Introduction

Water makes the Earth unique. Life, the climate and the weather all exist as they do because gaseous, liquid and solid forms of water can co-exist on the planet. The oceans, the ice masses, the clouds and the humidity of the atmosphere all play major roles in climate and weather [Peixoto and Oort, 1996]. The water vapour distribution in the upper troposphere (UT) and lower stratosphere (LS) is of central importance in several ways: it plays a major role in the balance of planetary radiation; it influences and responds to atmospheric motions; and it plays a key role in many aspects of UT/LS chemistry. Effects on the radiation balance are especially important because the water vapour molecule is strongly polar in shape giving it a strongly absorbing infrared spectrum. Consequently, it is the single most important "greenhouse gas" in the atmosphere, molecule-for-molecule. Harries [1997] pointed out that, because the effects of water vapour on the Earth radiative balance are so large, small errors in spectroscopic parameters and in radiative-dynamical models used to model the energy balance, can produce potentially large uncertainties in the prediction of climatic change. He presented evidence from sensitivity studies showing that the humidity concentration, particularly in the upper troposphere may need to be known with an accuracy in the range of 3-10% in order to avoid uncertainties in calculated radiative forcing that are of the order of the effect due to doubling CO2 concentrations. Forster and Shine [1999], extending earlier work of Rind and Lonergan [1995], calculate that if the increase in the lower stratospheric H2O mixing ratio reported over Boulder, Colorado, from 1979 to present, by Oltmans and Hofmann [1995] is occurring globally, the contribution to surface warming would be 40% of that from the CO2 increase over the same time period. Forster and Shine also emphasised that an increase of water vapour causes a cooling of the lower stratosphere that is comparable to the contribution due to ozone changes. These facts emphasise the urgent need to understand and assess the state of knowledge of atmospheric water vapour. In view of the potential effects on climate change, it is especially important to assess and attempt to understand long-term changes and decadal scale trends in water vapour.

Despite its abundance in the atmosphere and its importance for the climate system, many questions regarding H2O are presently unresolved. For example, the important question of feedback between water vapour mixing ratio in the upper troposphere and surface temperature is still unanswered. While increased temperature leads to increased moisture and further warming because of the "greenhouse" effect, it is unclear whether the warming produces more water due to further evaporation (positive feedback) or if the attendant increased upwelling causes a drying (negative feedback). Results of calculations by Manabe and Wetherald [1967] from a radiative-convective model with constant humidity suggested that the exponential increase of absolute humidity due to the sea surface temperature rise would exert a strong positive feedback. Other analyses from complex general circulation models are generally consistent with Manabe’s and Weatherald’s conclusions showing similarly large positive feedback. Ellsaesser [1984] on the other hand, argued that an increase in the strength of convection in the tropics would cause an increase in the Hadley Cell circulation. An increase in the strength of the circulation will lead to drying or negative feedback, rather than a moistening of the upper troposphere [Ellsaesser, 1984, Lindzen, 1990, and Sun and Lindzen, 1993]. In short, as far as the greenhouse effect and climatic change are concerned, it is still not clear whether thermodynamics or dynamics controls tropospheric water vapour.

Atmospheric water vapour is abundant in the atmosphere and possesses absorption features spread over a broad range of the electromagnetic spectrum. Therefore, it should be easy to measure. However, in practice, water vapour measurements have proved to be difficult. Water sticks to surfaces, thereby providing challenges for in situ techniques. Sharp vertical and horizontal gradients present difficulties for remote sensing techniques. A lack of understanding of the fundamental physics behind the observed spectrum complicates the analysis of remote sensing measurements. The radiative effects of clouds are yet another complicating factor in making water vapour measurements. Consequently, our understanding of the distribution of upper tropospheric and lower stratospheric water vapour is not as thorough as it should be.

