Potter T.D., Colman B.R. (co-chief editors). The handbook of weather, climate, and water: dynamics, climate physical meteorology, weather systems, and measurements

Подождите немного. Документ загружается.

Maintenance includes preventive and corrective actions. A solid preventive main-

tenance program can preclude a large portion of equipment failure events. Corrective

maintenance should be applied as promptly as possible to minimize downtime,

provided that proper documentation is made of the failure symptoms or condition

and that the replacement or repaired equipment is properly tested when installed.

Data Processing

Data processing begins during the measurement phase of the program; sensor

responses are translated to a numerical value that can be averaged, totaled, stored

as an extreme, or used to calculate a variance about the mean. Modern data recording

equipment offers the user numerous options, as well as pitfalls. As with other

activities, proper testing and documentation of the on-site processing routines is

an essential ﬁrst step in demonstrating that data processing is correctly applied.

Most programs are enhanced by including data-identiﬁer information with all data

records, so that raw ﬁles can be uniquely identiﬁed by time period and station

location. Once the data are collected from the ﬁeld station, the data should be

traceable throughout the data validation and editing processes.

Audits

An important step in demonstrating data credibility to outside groups is to document

the results of independent veriﬁcations of compliance with established procedures

and tolerance limits. The term performance audit implies tests made by knowledge-

able staff independent of those performing the routine site operations and checks

using independent testing equipment, which is also traceable to standards. The

complementary function is a system audit, in which independent staff examine the

work products and docum entation to ensure that activities comply with the estab-

lished procedures.

Reports

The ﬁnal link in the chain of demonstrating data quality is to produce credible

reports of the monitoring program results. Report content and format should ﬁt

the objectives of the program. It is wise to include summaries of the quality control

and quality assurance activities in sufﬁcient detail that the results can be accepted.

REFERENCES

American Nuclear Society (1984). ANSI=ANS-2.5: Standard for Determining Meteorological

Information at Nuclear Power Sites, American Nuclear Society.

American Nuclear Society (2000). ANSI=ANS-3.11: Determining Meteorological Information

at Nuclear Facilities, American Nuclear Society.

Slade, D. H. (Ed.) (1968). Meteorology and Atomic Energy, U.S. AEC, July.

866

REGULATORY APPROACHES TO QUALITY ASSURANCE AND QUALITY CONTROL

U.S. EPA (1994). Quality Assurance Handbook for Air Pollution Measurement Systems,

Vol. IV: Meteorological Measurements, U.S. EPA Ofﬁce of Research and Development,

Washington, DC, EPA=600=R-95-038d, April.

U.S. EPA (2000). Meteorological Monitoring Guidance for Regulatory Modeling Applications,

U.S. EPA Ofﬁce of Air Quality Planning and Standards, Research Triangle Park, NC,

EPA-454=R-99-005, February.

REFERENCES 867

CHAPTER 46

MEASURING GLOBAL TEMPERATURE

JOHN R. CHRISTY

Every part of the Earth system may be described by the variable of state we call

temperature. Fundamentally, temperature is the magnitude of the average molecular

kinetic energy of a substance—the higher the molecule’s average speed, the greater

the temperature. Today, temperature is estimated by several kinds of instruments

using a variety of techniques.

In situ measurements require, in some way, direct contact between the instrument

and the moving molecules of the targeted substance. The most familiar of these types

of instruments is the liquid-in-glass ther mometer whose bulb contains a liquid that

expands or contracts, allowing the liquid to move through an opening to a narrow,

graduated tube. The bulb is inserted into the medium of interest (e.g., air, water, ice,

or earth) and the liquid responds according to the amount of molecular motion

detected on contact. Another in situ device utilizes the fact that the electrical resis-

tance of a conductor is proportional to its molecular kinetic energy so that the

amount of electricity able to ﬂow throu gh the conductor indicates temperature.

The velocity of sound through a substance (usually air or water) is also directly

related to its temperature and thus is an indirect characteristic that can be used to

monitor temperature in situ.

Remote met hods, in contrast to in situ, have the advantage of measuring tempera-

ture from a distance. The most direct of these methods employs radiometers that

measure the intensity of radiation emitted by a substance. The magnitude of the

intensity is often proportional to the substance’s temperature. Satellite radiometers

fall into this category of devices as they monitor the upwelling radiation from the

various components of Earth’s system. Other relatively new remotely sensed meth-

ods estimate temperature by measuring (a) the speed and refraction of radio signals

through a material (e.g., Global Positioning System satellites), (b) the physical

Handbook of Weather, Climate, and Water: Dynamics, Climate, Physical Meteorology, Weather Systems,

and Measurements, Edited by Thomas D. Potter and Bradley R. Colman.

