Buren County, Michigan.
3 Chcboygan County, Michigan.
Rt = resistance at temperature t
t = temperature
CI,b, and c = constants
temperature correction were essential to
make conductivity measurements of Great fitted by least squares to measurements of
Lakes waters more accurate and mean- each water gave a fit that was highly sig-
ingful. nificant ( p far smaller than 0.0001). The
Experiments were accordingly carried equations were used, therefore, to calculate
out to obtain information on three major resistance values at intervals of 1°C; it was
points: the variations of the temperature held valid further to extrapolate to 0” and
coefficient with temperature in individual 35°C. All determinations of conductance
waters; extent of differences between and temperature coefficients have been
waters in these variations; and seasonal based on these computed values of R.
changes of temperature coefficient in the
RESULTS
same natural water.
The curves of Figure 1 show for each
METHODS
water and each temperature the tempera-
To determine the extent of variation in ture coefficient ( expressed as percentages)
the temperature coefficient in different required to adjust an observed conductiv-
waters, samples were analyzed from four ity value to conductivity at 18°C. The
lakes and one stream with conductivities shape and position of curves of tempera-
ranging from 46 to 442 pmhos/cm3; in addi- ture coefficient differed for each sample.
tion, samples were obtained from the Six of the samples fall into a cluster, but
Huron River at both the fall low-water (413 Weber Lake stands conspicuously apart.
pmhos/cm3) and the spring high-water Hooper ( 1954) reported that Weber Lake
(236 pmhos/cm”) stages (Table 1). The had unusually low alkalinity, and suspected
methyl-orange alkalinity of these waters that hydrogen sulfide was present in deeper
ranged from 19 to 240 ppm as CaC03. water where the sample for this study was
Measurements were made with an RC-1 collected.
bridge and dip-cell electrode manufactured The forms of the curves for the Huron
by Industrial Instruments, Inc., of Cedar River samples collected in the fall (low
Grove, New Jersey. Water samples were water ) and spring (high water) are sim-
placed in an ice-water bath and cooled to ilar, but the considerable differences in the
4°C. Then the ice was removed and resist- positions of the curves indicate that a shift
ance readings were made as the sample in temperature coefficient can result from
warmed to about 32°C. The sample was seasonal changes in chemical characteris-
constantly agitated with a mechanical stir- tics. These samples were taken from the
rer as well as with the dip-cell electrode river as it enters the city of Ann Arbor,
19395590, 1962, 3, Downloaded from https://aslopubs.onlinelibrary.wiley.com/doi/10.4319/lo.1962.7.3.0330 by IFSC - Instituto Federal de Santa Catarina, Wiley Online Library on [05/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
TEMPERATURE CORRECTION IN CONDUCTIV.CTY MEASUREMENTS 333
I I I I
2.9 I I
2.8
I WEBER LAKE
/
2 NORTH LAKE
3 HURON RIVER-SPRING
4 LAKE MICHIGAN
5 HURON RIVER- FALL
/ 6 WHITMORE LAKE
/ 7 FLEMING CREEK
/
I
2 .I
I I I I I I
2 .c L
‘0 I 5 IO 15 20 25 30 35
TEMPERATURE (‘Cl
FIG. 1. The relation between temperature and tcmpcraturc coefficient of conductance for selected
waters.
Michigan, above which there is some intro- determine the variation of the temperature
duction of industrial and domestic wastes. coefficient for the water under study for
Since the electrolytes represented would be the temperature range over which measure-
subject to little change, the differences ments are to be made. The data indicated
between temperature coefficients in the that the error of estimate of conductivity at
samples taken at low and high water in the lS?C rarely reaches 1% when temperature
Huron River probably reflect the influence coefficients, expressed to the nearest thou-
of differences in dilution. sandth, are determined for the water
An explanation for the differences in under study. If the coefficient is taken to
curves for temperature coefficients of vari- the nearest ten-thousandth the error seldom
ous waters might be found in a detailed reaches 0.1%.
study of differences in chemical content. Where it is not possible to determine the
Certainly more research is needed not only
variation of the temperature coefficient for
to gain a better understanding of the be-
the water, or where close precision (up to
havior of temperature coefficients, but also
to provide greater meaning to the use of 10% error) of results is not important, the
conductivity measurements in the descrip- use of the single coefficient 0.025 would
tion of aquatic conditions, be satisfactory for nearly all natural waters.
The error becomes smaller as the tem-
DISCUSSION
perature at the time of measurement ap-
It is obvious that to make precise con- proaches 18°C. If measurements are made
ductivity measurements it is necessary to over a wide range of temperatures, more
19395590, 1962, 3, Downloaded from https://aslopubs.onlinelibrary.wiley.com/doi/10.4319/lo.1962.7.3.0330 by IFSC - Instituto Federal de Santa Catarina, Wiley Online Library on [05/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
334 STANFORD I-1. SMITH
than one coefficient value should be used efficient for adjustment of conductivity
even though high precision is not required. measurements at 18°C. If adjustments are
Results of this study indicate that the fol- made to other temperatures, the appropri-
lowing values (from averages, excluding ate coefficients must be determined.
Weber Lake and Huron River in the fall)
may be suitable for adjusting conductivity REFERENCES
measurements of most natural waters to RREMNER, RAYMOND I?., AND THOMAS G. THOMP-
18°C. SON. 1937. The electrical conductances of
Temperntlue Temperature standard potassium chloride solutions through-
(“Cl cocfficicnt out the temperature range of 0 to 25”. J.
O-l 0.021 Am. Chem. Sot., 59: 2372-2374.
2-4 0.022 HARMON, B. V. 1958. The effect of pressure
5-8 0.023 on the electrical conductivity of sea water.
9-12 0.024 J. Mar. Res., 16: 83-89.
13-19 0.025 HOOPER, FRANK F. 1954. Limnological features
20-26 0.026 of Weber Lake, Cheboygan County, Michi-
27-31 0.025 gan. Papers Mich. Acad. Sci., 39: 229-240.
32-33 0.024 NATIONAL RESEARCH COUNCIL. 1929. Intcrna-
34-35 0.023 tional critical tables of numerical data-
physics, chemistry and technology. Vol. 3.
If, however, water is known to be polluted McGraw-Hill, New York. 471 pp.
or to have other unusual chcmical-charac- ROSENTHAL, ROBERT. 1950. Solution conductiv-
ity measurement instruments. Instruments,
teristics, reasonable precision can be as- 23 : 664-669.
sured only by determining the temperature TOWEIX, OLIN FREEXAN. 1905. The conductiv-
coefficients for that water over the tem- ity of liquids. Chemical Publishing Co.,
perature range within which conductivity Easton, Pennsylvania. 182 pp.
WASHBURN, EDWARD W. 1918. The equivalent
measurements are made. conductance of electrolytes in dilute solutions.
This study has been based entirely on I. The water correction. J. Am. Chem. Sot.,
the characteristics of the temperature co- 40 : 106-158.