The importance of water vapour and the fact that an assessment of current knowledge has not been undertaken before, led the Scientific Steering Group of the international research project on Stratospheric Processes and their Role in Climate (SPARC), a project of the World Climate Research Programme (WCRP), to initiate and sponsor the study documented in this report. Four key points were noted:

  1. The global, regional and seasonal distribution of water vapour in the UT/LS region of the atmosphere is not known with sufficient accuracy to validate climate models.
  2. It is important to examine what can be said about long-term change of the water vapour concentration in the UT/LS by analysing existing in situ and satellite data sets from a common vantage point.
  3. The various reports of the Intergovernmental Panel on Climate Change (IPCC) [IPCC, 1991, 1993, 1995] have all stressed the role of water vapour as a powerful greenhouse gas. However, the question of the magnitude and perhaps even the sign of the water vapour feedback still represents probably the single largest uncertainty in the quest for accurate prediction of global and regional warming caused by an increase of CO2 concentration.
  4. It is expected that new data requirements on both water vapour and related parameters (e.g. temperature) will come out of an assessment process, as has been the case for the SPARC Ozone Assessment [SPARC, 1998].

As an introduction to the report, we present a brief synopsis of the evolution of knowledge on the content and distribution water vapour. Key processes that influence these aspects of water vapour are only briefly discussed here, but more detail is provided in chapter 3.

Water Vapour in the Stratosphere

The extreme dryness of the stratosphere was first discovered by Brewer and his colleagues using data collected in the British Meteorological Research Flight (MRF) program with a manually operated frost-point hygrometer taking measurements over Southern England [Brewer, 1949; Bannon et al., 1952]. These flights started in 1943 and continued sporadically till the early 1980’s. During the first flights in the 1940s, they found tropopause frost point temperatures of -58ºC (~55 ppmv near 250 hPa), and mixing ratios ~2km above the tropopause as low as 1.6 ppmv. The MRF flight series continued through the early 1960's, then resumed again in the 1970's and early 1980's [Murgatroyd et al., 1955; Helliwell et al., 1957; Cluley and Oliver, 1978; Foot, 1984]. Oliver and Cluley [1978] identified a ?0.6°C temperature error in the early MRF frost point measurements. After the application of an appropriate temperature correction, the mean "representative stratospheric humidity (r.s.h.)" in the 1950's was ~3 ppmv while that in the 1970's was ~5 ppmv.

The first balloon-borne in situ and remote observations of stratospheric water vapour were reported in the literature of the 1960's. Results were mixed with some measurements indicating a considerable increase of water vapour mixing ratio with altitude above the tropopause and others showing little or no change. The controversy was reviewed by Gutnik [1961]. It was later understood that out-gassing of water vapour from the balloon skin during ascent, when most of the measurements were made, caused water vapour contamination of the in situ sensors or the atmosphere in the vicinity of the rising or floating balloon.

Mastenbrook, from the U.S. Naval Research Laboratory, adapted the frost-point hygrometer for use on stratospheric balloons. He undertook a series of soundings from 1964 to 1979 over Washington, D.C. and conducted a two-year measurement series from 1964 to 1965 in tropical air over Trinidad, West Indies, [Mastenbrook, 1968, 1971, 1974, Mastenbrook and Daniels, 1980]. Mastenbrook's data showed that the aridity of the stratosphere extended vertically up to the maximum reachable levels of his balloons, about 28 km, thereby settling the question of whether the stratosphere was "wet" or "dry". Whether there was a small increase of the mixing ratio with altitude over and above the values found in the lower stratosphere was not so much a concern at that time. The question was whether or not there was a substantial increase of the water vapour mixing ratio with altitude. For example, Sissenwine et al. [1968] had reported water vapour mixing ratio increases at 32 km about 6 times higher than values in the lower stratosphere. The 1976 U.S. Standard Atmosphere [U.S. Standard Atmosphere, 1976] reflects the uncertainty that prevailed at the time on this issue and lists two alternative vertical profiles.

The U.S. Naval Research Laboratory series of balloon soundings lasted until 1979 when the measurement program was handed over to NOAA in Boulder, Colorado. It has continued since 1980 [Oltmans and Hofman, 1995, Oltmans et al., 2000]. Just prior to the change from Washington to Boulder a switch to more modern electronics was also made which resulted in reduced scatter and improved time-height resolution of the instrument. Results from the Boulder soundings confirmed the principal earlier Mastenbrook results of a long-term strong seasonal change of water vapour mixing ratio in the lower stratosphere with maxima in summer. In addition, a positive trend of water vapour mixing ratio in the stratosphere of about 1% per year, which had been discovered by Mastenbrook over Washington D.C., was also found over Boulder.

