Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

MAY 2004 NOTES AND CORRESPONDENCE 821

NOTES AND CORRESPONDENCE

Uncertainties of Derived Dewpoint Temperature and Relative Humidity


X. LIN AND K. G. HUBBARD
High Plains Regional Climate Center, School of Natural Resource Sciences, University of Nebraska—Lincoln, Lincoln, Nebraska

25 July 2003 and 8 December 2003

ABSTRACT
This paper presents an evaluation of derived dewpoint temperature and derived relative humidity, in which
the dewpoint temperature is calculated using measured ambient air temperature and measured relative humidity
variables and the derived relative humidity is calculated from measured dewpoint temperature. The derived
dewpoint temperature and relative humidity are calculated using algorithms provided by the World Meteorological
Organization. The method of uncertainty analysis, provided by the National Institute of Standards and Technology,
is applied to calculate the uncertainties of an indirect measurement of derived dewpoint temperature and derived
relative humidity. The results from the uncertainty analyses of derived and observed variables suggest that the
use of derived dewpoint temperature and derived relative humidity involves risk because the uncertainties of
modern dewpoint temperature and relative humidity sensors can create several degrees Celsius of error in the
derived dewpoint temperature and several percent in the derived relative humidity.

1. Introduction cept has been internationally accepted as the basis for


A single air humidity sensor or system usually is the determination of the measurement uncertainty. Ap-
capable of directly measuring a single humidity-related plication of these guidelines has extended beyond cal-
variable such as relative humidity by use of a solid-state ibration and research laboratories and into the industrial
humidity sensor or dewpoint temperature from a chilled domain of manufactured products. Our intent is to obtain
mirror hygrometer. Use of these technologies is common the uncertainties of derived dewpoint temperature and
with weather-station networks or controlled environ- derived relative humidity when derived using the cal-
mental systems. However, in many agricultural and en- culation algorithms provided by the World Meteorolog-
gineering applications, the dewpoint temperature could ical Organization (WMO 1996).
be the required variable when the relative humidity was
measured, or vice versa. In such cases, we usually per- 2. Method
form some calculations to derive the desired humidity More than a dozen simple or complicated empirical
variables from existing data and available algorithms. equations exist for calculating the saturation vapor pres-
The question for us is, How accurate are these derived sure from ambient temperature and ambient air pressure
humidity variables and how can we quantify their un- or calculating the actual vapor pressure from the dew-
certainties? point temperature based on the Goff–Gratch (1946) for-
The concept of the uncertainty of measurements was mulation. The equations that are commonly used in at-
newly defined in a set of guidelines by the National mospheric sciences and other related disciplines are
Institute of Standards and Technology (NIST 1994) that from Wexler (1976), Buck (1981), and WMO (1996).
is a technical summary of a longer International Or- In this study, the WMO equations were taken for the
ganization for Standardization guide (ISO 1993), which uncertainty analysis as follows (WMO 1996):
was updated and corrected in 1995. The uncertainty is

1243.12 1 T 2
a parameter, associated with the result of a measurement, 17.62Ta
that characterizes the dispersion of the values that rea- e ws 5 f (P)6.112 exp and (1)
a
sonably could be attributed to the measurand. This con-
ew
RH 5 100 , (2)
Corresponding author address: Dr. Kenneth G. Hubbard, 244 L. e ws
W. Chase Hall, School of Natural Resource Sciences, University of
Nebraska—Lincoln, Lincoln, NE 68583-0728. where e ws is saturation vapor pressure (hPa) with respect
E-mail: khubbard1@unl.edu to water at the air temperature T a (8C), f (P) [51.0016

q 2004 American Meteorological Society


822 JOURNAL OF APPLIED METEOROLOGY VOLUME 43

FIG. 1. Uncertainty of derived dewpoint temperature under (top)