ISBN 0-471-21490-6 # 2003 John Wiley & Sons, Inc.

869

height of the sea surface (higher altitudes mean expanded or warmer water) and even

the thermal ‘‘color’’ of the ocean affected by microorganisms that preferentially

appear in waters of certain temperatures.

The distinction between in situ and remo te may be rather blurred. One thinks of a

bucket dropped over the side of a ship, ﬁlled with seawater, hoisted back up on deck

into which a thermometer is inserted, and from which a temperature reading is

determined after some time as in situ. However, one probably thinks of a satellite

radiometer, which measures the actual, unaltered photons representing the exact

character of the substance in question and traveling at the speed of light, as a

remote measurement. One can see that mere proximity to the intended medium

may not assure the most accurate estimate of temperature.

A unique and quasi-direct method is one that measures the temperature at various

depths of a very stable borehole (e.g., in bedrock or ice cap) and then estimates what

the surface temperature would have been in the past to produce the temperatures

observed at each given depth. Simply put, the deeper the temperature reading, the

longer in the past it was inﬂuenced by the surface temperature.

Temperatures from all of the above devices are referred to as being part of the

‘‘instrumental’’ record and in some sense may be called direct measurements. There

are, however, several indirect methods available in which some organism or physical

process preserves in its histo ry the character of the environmental factors, including

temperature, affecting it. In this category of ‘‘proxy’’ records are tree rings, ice core

composition and thickness, isotope ratios in ice, sea ﬂoor, and sediment cores, pollen

distributions in sediments, erosion rates, coral bands, sea level height, evidence of

the extent of mountain glaciation, plant and animal fossil types and distributions, and

many more. With these types of proxy and borehole records, some estimates of the

climate prior to the instrumental record are possible.

The global temperature is a rather ambiguous term since every part of the global

system has its own temperature. When used in the context of clim ate change, it

usually means the temperature of Earth’s atmosphere about a meter or two above the

surface, often termed near-surface air temperature. However, one could speak of

the temperature of the land itself, or of the sea water at various depths, or of the ice in

the ice caps, or of the atmosphere at any of several altitudes. It is even possible to

measure the temperature of the ‘‘cold space’’ in which Earth orbits and ﬁnd a value

of about 2.7 K. Because Earth is a syst em of many interactive components, the

temperature of each is truly necessary to document global temperature.

Most data sets of global temperature are in fact not global in extent nor systematic

in quantity measured. So, not only are there variations in ‘‘how’’ and ‘‘where’’

temperature is measured, one must be careful to know ‘‘what’’ aspect of the Earth

system is being measured. Because we as humans experience weather and sustain

our existence on the surface of this planet, the near-surface air temperature is usually

the quantity of ﬁrst importance to us.

For all of the temperature data sets above and to which the term global is applied,

there are issues of uncertainty—spatial and temporal homogeneity, calibration

(or lack thereof ) of sensors and techniques, changes in instrumentation (type,

method etc.), degradation of instruments over time, corruption of proxies over

870 MEASURING GLOBAL TEMPERATURE

time, changes in local environment due to nonclimatic factors, poor infor mation on

observational practices, and many others. Each of these present potential problems

and are usually difﬁcult to assess so that the total level of uncertainty in any

measurement is not completely quantiﬁable. In the attempt to understand precisely

what the climate system is doing in terms of temperature, the unpleasant notion of

potential error is never totally absent.

The various measures of global temperature have been described in publications

too numerous to list. The four Intergovernmental Panel on Climate Change reports

(IPCC, 1990, 1992, 1996, 2001) are probably the best sources of information that

document the more prominent data sets.

1 PROXY RECORD OF SURFACE TEMPERATURES

The average of proxy-estimated temperatures of widely scattered sites prior to instru-

mental record gives a rough idea of what the temperature may have been in the past.

One set of these time series is shown in Figure 1 (see also IPCC, 1990, Fig. 7.1). Even

in these very low time-resolution diagrams, there is clear evidence of large variability

in global temperatures. Fluctuations in volcanism, solar insolation, orbital para-

meters, atmospheric composition (including changes due to human activities),

aerosols, natural chaos, etc. certainly play their roles, sometimes in isolation and at

other times in interdependent ways, as probable explanations for the variations.

Efforts have been attempted to quantify these causes, usually by a combination of

empirical calibration and analytical theory (e.g., Mann et al., 1998; Tett et al., 1997;

Wigley and Raper, 1990), but such discussion is not the focus of this chapter.