The Lyman-a hygrometer, a new in situ instrument for the measurement of upper tropospheric and stratospheric water vapour, was introduced in 1978 [Kley and Stone, 1978, Bertaux and Dellanoy, 1978]. Using this technique Kley et al. [1979] showed that there is an increase of water vapour mixing ratio with altitude in the stratosphere. A tropical sounding over Brazil revealed the fact that an absolute minimum of water vapour mixing ratio, termed the hygropause, is located some 2-3 km above the tropical tropopause [Kley et al., 1979]. The minimum mixing ratio of 2.6 ppmv was later corrected down to 2.3 ppmv [Kley et al., 1983].

Global satellite measurements of water vapour commenced in 1978. The results from the Limb Infrared Monitor of the Stratosphere (LIMS) experiment on the Nimbus 7 satellite [Russell et al., 1984; Remsberg et al., 1984] extended the in situ measurements from 84°N to 64°S. The increase of the mixing ratio with altitude was confirmed and the measurements were extended to 55 km. LIMS also showed the hygropause to be a general feature of the tropical stratosphere. Work to combine the observations from different satellite instruments has proved valuable. For example the work by Jones et al. [1986] to merge water vapour from the LIMS experiment with CH4 measurements provided by the Stratospheric And Mesospheric Sounder (SAMS) experiment on the same satellite resulted in the first global view of total hydrogen variability (2 x CH4 + H2O). The Atmospheric Trace Molecule Spectroscopy (ATMOS) series of missions on several flights of the Space Shuttle [Farmer and Raper, 1986, Gunson et al., 1990] confirmed the LIMS and SAMS findings.

Long-term satellite observations of stratospheric water vapour concentration began in 1986 with the Stratospheric Aerosol and Gas Experiment II (SAGE II) [McCormick et al., 1993, Chiou et al., 1997]. Due to a strong sensitivity of the SAGE II water vapour retrieval algorithm to aerosol interference and the Pinatubo volcanic eruption that put large amounts of aerosols into the stratosphere, the continuous record was interrupted in 1991. The Halogen Occultation Experiment (HALOE) was launched in 1991 and has provided nearly continuous measurements of water vapour over the pressure range from 0.01 down to 200 hPa [Russell et al., 1993; Harries et al., 1996]. Several other satellite, shuttle borne, and balloon instruments were operated during the 1980s and 1990s. These will be discussed in detail in subsequent chapters of this report.

Water Vapour in the Upper Troposphere

Water in the troposphere occurs in all the possible aggregate states: liquid, gaseous and solid. The strong dependence of water vapour partial pressure on temperature, the fact that atmospheric temperatures from ground to tropopause span a range of some 60°C, and that transport in the troposphere is rapid, combine to cause large variability in tropospheric water vapour mixing ratios that are quantitatively difficult to understand. Moreover, very few methods exist to measure upper tropospheric water vapour mixing ratio (specific humidity) and relative humidity with high accuracy. Water vapour soundings by the operational national radiosonde networks do not produce high quality data in the upper troposphere. Research balloons and research aircraft have yielded a fair amount of information. However, compared to the large variability of upper tropospheric humidity on a variety of spatial and temporal scales, the database obtained by in situ techniques is inadequate. Satellite techniques, which are less influenced by spatial constraints, have begun to contribute to our knowledge.

Tropospheric water vapour concentrations are strongly coupled to dynamics and to temperature. Although the water vapour partial pressure of ascending air is not known to exceed saturation with respect to the liquid phase, water vapour concentrations can still vary by more than 4 orders of magnitude between the ground and the tropical tropopause. In the upper troposphere where temperatures are persistently below -40°C, it is convenient to relate the water vapour concentration to relative humidity with respect to ice (RHi). Based on tropical in situ observations, it has been found that upper tropospheric RHi is close to 100% over convective regions and RHi is often less than 10% in regions with subsidence [Kley et al., 1997]. However, formation of ice hydrometeors depends on the availability of ice condensation nuclei and on microphysics. Supersaturation (RHi >100%) has been observed, probably caused by the virtual absence of aerosols that can act as nucleation centres. Jensen et al., [1999] showed that supersaturation occurs below the tropical tropopause. Over mid-latitudes of the Northern Hemisphere large persistent supersaturations of up to 150% are a common occurrence [Gierens et al., 1999]. Generally speaking, because of the upper limit on the water vapour partial pressure, set by the temperature dependent phase equilibrium between vapour and condensate, the maximum tropospheric water vapour mixing ratios are strongly coupled to atmospheric temperature. Hence, the water vapour mixing ratios tend to undulate with seasonally varying temperature, modified by meteorological processes which, together, can produce changes of relative humidity from a few to one hundred percent.

