Continued from previous page
Since the DTR is the maximum temperature minus the minimum temperature, the
DTR can decrease when the trend in the maximum or minimum temperature is downward,
upward, or unchanging. This contributes to less spatial coherence on the DTR
map than on maps of mean temperature trend. Maximum temperatures have increased
over most areas with the notable exception of eastern Canada, the southern United
States, portions of Eastern and southern Europe (Brunetti et al., 2000a), southern
China, and parts of southern South America. Minimum temperatures, however, increased
almost everywhere except in eastern Canada and small areas of Eastern Europe
and the Middle East. The DTR decreased in most areas, except over middle Canada,
and parts of southern Africa, south-west Asia, Europe, and the western tropical
Pacific Islands. In some areas the pattern of temperature change has been different.
In both New Zealand (Salinger, 1995) and central Europe (Weber et al., 1994;
Brázdil et al., 1996) maximum and minimum temperatures have increased
at similar rates. In India the
DTR has increased due to a decrease in the minimum temperature (Kumar et al.,
1994). Eastern Canada also shows a slight increase in DTR due to a stronger
cooling in maximum temperatures relative to minimum temperatures (Easterling
et al., 1997). However, recently annual mean maximum and minimum temperatures
for Canada have been analysed using newly homogenised data (Vincent, 1998; Vincent
and Gullet, 1999); these have increased by 0.3 and 0.4°C, respectively,
over the last fifty years (Zhang et al., 1999). Central England temperature
also shows no decrease in DTR since 1878 (Parker and Horton, 1999). Similarly,
a new temperature data set for north-east Spain (not available on Figure
2.2 below, Brunet-India et al., 1999a,b), shows an increase in maximum temperature
over 1913 to 1998 to be about twice as fast as that of minimum temperature.
Recent analyses by Quintana-Gomez (1999) reveal a large reduction in the DTR
over Venezuela and Colombia, primarily due to increasing minimum temperatures
(up to 0.5°C/decade). In northern China, the decrease in DTR is due to a
stronger warming in minimum temperature compared with maximum temperatures.
However, in southern China the decreased DTR is due to a cooling in maximum
with a slight warming in minimum temperature (Zhai and Ren, 1999).
Figure 2.4: Tests of bias adjustments to sea surface temperature
(SST) using a climate model (Folland et al., 2001). Black line - annual
mean observed land surface air temperature (SAT) anomaly (°C) from
a 1946 to 1965 average (Jones, 1994), a period before major anthropogenic
warming. Red line – annual averages of four simulations of SAT
anomalies using uncorrected SST data, 1872 to 1941, and an average
of six simulations for 1941 to 1998. Blue line – average of six
simulations of SAT, forced with SST data corrected up to 1941. |
|
Figure 2.5: (a) Annual anomalies of global SST (bars and solid
curve) and global night marine air temperature (NMAT, dotted curve),
1861 to 2000, relative to 1961 to 1990 (°C) from UK Met Office analyses
(NMAT updated from Parker et al., 1995). Smoothed curves were created
using a 21-point binomial filter to give near-decadal averages. Also
shown are the equivalent SST anomalies from the SAR – dashed
curve. (b) Smoothed annual global SST (°C), 1861 to 2000, relative
to 1961 to 1990, from USA National Climate Data Centre, Quayle et
al. (1999) (thin dashed line, includes satellite data); USA National
Centres for Environmental Prediction, Reynolds and Smith (1994) and
Smith et al. (1996) (thin solid line, includes satellite data, to
1999 only), and UK Met Office (Jones et al., 2001) (thick line). (c)
UKMO SST and NMAT anomaly time-series from a 1961 to 1990 average
for the Northern Hemisphere. (d) As (c) but for the Southern Hemisphere.
