Figure 1
Changes in mean global
surface temperature from the 1951-1980 average from 1880-1994.U.S. Global Change Research Program.
Figure 2
Changes of maximum and minimum temperatures.IPCC, 1990.
Figure 3
Annual precipitation anomalies.IPCC, 1990.
Figure 4
Temperature anomalies in the troposphere and lower stratosphere 1958-1989. IPCC, 1990.
Figure 5
Sub-surface ocean temperature changes and Northern hemisphere snow extent anomaliesIPCC, 1990.
Figure 6
Sea-ice changesIPCC,
1990.
Introduction: Measurements of meteorological variables with sufficient spatial coverage to represent global observations began in the mid to late 1800s. Measurements are taken (1) a few feet (~1.6 m) above the earth's surface over land, (2) at various elevations above the surface over land by launches of weather balloons every 12 hours, and ( 3) a few feet above the water surface over oceans. On-going measurements also are taken by satellites, but this topic is covered in Unit 2-11 . It is important to remember that, in our discussions about global warming, the term "surface temperatures" really means air temperatures taken a few feet above the surface of the earth. These are the temperatures that normally are reported in the media associated with weather forecasts.
Measurements
Measurements made at land stations include temperature, dew point, relative humidity, precipitation, snowfall, snow depth on ground, wind speed, wind direction, cloudiness, visibility, atmospheric pressure, evaporation, soil temperatures, and various types of weather occurrences such as hail, fog, thunder, glaze, etc. In the US, weather observations are normally taken at hourly intervals, but sometimes are taken more frequently under rapidly changing conditions. New and automated weather stations report a more limited list of meteorological variables at 20 minute intervals.Weather data from the atmosphere above the earth's surface (above about 3 meters) are taken by instrument packages (radiosondes, rawinsondes, and rocketsondes) that are carried aloft by weather balloons that are launched every 12 hours (in the US and many other locations, but less frequently in developing countries) and transmit data to a receiving station on the ground. These "upper air" data include temperature, relative humidity, atmospheric pressure, and wind. Vertical profiles of atmospheric measurements taken at approximately the same time are called atmospheric soundings.
Measurements over oceans are taken at the locations of buoys that may be either moored or floating and from ships at sea. These observations normally are reported at hourly intervals. A global ocean observing system, GOOS, is in the planning stages that envisions "a global network to systematically acquire, integrate and distribute marine and oceanic observations, and to routinely produce analyses and forecasts and related products to enable governments, industry, science and the general public to cope with critical marine-related issues including the ocean's effects on climate."
Research teams from the US, Great Britain, and Russia have independently analyzed global records of air temperature taken at surface. These data sets begin about 1860 and extend to the present.
Global Mean Surface Temperature
Studies of global warming that rely on surface air temperature measurements generally use plots of global (or hemispheric) mean temperature, Figure 1, that are expressed relative to some base temperature. Typically the 1951-1980 average or the 1951-1990 average is used as a reference and individual annual temperatures are expressed as their departure from this mean value.
The global mean surface temperature as constructed from thermometer records taken over land shows that temperature increased from the relatively cool late nineteenth century to the relatively warm 1990s. The pattern of change differs somewhat between the Northern and Southern Hemispheres and between land and sea. In the Northern Hemisphere the temperature changes over land are irregular but fluctuate about a mean level of about -0.3°C and shows a rather abrupt warming in the 1920s. From this point the temperature continues to rise to a maximum in the late 1930s and then starts a downward trend lasting about 30 years. Since about 1970, the temperature record shows a very substantial rise, with the rate being equal or larger than the rise of the 1920s.
The record for the Southern Hemisphere is similar but does not have the abrupt rise in the 1920s. Rather the record shows a more gradual increase from about 1890 to the late 1930s. The 30-year cooling trend starting in the late 1930s also is apparent, although of lower amplitude in the Southern Hemisphere . The warming since 1970 also shows up in the Southern Hemisphere record, although again with lower amplitude.
Problems Leading to Uncertainty
There are several problems that lead to uncertainty in constructing these hemispheric and global mean temperature records (Houghton et al, 1990):
- Spatial coverage of the data is incomplete and varies greatly.
- Changes have occurred in observing schedules and practices.
- Changes have occurred in the exposures of thermometers.
- Stations have changed their locations.
- Changes in the environment, especially urbanization, have taken place around many stations.