Relatively few papers appear in the literature describing the upper tropospheric water vapour large and small scale distribution and seasonal, annual and longer time scale changes. Moreover, when we consider that the radiosonde network has not, and does not, provide accurate information on upper tropospheric water vapour, we are left with few data sets, especially those of a regional or global nature. Kelly et al., [1991] found lower mixing ratios in the upper troposphere at middle to high latitudes of the southern hemisphere than those at the equivalent winter season of the northern hemisphere. In situ measurements from high quality research balloons or aircraft have provided sporadic data on upper tropospheric water vapour. They have shown that a strong seasonal cycle of upper tropospheric water vapour mixing ratio exists at mid-latitudes with a summer maximum and a winter minimum. This information comes from measurements over southern England, Washington and Boulder [Oliver and Cluley, 1978, Mastenbrook and Daniels, 1980, Oltmans and Hofmann, 1995]. Dethof et al., [1999] showed that the surface topography has an impact on the dynamics of the UT/LS region and on the transport of water vapour into the stratosphere. Duhnke [1998], in the line of earlier work by Briggs and Roach [1963], investigated upper tropospheric water vapour mixing ratio and relative humidity in the coordinate system of the north Polar jet stream and found a downward transport of dry stratospheric air through the tropopause on the cyclonic side and an upward transport of high water vapour mixing ratio into the stratosphere on the anticyclonic side of the jet stream.

Bates et al., [1996], used information from the Tiros Operational Vertical Sounder (TOVS) instruments to construct a quasi-global picture of upper tropospheric relative humidity. Unfortunately, the TOVS humidities are weighted over a fairly deep vertical altitude range and they are not well validated. The Microwave Limb Sounder (MLS) on the Upper Atmosphere Research Satellite (UARS) has produced results on the large scale distribution and time variability of upper tropospheric water vapour [Stone et al., 2000] and can be used to determine the vertical distribution of upper tropospheric relative humidity with a resolution of approximately 3 km. Sandor et al., [1998] presented the seasonal distribution of tropical to mid-latitude upper tropospheric relative humidity, derived from the MLS product. Newell et al., [1996b, 1997] analysed MLS-derived upper tropospheric humidity fields and found moist plumes in tropical air over regions with rising motion and dryness in sinking regions. Based on their analysis, they concluded that a substantial fraction of upper tropospheric humidity to the west of South America originates over the South American continent. They also found a positive relationship between the zonal mean tropical moisture and Eastern Pacific sea surface temperature. Zhu et al. [2000] found from further MLS UT/LS analysis that the moisture in the tropical upper troposphere is mainly increased by intensified small-scale local convection.

Chemical processes

The primary process for the increase of the water vapour mixing ratio with altitude in the stratosphere is photochemical CH4 oxidation [Brasseur and Solomon, 1984; LeTexier et al., 1988]. Oxidation of methane leads predominantly to production of H2O and to a small extent H2. The proportion of each molecule produced is a function of altitude and latitude and is dictated mainly by the age of air relative to the chemical lifetimes of the gases involved. Generally, one CH4 molecule will lead to production of approximately two H2O molecules. Entry level air from the troposphere in the 1990's carried ~1.7 ppmv of CH4 into the stratosphere [Dlugokencky et al., 1998].

