ÐÓ°ÉÔ­´´

There are those who predict imminent apocalypse as a result of more
greenhouse gases in the atmosphere and those who think that nothing much
will change. This dichotomy is possible because the debate is founded not
on fact but on the predictions of models which no one claims are very precise.
To recognise, let alone understand, changes in our climate, we need to know
what is actually happening. We do not. In fact, all our climate data are
pretty poor stuff.

One definition of climate is that it is the average of the weather over
decades and centuries, so that a climate record is the product of the day-to-day
readings that comprise weather data. Phenomena measured include air temperature
and humidity, wind speed and direction, amounts of rainfall, snow, solar
radiation (and how it is redistributed), cloud cover, the hydrological cycle,
the temperature of the sea and its currents, and ice and snow cover. To
detect the full spectrum of climate change, all of these need to be measured,
but my concern here is with the basic components of ‘weather’ – temperature,
wind, sunshine, rainfall – and how to measure them.

Many data now come from satellites, producing images in visible and
infrared light. From this information, meteorologists can pick out cloud
patterns, and estimate rainfall from the temperature of the tops of clouds.
Images using different wavelengths of radiation focus on specific weather
characteristics such as ground temperature, air humidity and snow cover.
But satellites cannot provide everything that the climatologists need, although
in the future they will be increasingly important sources of information.
Data from satellites must be supplemented with ground-level data to calibrate
their results; alone their accuracy is not high.

But this ‘ground-truth’ is the very information that is missing or inadequate.
Historically, most of these records have been made by national meteorological
services for weather forecasting, which gained importance in the first half
of this century because of the demands of shipping and aviation. Most of
the instruments used today to collect these data are still of the simple
types developed in the 19th or early 20th century. At a typical site, a
meteorologist reads air temperature from a mercury-in-glass thermometer
surrounded by screens of slatted wood which shield it from sunlight while
allowing air to flow freely around it. Thomas Stevenson, a Scottish lighthouse
engineer and the father of Robert Louis Stevenson, designed this type of
screen in 1866. A second thermometer within the Stevenson screen is equipped
with a wick dipped in water that keeps its mercury bulb wet. Evaporation
keeps this thermometer cool, and the drop in temperature relative to the
other instrument is used as a basis for calculating air humidity.

Meteorologists still estimate wind direction by looking at a vane, although
no longer one shaped like a cock or an arrow. The speed of the wind is measured
by an instrument consisting of three cups which rotate on a shaft, called
the cup anemometer, designed in 1846 by Thomas Romney Robinson, a clergyman
and astronomer from Armagh. The wind produces more pressure on the open
side of each cup than on its back to make the shaft spin, giving surprisingly
precise readings.

The most elegant of these old instruments was the sunshine-hours recorder,
designed in 1853 by John F. Campbell of Islay, near London, an employee
of the Board of Health, and modified in 1880 by Sir George Stokes, a member
of the Meteorological Council. It is known internationally as the Campbell-Stokes
recorder and can still be seen at most of the world’s weather stations,
as well as at many seaside resorts. A glass ball focuses an image of the
Sun into a bright dot which burns a line onto a strip of cardboard through
hours of the day. The card is calibrated so that the length of the track
gives the total number of sunshine hours each day. Many national meteorological
services still use this instrument, even though much more precise solar
radiation sensors now exist.

The quality of data from these traditional instruments very dependent
on how well they are maintained and how good their operators are. These
factors vary a great deal from place to place and from time to time; much
research must have been based on data of unknown and, worse, unquestioned
quality. But the quality of the data is not just a matter of the operator
taking care, as the example of rainfall shows. Rain is caught in a funnel,
collected in a bottle, and measured by the meteorologist each day. But the
process of ‘catching’ rain turns out not to be as easy as it sounds.

In 1766, William Heberden the elder, a celebrated London physician,
and one of a growing number of amateur meteorologists, noticed that the
higher a rain gauge was placed the less rain it tended to catch, compared
with others lower down. He experimented with rain gauges on the towers of
Westminster Abbey, but the same effect happens with gauges on the ground
at different altitudes. Heberden could not explain why.