the accuracy combination of 60.28C for a known ambient temperature
and 62% for a known relative humidity and (bottom) the accuracy
combination of 60.38C for a known ambient temperature and 65%
for a known relative humidity. FIG. 2. Uncertainty of derived relative humidity under (top) the
accuracy combination of 60.28C for a known ambient temperature
and 60.38C for a known dewpoint temperature and (bottom) the
accuracy combination of 60.38C for a known ambient temperature
1 (3.15 3 10 26 )P 2 0.0074P 21 ] is the enhancement and 60.58C for a known dewpoint temperature.
factor at the air pressure P (hPa), RH is relative humidity
(%), and e w is actual vapor pressure (hPa).
The saturation vapor pressure e ws is a function of air the derived RH from known values of the T d and T a .
temperature and site atmospheric pressure, whereas the The calculations for both RH and T d obviously require
actual vapor pressure e w is a function of dewpoint tem- the measurements of site atmospheric pressure. In our
perature T d and site atmospheric pressure if the satu- uncertainty analysis, the site or station atmospheric pres-
ration vapor pressure is with respect to the water or is sure is considered to be a constant (1000 hPa).
a function of frostpoint temperature if the saturation In uncertainty analysis, the individual standard un-
vapor pressure is with respect to ice. Thus, the WMO certainty u i is defined as the uncertainty of the result of
Eqs. (1) and (2) are often used to calculate the derived a measurement expressed as its standard deviation
T d from known values of RH and T a and to calculate (NIST 1994). For the manufacturer’s stated ‘‘accuracy’’
MAY 2004 NOTES AND CORRESPONDENCE 823

specifications of air temperature, relative humidity, and TABLE 1. Uncertainty of derived dewpoint temperature.
dewpoint temperature sensors, it is reasonable to use Inputs Uncertainty of derived T d (8C)
rectangular distributions in uncertainty analysis (NIST
T a (8C) RH (%) (60.28C, 62%)* (60.38C, 65%)*
1994). Thus, one must assume a rectangular distribution
for all sensors that measure T a , RH, and T d based on 240 10 1.8 4.5
220 10 2.1 5.4
the manufacturer specifications with a lower limit a2 0 10 2.5 6.2
and an upper limit a1 (e.g., 60.28C for T a or T d and 20 10 2.9 7.2
62% for RH). Therefore, the standard uncertainty 40 10 3.3 8.1
(NIST 1994) of T d and RH is 240 50 0.5 1.1
220 50 0.5 1.3
a 0 50 0.6 1.5
u(x i ) 5 , (3) 20 50 0.7 1.7
Ï3 40 50 0.8 2.0
where a 5 (a1 2 a2 )/2 for both dewpoint temperature 240 98 0.3 0.7
220 98 0.4 0.8
and relative humidity, and x i represents T a , T d , and RH. 0 98 0.4 0.9
The combined standard uncertainty u c can be derived 20 98 0.4 1.0
from the original Eqs. (1) and (2) using the law of prop- 40 98 0.5 1.2
agation of uncertainty, commonly called the root sum-
of-squares (RSS) method for a specific output quantity * The first term is for T a’s stated accuracy, and the second term is
for RH’s stated accuracy.
(T d or RH). Multiplying a coverage factor k (NIST 1994)
by the combined standard uncertainty, one can obtain
the expanded uncertainty U. In accord, a normal cov-
3. Results and discussion
erage distribution will be used for expressing uncer-
tainty in this study and is defined by setting k 5 2 such a. Uncertainty of derived dewpoint temperature
that the uncertainty interval for U has a 95% confidence
level. We assume that, in each system, the measurement Figure 1 illustrates the variation of the uncertainty of
of dewpoint temperature or the measurement of relative derived dewpoint temperature with ambient temperature
humidity is statistically independent of air temperature and relative humidity. The uncertainty of derived dew-
measured in observations. Therefore, the expanded un- point temperature (UTd) increases as the known relative
certainties for derived dewpoint temperature UTd and humidity decreases. With decreasing ambient air tem-
derived relative humidity U RH in each system can be perature, the uncertainty of derived dewpoint temper-
calculated as ature decreases. When the known relative humidity is
10%, the uncertainty of derived dewpoint temperature