Figure 1 (top) shows two periods of interglacial (warm) temperatures over the

past 150,000 years, the current being the Holocene that began over 10,000 years ago.

The mid-Holocene ( 6000 years ago) was relatively warm as was the period about

1000 years ago (Fig. 1, middle and lower). There is considerable evidence that the

period between 1400 and 1850 was somewhat cool and is commonly called the Little

Ice Age, though recent evidence suggests that it was a period of several types of

climate ﬂuctuations in various parts of the globe. Solar variability and signiﬁcant

volcanism are among those forcing parameters thought to play roles in causing these

‘‘recently’’ depressed temperatures.

2 INSTRUMENTAL RECORD OF SURFACE TEMPERATURES

For most of the instrumental period, the past 150 years or so, temperature has been

determined from the familiar liquid-in-glass thermometers housed in various types

of shelters, usually 1.5 m above the ground, around the world’s land areas. Over the

oceans, the sea surface temperature (or SST, i.e., the temperature of the seawater, not

the air) is the preferred quantity because of the ocean’s more spatially and temporally

coherent temperature ﬁeld. The manner by which the SSTs are measured has varied

considerably through the years and requires careful adjustment factors to account for

2 INSTRUMENTAL RECORD OF SURFACE TEMPERATURES 871

Figure 1 Estimates of global temperature variations over three long periods ending with the

present day determined from proxy information (EarthQuest, Ofﬁce of Interdisciplinary Earth

Studies, Spring 1991, Vol. 5, p. 1).

872

MEASURING GLOBAL TEMPERATURE

biases introduced when, for example, buckets or ship- intake values are used

(Folland and Parker, 1995). The ‘‘global’’ temperature most commonly reported is

a combination of the near-surface air over land and SST reports over oceans,

however geographically sparse they might be.

Only a handful of serious efforts have achieved the goal of collecting as many

reports as are presently accessible to document the temperature in several places

around the globe over the past 150 years. (Scattered records go back further, one to

the midseventeenth century in central Eng land; see Manley, 1974.) The IPCC

usually focuses on the data sets produced by (1) the Climate Research Unit of the

University of East Anglia (near-surface air temperature over land; Jones and Briffa,

1992), (2) UK Meteorological Ofﬁce (SSTs; Parker et al., 1995), (3) Goddard

Institute for Space Studies, NASA (near-surface air temperature over land; Hansen

and Lebedeff, 1988), (4) Russian (near-surface air temperature over land, Vinnikov

et al., 1990), and (5) NOAA satellite-based SSTs (Reynolds and Smith, 1995). Two

other signiﬁcant efforts are the Global Historical Climate Network (NOAA=NCDC;

Vose et al., 1995) and a database of individual SST reports known as COADS

(Woodruff et al., 1987). At present, some of these groups are combining the data

sets of temperatures over land with SSTs to produce global coverage (e.g., IPCC,

1996; Hansen et al., 1996).

The time series of global, annual surface temperature anomalies for three of the

groups is displayed in Figure 2. (All of the time series produced from the various

groups are quite similar, and this is to be expected since the critical broad variations

depend on basically the same primary source data for each data set.) These time

Figure 2 Annual, global anomalies of surface temperature 1851–1997 as a combination of

near-surface air temperatures over land and SSTs over oceans (CRU=UEA, UKMO) and land-

only (NASA=GISS).

2 INSTRUMENTAL RECORD OF SURFACE TEMPERATURES 873

series were described in the IPCC (1996) as showing a 0.3 to 0.6



C rise over the past

century, the range being due to the uncertainty factors mentioned above.

The features of this time series are well-known; a fairly signiﬁcant rise from about

1910 to 1940 (0.4



C), then somewhat random variations to 1980 and a rise since

that time (an additional 0.2



C). A comparison of the most recent 30 years with

those of 1870–1899 reveal an incre ase in temperature of about 0.4



C while a

comparison of the most recent 10 years versus the earliest 10 (1860–1 869) shows

a rise of about 0.6



3 BOREHOLE TEMPERATURES

An insulated, homogeneous column of material with one end exposed to temperature

variations over a long period will reveal temperature variations throughout the

column according to the speed at which the ﬂuctuations propagate from the exposed

end. An analysis of the temperatures throughout the column at a given point in time,

then, may contain enough information to recover those external temperature varia-

tions. The general idea is that the greater the depth from the exposed end, the further

back in time the temperature at that depth might represent. Of course, as time passes,

the perturbations tend to smear, and it becomes difﬁcult to extract meaningful

information about what may have happened at the surface.