A century passed before William Stanley Jevons showed that the rain
gauge itself acts as an obstacle to the flow of air over it. In the same
way that an aerofoil gains its lift, the wind speeds up as it flows across
the mouth of the gauge, carrying away some of the raindrops that would otherwise
have fallen into the gauge. The stronger the wind, the greater the effect
and, as Jevons observed, winds become faster with height.

As much as 80 per cent of the catch can be lost in some conditions,
especially in drizzle when the drops of water are so small that the wind
can carry away a high proportion of them. This can be an enormous source
of error, especially compared with the other rain gauge errors – from water
lost by splashing, evaporation and the inclination of the rain gauge, which
each affect readings by less than 1 per cent.

Building a better rain gauge

In 1878, to reduce the wind effect, Francis Nipher, an American meteorologist,
devised a shield to deflect the wind downwards. Earlier, in 1842, Stevenson
had tried a different approach, putting the gauge in a pit with its opening
at ground level; he also laid bristle matting around the gauge to cut down
on splashes. This system was improved in 1934 by the German meteorologist
Koschmieder, who replaced the bristles with a metal grating resulting in
even fewer splashes. In 1928 an English meteorologist, Frank Huddleston,
further refined the landscape of the rain gauge by surrounding the instrument
with a turf wall, a sloping circular mound of earth about 4 metres in diameter
and 30 centimetres high that surrounded the gauge and shielded it from the
wind.

In the past few decades, the search for a better rain gauge has continued,
taking in the aerodynamics of funnels by testing new streamlined profiles.
But despite centuries of investigation, most gauges are still without any
protection against wind. And even a perfect rain collector cannot do a perfect
job. Rain does not fall evenly; two gauges only a few metres apart will
collect amounts that differ by several per cent. The way to resolve this
mismatch could be by combining ground truth measurements with remote sensing
methods, such as ‘weather radar’ which detects raindrops over a large area
in the same way that air traffic controllers detect aircraft.

But the worst shortcoming of the weather network is the distribution
of its weather stations. Because the old-style meters and gauges need operators,
most established stations are close to populated areas; the Earth’s uninhabited
areas and the oceans hardly exist in the weather records. Instruments near
or in cities can also introduce bias in the readings, because cities are
warmer than their surroundings – the ‘heat-island’ effect. With so many
engines, people and heated and air-conditioned buildings, cities are warmer
than the surrounding countryside and do not cool as much at night. Other
climate variables, such as wind speed and direction, and rainfall, also
tend to be different in cities. Ideally, weather stations should represent
the surrounding country and should not change substantially over the years.

Measurements representative of ocean climates are another special case.
Of the rare readings, most come from instruments on islands, by definition
different from their watery surroundings. Many instruments are situated
conveniently close to airstrips, to provide relevant information for pilots.
There they are also handy for the observers who take the readings. At an
airport, instruments are surrounded by acres of tarmac; the local climate
is very unlike that of the sea. But the majority of the islands do not even
have instruments. What few open sea readings there are come from merchant
ships, but otherwise meteorologists depend on satellite information.

Nor are matters helped by there being few widely followed standards
for the construction or siting of the instruments themselves. The only agency
with any voice in this is the World Meteorological Organisation, based in
Geneva, which holds regular conferences and has a committee dealing with
the topic. But national meteorological services are highly conservative,
sticking to their long-established practices, partly to keep costs down
and partly to maintain continuity – even if this also means maintaining
imprecision.

Yet the advent of microelectronics has brought precise automatic monitoring
of the weather within reach, even in the toughest conditions. At less than
£1000 each, today’s data loggers – the recording part of a modern
automatic weather station – can run for several years on one lithium battery
or indefinitely on solar power, storing a year’s weather data in a small
memory. Similarly, outstations anywhere on Earth can relay data to base
via satellites.