[1 ]
1/2

2 1 2
]Td 2 ]Td
2 2
ranges from 1.88 to 3.38C, resulting in a system with an
U Td 5 k u (Ta ) 1 u 2 (RH) and (4a) air temperature accuracy of 60.28C and a relative hu-
]Ta ]RH
midity accuracy of 62% (Table 1). From Fig. 1 and

[1 ] Table 1, in principle, for any dewpoint temperature in-


1/2

2 1 2
]RH 2 ]RH 2
2 2

URH 5k u (Ta ) 1 u (Td ) . (4b) strument or sensor based upon the relative humidity
]Ta ]Td measurement, the uncertainty of dewpoint temperature
The partial derivatives in Eq. (4) are often referred to measurement cannot fall below 60.58C unless the man-
as sensitivity coefficients and can be obtained analyti- ufacturer uses a more accurate saturation vapor pressure
cally. From Eq. (3), u(T a ), u(T d ), and u(RH) are the equation than that provided by the WMO in Eq. (1).
standard uncertainty associated with the manufacturer’s Furthermore, if the known ambient temperature and rel-
specifications for each sensor. The value of k is 2 as ative humidity have an accuracy of 60.38C and 65%,
described above. To illustrate the uncertainty of derived respectively, this combination creates a range of un-
dewpoint temperature and derived relative humidity, we certainty for derived dewpoint temperature of 0.78–
assume that the accuracy specifications from the man- 8.18C (Table 1). In reality, the accuracy combination of
ufacturers are 60.28 or 60.38C for air temperature, 60.38C and 65% for the ambient temperature and rel-
62% or 65% for relative humidity, and 60.38 or ative humidity is a realistic expectation when quality
60.58C for dewpoint temperature. For the derived dew- sensors are deployed in remote field conditions.
point temperature calculations, we took the known am-
bient temperature for a range extending from 2508 to b. Uncertainty of derived relative humidity
1508C and the known ambient relative humidity ex-
tending across a range from 2% to 100%. For derived The uncertainty of derived relative humidity is illus-
relative humidity calculations, the ambient temperature trated in Fig. 2. In general, the uncertainty of derived
range was set between 2508 and 1508C and the dew- relative humidity exponentially increases with a de-
point temperature depression (difference between am- creasing dewpoint temperature depression or an increas-
bient air temperature and dewpoint temperature) was ing relative humidity. When the dewpoint temperature
limited to a range of 08–508C. depression is relatively small, increasing the ambient
824 JOURNAL OF APPLIED METEOROLOGY VOLUME 43

TABLE 2. Uncertainty of derived relative humidity.

Inputs Uncertainty of derived RH (%)


T a (8C) T a 2 T d (8C) (60.28C, 60.38C)* (60.38C, 60.58C)*
240 2 6.6 10.9
220 2 5.7 9.5
0 2 2.9 4.7
20 2 1.4 2.1
40 2 2.0 3.3
240 10 3.3 5.6
220 10 3.6 6.0
0 10 2.3 3.8
20 10 0.9 1.3
40 10 1.0 1.6
240 20 1.3 2.1
220 20 1.8 3.0
0 20 1.5 2.5
20 20 0.7 1.1
40 20 0.4 0.6

* The first term is for T a’s stated accuracy, and the second term is for T d’s stated accuracy.