Certain types of stable, homogeneous bedrock and large ice plateaus lend them-

selves to temperature recovery through measurements taken in boreholes. The theory

and complications regarding the inversion of the temperature proﬁles into time series

are quite complex, but the general results show that many land regions have experi-

enced warming in the past three centuries, though the results show variations among

the individual sites (Deming, 1995). Also, evidence from boreholes on the Green-

land Plateau suggest that the period around 1000 years ago was warmer than the

present century (Cuffey et al., 1994; Dahl-Jensen et al., 1997). Both of these results

are consistent with the temperature estimates in Figure 1.

4 UPPER AIR TEMPERATURES

Upper air measurements, which could be compiled into large-scale averages, became

possible in the late 1950s due to the expansion of the network of radiosonde

stations—sites that release balloon-borne instruments to measure temperature,

wind, and humidity to altitudes up to and exceeding 10 km. Sources of large-scale

averages of temperatures at various levels and for various layers have been provided

by NOAA (Angell, 198 8; 63 stations; Oort and Liu, 1993; 800þ stations), UK

Meteorological Ofﬁce (UKMO) (Parker et al., 1997; 350þ stations), and Russian

IHMI-WDC (Sterin et al., 1997; 800þ stations).

Temperature time series constructed from individual radiosonde sites often

display inhomogeneities most often related to changes in the instrumentation or

874 MEASURING GLOBAL TEMPERATURE

the algorithms by which the raw data are processed into pressure-level data (Gaffen,

1994). These inhomogeneities can be quite prominent at the highest elevations

because errors or changes tend to accumulate as the balloon ascends. Oort and

Liu (1993) generate global maps by object ive analysis of their world-wide radio-

sonde dataset. The UKMO product uses selected stations with better records of

homogeneity and length and produce s a quasi-global analyses with limited inter-

polation for ﬁlling in vacant grids. As with Oort and Liu (1993), the RIHMI-WDC

produces global analyses by objective means but also applies a complex quality

control algorithm to the data to remove obvious inconsistencies that viola te

hydrostatic constraints and those of spatial coherency.

Since 1979, nine NOAA polar orbiting satellites have carried an instrument, the

microwave sounding unit, or MSU, designed to provide temperature information in

both clear and cloudy areas where infrared (IR) methods are ineffective. The sensor

measures the intensity of radiation near the 60-GHz oxygen absorption band, which

is proportional to atmospheric temperature. Though not intended to provide long-

term climate information, data from the nine MSUs, which have orbited since 1979,

have been calibrated and merged into a single time series (Christy et al., 1995). With

daily global coverage (over 30,000 observations per day) of a very robust quantity

(a volume of air over 50,000 km

per observation), these data have some advantages

over other types of data sets.

Two MSU temperature products are widely used: the lower troposphere (the

average temperature of the surface to about 7 km or 1000 to 400 hPa) and the

lower stratosphere (17 to 22 km layer or about 120 to 70 hPa). Version D of the

data described in Christy et al. (2000) takes into account recently discovered inﬂu-

ences on the satellites that affect the observations.

We have found in the several years of constructing the MSU data sets that the

issues of greatest impact on the long-term record are (1) the intersatellite biases, (2)

the orbital time drift, and (3) the orbit decay (see Wentz and Schabel, 1998). Because

the MSU data sets are products we produce, I shall devote a relatively large amount

of discussion to them.

Though calibrated to high precision on Earth in thermally controlled vacuum

chambers, the MSU, once in the environment of space, may acquire or display

unexpected characteristics. In one case (NOAA-12) an electronic gain change

occurred after launch so that the anticipated calibration target temperature of cold

space (2.7 K) was not correct, being measured at about 6 K. Corrections for this

effect were generated by Mo (1995). Also, the MSU, as a cross-track scanner,

monitors temperatures to the left and right of the track. In an unexpected result,

temperature comparisons of these two sides produce average differences as great

as 3



C and as little as 0.05



C in different instruments. This asymmetry is likely due

to variations in the antenna beam patterns from instrument to instrument—i.e., the

actual location of the ‘‘cone’’ through which the energy upwells to the sensor is not

as precisely located as anticipated. This is a systematic effect (it will not change

during the instrument’s life) but will produce overall biases if not accounted for.

Continuing with nonclimatic effects, there are two spurious consequences of the

slow east–west drift of a satellite. One is that the MSU observes Earth at later or

4 UPPER AIR TEMPERATURES 875