Problems due to item 1 are most pronounced before 1900. Studies on the data sets after 1900 where newer stations are successively included and excluded show that the spatial coverage beginning in 1900 is relatively representative of the entire globe.
Houghton (1990) has considerable discussion on each of these items. They conclude that item 5, and urbanization in particular, is the largest source of possible error in the temperature record. As cities have expanded, formerly rural weather stations become surrounded by, or at least are influenced by urban landscapes, which are known to have higher surface air temperatures. This source of error differs from the others primarily because, whereas the others may have some stations giving positive errors while other may give compensating negative errors, all stations subjected to urbanization will cause a monotonic warming of the record. However, analysis of temperature records from stations known to be continuously located in rural areas show the same general pattern as the complete data set of all stations.
Oceans account for 61% of the surface area in the Northern Hemisphere and 81% of the area in the Southern Hemisphere. Over the ocean the temperature patterns generally resemble the land-based record. The Northern Hemisphere has a general cooling from the early part of the record until about 1910 when a relatively abrupt warming, similar to that observed over land, begins and lasts until about 1940 when the record shows a period of fluctuations but little change in the mean. A brief drop in temperature starting in the late 1960s is followed by a moderately steep rise to the present values.
In the Southern Hemisphere ocean, there is little trend before about 1930 when a brief but abrupt rise is followed by a leveling off from about 1940 to 1970. After 1970 a moderate upward trend is revealed.
The combined temperature record shows a relatively stable period from 1860 - 1910 followed by a rise to about 1940. After this time the record has relatively low trend but with higher variability until about 1970 when the rather abrupt rise to the present value begins.
If we look separately at daily maximum and daily minimum temperatures, Figure 2 , we find that a rise in daily minimum temperatures is accounting for most of the warming that has been observed, at least in the US and certain parts of the world. This could be attributed to both rise in greenhouse gases and increases in cloudiness, which will have a larger influence on nighttime minimum temperatures than daytime maximum values.
Analysis of Precipitation
Analysis of precipitation records is even more difficult than temperature. Errors in precipitation records include wind speed during rain/snow events that causes droplets/snowflakes to be deflected from falling in the gauge by flow patterns around the instrument. This error always leads to reports of rain being less than actually occurred, so it (like urban heat islands were for temperature) does not experience a compensation effect when a large number of stations are aggregated. Elaborate means can be taken to compensate for this, but such accurate instruments require more expense and maintenance. Wetting of the gauge in light precipitation events also can lead to under-reporting of precipitation amounts. Even if the gauges were accurate, the spatial variability of rainfall is much larger than that for temperature, so intense rain events can be missed completely by a sparse network. Also, intense but isolated rain even at one station in a sparse network can be erroneously inferred to be representative of rainfall over a large area. Because this representativeness problem may cause either positive or negative errors, a large measure of compensation usually occurs.
Analysis of precipitation records, Figure 3 , over the globe show some regions seemingly experiencing a trend toward increasing precipitation and others decreasing. Subtle changes in large-scale flow patterns can influence precipitation patterns, so it is difficult to draw general conclusions.
Temperatures in the atmosphere above the surface, Figure 4 , as measured by balloon-borne instruments and satellites have been analyzed for trends over the period of record which, for these observations, is only about 50 years long. The middle troposphere from about 1,500 m to 7,000 m (850 mb - 300 mb) shows a general warming since about 1965 of about 0.4°C. The upper troposphere (300mb - 100 mb) shows a cooling trend in general disagreement with most global climate models. The lower stratosphere generally shows a substantial cooling, which is in agreement with most model results and is due to reduced ozone absorption of UV radiation and increased warming at the surface which deepens convection.
There are some measurements, Figure 5, taken from the North Atlantic and North Pacific Oceans that suggest a warming of the ocean layer from 500 m to 3000 m depth. Areal coverage by snow, Figure 5 , in the Northern Hemisphere shows a gradual decline over the last 20 years during which satellite measurements have been made. Sea ice area, Figure 6, over the same period shows a slight decline in the Southern Hemisphere and essentially no change in the Northern Hemisphere. Perhaps a more meaningful measure of change in sea ice volume, however, is sea-ice thickness rather than horizontal areal coverage. And there have been reports of substantial thinning of ice in recent years, although the record is quite short.
Houghton, J. T., G. J. Jenkins, and J. J. Ephraums, 1990: Climate Change. The IPCC Scientific Assessment. Cambridge University Press. 365 pp.