temperature decreases the uncertainty of derived relative observations from remotely deployed sensors (bottom
humidity (Fig. 2 and Table 2). Under the accuracy com- panel of Figs. 1 and 2).
bination of 60.28 and 60.38C for ambient air temper-
ature and dewpoint temperature depression, the uncer-
4. Conclusions
tainty of derived relative humidity ranges from 0.4% to
6.6% (Table 2). The uncertainty of derived relative hu- Significant uncertainties exist when dewpoint tem-
midity increases when an accuracy combination of perature and relative humidity are derived using data
60.38 and 60.58C is used for ambient air temperature from automated weather stations, climate databases that
and dewpoint temperature depression (Fig. 2). In a sim- include derived humidity variables (T d or RH), and hu-
ilar way, the latter combination of measurements is a midity sensors designed to measure one variable and
more reasonable expectation for users using modern programmed to output another. The magnitude of the
sensors and practices under field conditions. Thus, the uncertainty for derived dewpoint temperature not only
uncertainty of derived relative humidity should range increases with increasing ambient temperature, but also
from 0.6% to 10.9% (Table 2 and Fig. 2). increases with the falling ambient relative humidity. The
The results in Figs. 1 and 2 reflect the imperfections magnitude of uncertainty for derived relative humidity
in the WMO equation under certain conditions when increases with decreasing dewpoint temperature de-
humidity variables are derived. However, even when pression. It also changes with ambient temperature, es-
other equations such as the Wexler (1976) equation or pecially when ambient relative humidity is high, which
American Society of Heating, Refrigeration, and Air in reality is a common condition. We recommend that
Conditioning Engineers (ASHRAE 1997) equation were one should use, when possible, the actual direct mea-
used to calculate saturation vapor pressure, similar re- surements for air humidity–related observation. How-
sults were produced. In addition, Gates (1994) presented ever, if derived air humidity data must be used, the
the expected error in derived dewpoint temperature us- uncertainty should be calculated, especially for derived
ing a measured dry-bulb temperature and relative hu- dewpoint temperatures when the relative humidity is
midity and the ASHRAE equation. Note that the ex- low and for derived relative humidity when the relative
pected error or root-mean-square error used in Gates humidity is high. The hope for a reduction in uncertainty
(1994) indicates one standard deviation in the units of of derived humidity variables lies in examining the bias
measure (a 68% confidence level for normal distribu- relationship between measured humidity variables and
tion). However, the method used by Gates to simulate other variables (e.g., ambient temperature of sensors and
the results was based on the RSS rather than the ex- ambient wind speed). Transfer functions that remove
pected error or the root-mean-square error (Gates 1994). this bias will reduce the input uncertainty in the uncer-
The results presented in our study reflect the uncertain- tainty analysis and will result in lower uncertainty of
ties of derived dewpoint temperature and derived rel- the derived variables. We intend to study this aspect
ative humidity at the 95% confidence level. To the au- more in the future.
thors’ knowledge, the accuracy combinations shown in
the third column of Tables 1 and 2 (top panel of Figs. Acknowledgments. This study was supported in part
1 and 2) are achievable in the laboratory, but the results by the National Climatic Data Center (NCDC) in our
illustrated in the fourth column are more realistic of the project of ‘‘Performance Study of Air Humidity/Water
MAY 2004 NOTES AND CORRESPONDENCE 825

Vapor Monitoring Systems for the U.S. Climate Ref- Gates, R. G., 1994: Dew point temperature error from measuring dry
erence Network (USCRN).’’ We thank Nebraska State bulb temperature and relative humidity. Trans. ASAE, 37, 687–
688.
Climatologist Allen L. Dutcher and Dr. Qi Hu for their Goff, J. A., and S. Gratch, 1946: Low-pressure properties of water
valuable comments and suggestions on the original man- from 2160 to 12128F. Trans. Amer. Soc. Heat. Vent. Eng., 52,
uscript. This paper has been approved as Journal Series 95–121.
No. 14227 by the Agricultural Research Division, Uni- ISO, 1993: Guide to the Expression of Uncertainty in Measurement.
versity of Nebraska at Lincoln. International Organization for Standardization, 101 pp.
NIST, 1994: Guidelines for evaluating and expressing the uncertainty
of NIST measurements results. National Institute of Standards
REFERENCES and Technology Tech. Note 1297, 20 pp.
ASHRAE, 1997: ASHRAE Handbook—Fundamentals. American So- Wexler, A., 1976: Vapor pressure formulation for water in range 0
ciety of Heating, Refrigerating and Air-Conditioning Engineers, to 1008C—A revision. J. Res. Natl. Bur. Stand., 80A, 775–785.
576 pp. WMO, 1996: Guide to Meteorological Instruments and Methods of
Buck, A. L., 1981: New equations for computing vapor pressure and Observation. 6th ed. WMO-No. 8, World Meteorological Or-
enhancement factor. J. Appl. Meteor., 20, 1527–1532. ganization, 501 pp.

You might also like