Modern loggers can also extract more useful information from traditional
sensors. A switch on the shaft of a cup anemometer can measure wind speeds
at any instant as well as average speeds and gusts. Electronic vanes can
monitor instantaneous wind direction far more precisely than a simple weather
vane, as well as giving average values. Miniature platinum resistance thermometers
respond rapidly and give precise average temperatures, while sensors based
on a water-absorbing polymer film only micrometres thick give humidity measurements
to 3 per cent. The Campbell-Stokes recorder, with its spots of sunlight
on a card, has been replaced by electronic devices to measure solar energy.
There are also sensors that record the difference between the total incoming
energy from the Sun and atmosphere and radiation from the ground – the only
weather variable directly influenced by greenhouse gases.

Rain is still collected by a funnel, but modern instruments can also
record the rate at which it falls. Each tenth of a millimetre of rain fills
a ‘tipping bucket’, sending an electric pulse to a recorder to provide totals
over any period or intensities from minute to minute. The bucket itself
is another old idea, originated by Sir Christopher Wren in 1662.

Good as these new techniques are, there are situations that require
special strategies – the oceans, for one. Automatic equipment can be fitted
to buoys, but each one sited on the deep ocean costs £100 000. Although
ships have collected data for years, Britain, for example, now has only
one permanently moored weather ship in the north Atlantic. The rest of the
national data from this part of the Atlantic, which influences so much of
Britain’s weather, comes from merchant ships, some of which launch transmitting
sensors attached to weather balloons. They provide fewer than 50 readings
a week, data that do not come at set times nor from set positions. It was
this lack of data that caused the gales of 1987 to go undetected. The Pacific
is similarly served, largely by Canada.

Automatic monitoring could be a cheap way to improve coverage on islands,
because the very compact equipment could be set up on isolated rocks or
far from human habitation on islands small enough not to modify the oceanic
climate. But there are also areas of land without even traditional weather
stations; notable among these are the cold deserts of the poles, where models
suggest that the effects of climate change may be extreme. Here, even setting
up instruments presents major difficulties. Access is an obvious problem,
but others are more subtle. One is the type of weather: snowfall is one
of the most difficult meteorological events to measure, especially in stormy
regions. The snow that settles at one point may contain a significant amount
that has drifted in from elsewhere; fallen snow can blow from place to place
as spindrift, complicating the pattern even more.

The problem of interference with patterns of wind flow, so often experienced
with rain gauges, is more severe with snow collectors because snowflakes
are more easily carried by wind. Once snow is in the collector, the usual
way to measure the amount is to melt it – and this demands a power source.
Solar power is no good, not only because of the long polar winters in which
the Sun does not rise, but also because solar panels cannot deliver the
large amount of power required. Meteorologists are developing other methods,
such as one based on the scintillation produced as snowflakes pass through
a light beam; these have yet to be fully tested. Also, they are expensive
and take considerable power to keep their optics free of ice.

Because of the problems of catching snow, it is often meas-ured after
it has settled on the ground. Using its depth can be misleading because
its density will vary, so meteorologists try to find out how much water
the snow represents. Even for an observer in the field, this needs sophisticated
nuclear radiation instruments; automating the process is even more complex.
An alternative way is to measure the snow as it melts, collecting the water
in a large gauge. But layers of ice in the packed snow can divert the trickle
of water, which in turn disrupts the readings. Nor is this method useful
in telling you how much snow fell, or when; it cannot be used at all in
places where the snow does not melt.

An additional difficulty in cold regions is that snow and ice sticks
to the sensors. The usual solution is to throw kilowatts of heat at the
problem, but this is possible only where power is plentiful – rarely the
case at these remote sites. Britain’s Institute of Hydrology has investigated
new de-icing methods which do not rely on power or heat. Instead, they suggested
that the outer structure and sensors of an automatic system should be made
of hydrophobic materials, which both repel water and reduce the adhesive
strength of ice. This was combined with jolting the sensors mechanically,
using air from an aqualung, which also flexes the outer surfaces to shake
the ice off. This three-pronged defence kept experimental stations free
of ice through several winters in the Cairngorm mountains and for a winter
at the British Antarctic Survey base at Faraday. The BAS is now in the process
of installing semi-automatic stations with manual de-icing at five of its
bases. But the great expanse of Antarctica remains largely unmonitored as
do both the Arctic and most of the world’s great mountain regions.

