The application of the instruments described in the previous sections to studying the processes involved in climate change is fairly obvious. For example, absolute radiometers (ACRIM) measure the output of the Sun; broadband radiometers (ERBE, GERB, CERES) monitor the Earth's radiation budget; spec-troradiometers (ATSR, MODIS, HIRS, AIRS) measure surface and atmospheric temperatures; interferometers (IRIS, ATM OS, TES) measure composition, including greenhouse-gas amounts. Collectively and in various combinations, these instruments and others can monitor and study the general circulation of the atmosphere and its cloud and aerosol regimes, both crucial but elusive components of the climate system.
However, the measurement and detection of climate change is much more difficult. The traditional method has been to synthesize a global mean surface temperature from thousands of measurements made by different thermometric instruments around the world, over an extended period of time. This is obviously an arduous procedure, and one that is subject to many different types of error. Given the urgency of detecting quite small changes and trends before they become large enough to be detrimental to human affairs, there is a clear need to bring modern technology to bear. Radiometric instruments in space offer convenient global coverage by a single instrument, or at least a set of nominally identical and intercalibrated instruments, and therefore a potential solution to many of the problems inherent in the use of fixed, ground-based networks.
What, however, are the best parameters to measure, and how should the data be utilized, in order to produce the most sensitive and least error-prone estimates of global change, if indeed change is occurring? Since we are interested in future projections of climate change, we have no choice but to use forecasting models. The role of measurement then is to test and validate the models using relevant global data that are accurate and calibrated absolutely. The more obvious approaches, like using an ATSR-type instrument to map global temperatures and then taking global and annual averages, work in principle but in practice are affected by surface emissivity variations and cloud contamination of the radiances. Also, this approach does not use all of the available information in radiance measurements from space. Attempts to use radiometric measurements of the tropospheric temperature profile have focused on microwave sounders, which are less prone to interference by cloud effects, since most clouds are transparent at these long wavelengths. However, these also have problems with variable surface emissivity, and in particular with the vertical response function, which covers the whole troposphere and part of the lower stratosphere. The effect of this is to confuse and mask the climate signal since the temperature changes at various heights, which can oppose each other, are convolved together in an inconvenient way.
Goody et al. (1998) considered this whole question carefully and published a summary with recommendations. In this they stress the difference between observing forcings (changes in the key climate processes, such as those described in the opening paragraph of this section) and observing a response (such as a change in global or regional mean surface temperature or precipitation patterns). The latter requires the climate-change signal to be specified and benchmarks for this signal to be recorded at different times. The existence of anthropogenic forcing is not in serious doubt, and therefore we expect that the signal should, in time, emerge from the natural variability. The natural variability in the signal is itself of interest, both intrinsic and because knowledge of the statistics of natural variability is required to determine when the signal will emerge. Natural variability on timescales longer than the measurement period cannot be distinguished from an emerging signal; we must have independent data on the natural variability before we can interpret the data. This may come from the record to be found in archives of climate data, but because of the many limitations of old data it is more likely to be obtained from the chaotic behaviour of a general circulation model.
The chances of detecting a climate-change signal tend to improve if the signal selected is as complex as possible, in order to use as much measured information as possible and to maximize the chance that the signal will be distinct from natural variability. Goody et al. consider two examples of such 'fingerprints': the atmospheric refractivity, measured by transmitting radio waves between satellites in orbit along a path that intersects the atmosphere as one enters occultation behind the Earth as seen from the other, and the spectral radiance emerging from the top of the atmosphere, as measured by an interferometer equipped with an absolute calibration system. In order to maximize its complexity they manufacture the fingerprint from refractivities at 20 different heights below 25 km, 30 different locations around the globe, and all four seasons. For a fingerprint based on radiances the 20 heights are replaced by 1000 different quasimonochromatic wavelengths between 400 and 1400 cm-1, a range that accounts for 98.3% of the radiance from a blackbody at the mean radiating temperature of the Earth.