Both for 1861 to 2000. |
|
The DTR is particularly susceptible to urban effects. Gallo et al. (1996) examined
differences in DTR between stations based on predominant land use in the vicinity
of the observing site. Results show statistically significant differences in
DTR between stationsassociated with predominantly rural land use/land cover
and those associated with more urban land use/land cover, with rural settings
generally having larger DTR than urban settings. Although this
shows that the distinction between urban and rural land use is important as
one of the factors that can influence the trends observed in temperatures, Figure
2.2 shows annual mean trends in diurnal temperature range in worldwide non-urban
stations over the period 1950 to 1993 (from Easterling et al., 1997). The trends
for both the maximum and minimum temperatures are about 0.005°C/decade smaller
than the trends for the full network including urban sites, which is consistent
with earlier estimated urban effects on global temperature anomaly time-series
(Jones et al., 1990).
Minimum temperature for both hemispheres increased abruptly in the late 1970s,
coincident with an apparent change in the character of the El Niño-Southern
Oscillation (ENSO) phenomenon, giving persistently warmer sea temperatures in
the tropical central and east Pacific (see Section 2.6.2).
Seasonally, the strongest changes in the DTR were in the boreal winter (-0.13°C/decade
for rural stations) and the smallest changes were during boreal summer (-0.065°C/decade),
indicating some seasonality in the changes. Preliminary extensions of the Easterling
et al. (1997) analysis to 1997 show that the declining trends in DTR have continued
in much of North America and Asia.
Figure 2.3 shows the relationship between cloudiness
and the DTR for a number of regions where long-term cloud cover data are available
(Dai et al., 1997a). For each region there was an increase in cloud cover over
the 20th century and generally a decrease in DTR. In some instances the correlation
between annual cloud cover and annual DTR is remarkably strong, suggesting a
distinct relationship between cloud cover and DTR. This would be expected since
cloud dampens the diurnal cycle of radiation balance at the surface. Anthropogenically-caused
increases in tropospheric aerosol loadings have been implicated in some of these
cloud cover changes, while the aerosols themselves can cause small changes in
DTR without cloud changes (Hansen et al., 1998 and Chapter
6).
|
Box 2.2: Adjustments and Corrections to Marine Observations.
The SST data used here comprise over 80 million observations from the
UK Main Marine Data Bank, the United States Comprehensive Ocean Atmosphere
Data Set (COADS) and recent information telecommunicated from ships and
buoys from the World Weather Watch. These observations have been carefully
checked for homogeneity and carefully corrected for the use of uninsulated
wooden and canvas buckets for collecting seawater prior to 1942. However,
corrections prior to about 1900 are less well known because of uncertainties
in the mix of wooden and canvas buckets. Nevertheless, Figure 2.4 provides
good evidence that even in the 1870s, SST was little biased relative to
land-surface air temperatures globally. Since 1941, observations mainly
come from ship engine intake measurements, better insulated buckets and,
latterly, from buoys. SST anomalies (from a 1961 to 1990 average) are
first averaged into 1° latitude by 1° longitude boxes for five-day
periods; the anomaly for a given observation is calculated from a 1°
box climatology that changes each day throughout the year. The five-day
1° box anomalies are then aggregated into 5° boxes for the whole
month with outlying values rejected, and monthly average anomalies calculated.
Further adjustments are made to monthly SST anomalies for the varying
numbers of observations in each 5° box because when observations are
few, random errors tend to increase the variance of the monthly mean.
NMAT data are treated similarly and have quite similar characteristics.
However, a variance adjustment to NMAT data is not yet made. NMAT data
are also corrected for the progressive increase in the height of thermometer
screens on ships above the ocean surface, though no corrections have been
made since 1930. Because there are only about half as many NMAT as SST
data and NMAT have smaller temporal persistence, monthly NMAT anomalies
may be less representative than SST anomalies even on quite large space
scales. On longer time-scales, and over the majority of large ocean regions
in the 20th century, there is good agreement between NMAT and SST. 19th
century NMAT anomaly time-series should be
|