Although there are still problems to be solved in collecting the sort
of data that climate modellers need to verify their predictions, we have
the technology. There are instruments capable of monitoring the climate
to a much higher precision, and at a lower cost, than the old-style instruments
can achieve. Gone is the need for a daily operator and with it the need
to keep instruments close to populated areas. In contrast, most of the existing
long-established instrument networks of the world’s national meteorological
services, on which the modellers still have to rely, are labour-dependent,
archaic, imprecise and cover barely one-tenth of the Earth’s surface adequately,
making them particularly unsuited to the task of monitoring the climate
change that may well just be upon us.

For if the climate modellers are right we are about to experience over
the next 50 years changes that would normally have taken thousands of years.
Change will be apparent, some modellers say extreme, within a generation,
yet we are totally unprepared to measure what happens, still being largely
dependent on last century’s technology to record events.

The national meterological services of a few of the richer countries
have begun to automate their networks of instruments, but there is no coordinated
programme. Indeed there is now less rather than more investment in ground-based
data collection, with an increasing reliance on the more easily obtained,
but low-resolution satellite data – not so good for forecasting, even less
so for climatology.

Unless the situation is improved, perhaps by a World Meteorological
Organisation initiative to modernise and expand ground-based instrumentation
globally, we run the risk of going to considerable expense trying to predict
what the climate might do, while neglecting to find out what it actually
does.

Ian Strangeways was formerly head of applied physics at the Institute
of Hydrology and now runs a consultancy in environmental monitoring and
instrumentation.

Further Reading: The Invention of Meteorological Instruments by WE
Knowles Middleton, published in 1969 by the Johns Hopkins Press

* * *

Climate figures at your fingertips

Climate data – some of it dating back to 1832 – is readily available
to personal computer users in Australia. In what is being billed as a world
first, a small company based in Melbourne has transferred data from 15 000
climate and rainfall stations onto a compact disc (CD) with read-only-memory
(ROM). A software program called Supermap allows access to the data base.

The data has been compiled by the Australian Bureau of Meteorology.
Rainfall has been recorded at some 15 000 stations, in many cases for well
over 100 years. Just over 900 of these stations have also recorded temperature
and humidity on a daily basis. The data have been used by a software company,
Space-Time Research, to provide a range of climatic information including
dry-bulb and wet-bulb temperature, dew-point temperature, relative humidity,
mean daily maximum and minimum temperatures, and mean rainfall.

Rainfall data from the 15 000 stations includes the number of ‘raindays’
and the total rain for each month, each year. Each site’s latitude and
longitude is given.

‘We have taken an extraordinarily large historical rainfall and climate
database, previously only available on main frames or very powerful minicomputers,
and brought it down to the level of a notebook computer,’ said Jack Massey,
director of the software company.

Massey stressed that the information has great practical value. A farmer
could plan for the best yield from crops on the basis of the typical climate
for a particular area at a particular time of the year. And engineers and
town planners could use the data to plan construction work.

But this data, with its ease of access, is equally valuable in other
areas. The system will be sold relatively cheaply to schools and libraries
– about a sixth of the price for corporate users. ‘We see this as a very
good way of students understanding how to access and use a data base and
learning something about climate in the process,’ said Massey. Curriculum
materials to support the use of the system in secondary schools are being
developed.

The CD ROM system can display weather patterns in a number of ways including
dot maps, pie charts, histograms, and bar charts. It can be accessed through
a computer network or used in isolation. The data will be updated annually.

The ability to analyse climate trends and to make forecasts will be
included in future software. Also, Space-Time Research is negotiating with
a US company to make a similar product available in North America.

Ian Anderson