Figure 10.13 illustrates the power of these complex fingerprints. The four frames show the change in the radiance leaving the top of the atmosphere, at the spectral resolution of the IRIS instrument (§10.8) calculated for four different forcing factors. These factors are (a) a doubling of CO2 from 330 to 660 ppm; (b) an increase in the water-vapour amount, through the effect of temperature on the
800 1000 1200 1400
800 1000 1200 1400
flG. 10.13. The calculated effect on the outgoing radiance spectrum of the Earth of (a) doubling the CO2 concentration, (b) increasing the relative humidity of the entire troposphere, (c) increasing the solar flux, and (d) changing the cloud cover. In each case the change in radiance over the unperturbed value is shown. The perturbations were chosen in cases (b), (c) and (d) to give the same change in surface temperature as in (a), namely 1.63 K. (Goody et al. 1998)
relative humidity; (c) an increase in the solar flux from 300 to 305.5 W m-2; (d) the introduction of a model cirrus cloud. These all produce a readily measurable change in the climate signal, i.e. one that is much larger than the instrumental noise or calibration uncertainty, and are chosen so they all change the surface temperature by the same amount, namely 1.63 K. However, even before numerical analysis it is clear that they produce different and distinguishable fingerprints. With some assumptions, probably optimistic, Goody et al. found that the signature of global warming due to increasing CO2 should be detectable when the concentration of the gas has risen by just 6.2% and, in the model they used, the surface temperature has risen by 0.2 K.
Attempts to apply this promising approach to real measurements have been made by Harries et al. (2001), Anderson et al. (2003), Brindley and Harries (2003), and Brindley and Allan (2003). The data used were from IRIS and IMG, which offered a time separation of 27 years, during which time the theoretical expectation would be that change should be detected, since CO2 increases of nearly 10% have been measured in that time. The task is harder with real, as opposed to calculated, data, however. Corrections have to be applied for the different
it 270 E
800 900 1,000 1,100 1,200 1,300 Wavenumber (cm-1)
flg. 10.14. The results of a comparison between IRIS data from 1970 and IMG data from 1997, by Harries et al. (2001). The top frame compares a 3-month average of the two over selected areas; the central frame is the difference spectrum for three areas of the planet; the bottom frame shows the calculated difference due only to the known change in greenhouse gases, with no response by the climate.
800 900 1,000 1,100 1,200 1,300 Wavenumber (cm-1)
spectral resolutions of the instruments and their footprints on the planet, as well as for the limitations in space and time of the sampling in each case. The difficulties of doing this are shown to be formidable, and lead to large random errors.
The results of Harries et al. (2201) are summarized in Fig. 10.14. They show that the largest differences between 1970 and 1997 are seen in the strong bands of CO2 and (especially) CH4, and are attributable to the increases in the concentrations of those gases that have been measured chemically on the ground. These, of course, are examples of increases in forcing; any indications of a response to these forcings are most likely to appear as an increase in radiance seen in the 1100 to 1200 cm-1 window region, but very little is apparent. At the same time, an unexplained cooling is seen in the 800 to 900 cm-1 region, which could be contamination by thin cloud, and that may also be offsetting the global-warming signal in the shorter-wavelength window.
The problem with clouds is a crucial one for this approach. Not only does it make the result very sensitive to instrument field-of-view and sampling characteristics, it also raises the possibility that secular changes in cloud characteristics, for example in the abundance of an otherwise invisible thin layer of high ice clouds, or in the general prevalence of background aerosols, might mask other changes and remain undetected if their spectral signature is not sufficiently specific. A separate means of monitoring these may need to be contemplated, while simultaneously the measurements might be made at a large distance from the Earth, where most of a hemisphere remains in the field-of-view while measurements are made over an extended period by a single instrument with a reliable long-term calibration system. Alternatively, a less cloud-dependent approach may be better, for instance the refractivity measurements that Goody et al. originally postulated as an alternative to the spectral radiance approach. No instrumentation is in place at present that could be used to test and assess this.
Overall, the message is that the spectral radiance method for monitoring climate change (as opposed to forcing) is more complicated and difficult than it may at first seem, and so far has not proven conclusive. However, an approach to the reliable and early detection of climate change is still needed urgently, and this remains one of the most promising. Much better results could be expected from an observing system that was designed specifically for the purpose, rather than trying to reconcile disparate data sets originally generated for other purposes.
Was this article helpful?
Your Alternative Fuel Solution for Saving Money, Reducing Oil Dependency, and Helping the Planet. Ethanol is an alternative to gasoline. The use of ethanol has been demonstrated to reduce greenhouse emissions slightly as compared to gasoline. Through this ebook, you are going to learn what you will need to know why choosing an alternative fuel may benefit you and your future.