Transport from the troposphere to the stratosphere

Progress in understanding the processes that control atmospheric water vapour and exchange between the troposphere and stratosphere has been slow. A major advance was provided by Brewer [1949] who used observations of low H2O mixing ratio in the stratosphere over Southern England to propose that influx to the stratosphere occurs by slow, large-scale upwelling through the tropical tropopause where the water vapour partial pressure is in equilibrium with the ice phase. At a tropical tropopause temperature of ?80°C at 100 hPa the corresponding mixing ratio is 5.4 ppmv. Egress from the stratosphere occurs in conjunction with large-scale downward motion through the extratropical tropopause where the downward directed advective flow of air counteracts the effects of eddy diffusion which, in the absence of the advective component of the flow, would seed the stratosphere with much higher water vapour mixing ratios as they prevail in equilibrium at the higher extratropical tropopause temperatures. Newell and Gould-Stewart [1981] extended this picture by arguing that the tropical tropopause is generally too warm to explain the extremely dry stratosphere. Based on this, they further narrowed the region where air slowly ascends into the stratosphere to those regions and times in which the tropical tropopause is much colder than average. These include the western tropical Pacific, northern Australia, and Indonesia during the November to March period and over the Bay of Bengal and India during the monsoon. This space and time limited region has become known as the "stratospheric fountain". Atticks and Robinson [1983] and Frederick and Douglass [1983] on the other hand showed results that were compatible with the idea that water enters the stratosphere year-round at all longitudes except 0° to 30°W. Later, results from satellite and in situ instrumentation (HALOE, MLS, SAGE II and Lyman-a ) [e.g. Mote et al., 1996; Weinstock et al., 1995] were published supporting the idea that air enters the stratosphere not just at limited times, but throughout the entire year. Satellite results showed that the tropical lower stratosphere is continually being modulated on an annual cycle with water vapour entering from below at values governed by the tropopause temperature. This phenomenon has been termed the "atmospheric tape recorder" [Mote et al., 1996].

The observations of a minimum water vapour mixing ratio above the tropical tropopause over Panama [Kley et al., 1979] and those that support a minimum in H2O in the Western tropical Pacific very near the tropopause [Kelly et al., 1993; Rosenlof et al., 1997; and Jackson et al., 1998] suggest that the process of exchange is more complicated than the "fountain" theory implies. While other plausible ideas have been put forward such as overshooting deep convection [Danielsen, 1982], no scientific consensus exists on what the dominant processes are that transport air across the tropical tropopause to the stratospheric "overworld", the region consisting of isentropes (surfaces of constant potential temperature) that lie above the tropopause everywhere.

Mechanisms of troposphere-stratosphere exchange in the tropics

The current view on transport of trace gases between the low stratosphere (Q < ~380 K) and the "overworld", Q > 380 K, is based on the top-down control principle of Haynes et al., [1991] and expressed in a review article by Holton et al., [1995]. According to Holton et al., wave-induced forces in the extratropical overworld drive a global scale extratropical fluid-dynamical suction pump which draws air upward and poleward from the tropical lower stratosphere and pushes it poleward and downward into the extratropical troposphere. An important point in Holton’s review is the implication that the tropical troposphere-stratosphere exchange rate is not determined by details of near-tropopause phenomena, such as penetrating cumulus convection or small scale mixing. That is, non-local forces that drive the general circulation also control the global troposphere - to - stratosphere rate of transport.

Goals and Purposes of this Report

The overall goal of this study is to provide a current assessment of the state of knowledge of the water vapour distribution, variability and long-term changes. While there is some review of the literature included in the report it is minimal and only used where needed to aid the assessment process. The same is true of historical records, references and limited measurement campaigns. Results from short campaigns are applied where needed to assist in validating the data used for the assessment. The report is organised into three chapters. Chapter 1 describes the physical parameters measured, algorithms used, estimated errors, potential drifts, data record length, measurement frequency and spatial resolution for each instrument. This is an essential aid in assessment of data quality and interpretation of intercomparisons. Chapter 2 provides definitive statements regarding data quality for each measurement system used in the study, it shows how results intercompare with one another, discusses what the limitations are in vertical resolution, altitude and latitude and provides an overall description of what can and cannot be believed in the distributions and long-term changes. Chapter 3 provides a description of what is known about H2O in the context of the findings of chapters 1 and 2. It describes geographic distributions as a function of altitude, latitude and longitude in the Northern and Southern Hemispheres, seasonal cycles and the effects of the Quasi-Biennial Oscillation (QBO), El Nino and the Southern Oscillation (ENSO) and Tropical Intraseasonal Oscillation (TIO) on the water vapour distribution. It also discusses stratospheric entry level mixing ratios and long-term changes. Summaries, conclusions and recommendations are provided as needed in each chapter.