Chapter 2

2.1 Introduction

The need for improvements in the comparability of water measurements in both the stratosphere [Harries, 1976] and troposphere [Elliott and Gaffen, 1991] has long been recognised. Water vapour is measured in the atmosphere from multiple platforms with a large variety of sensors. Instruments developed in more recent years, particularly satellite-based sensors, have included correlative measurement campaigns as part of their validation procedure. For those that measure tropospheric water vapour, the conventional radiosonde has often been used for validation. The measurement of upper tropospheric water vapour by radiosondes is, however, fraught with numerous problems. In the stratosphere, the most commonly used in situ water vapour measurement instruments have shown significant biases between each other [Albritton and Zander, 1985]. As a result of correlative measurement programs, special intercomparison activities, and continuing measurement and algorithm improvements, discrepancies between measurement systems have been better quantified. In some instances, revised data sets have been prepared. Extensive use is made of these revised data sets for this report. In several of the cases, the results presented here are updates of validation studies carried out at the inauguration of the instrument measurement program. Revisions have also been made in some cases to the correlative data as well as the data to be validated.

Data quality questions are addressed from two perspectives. The first is the comparability of data obtained using a wide variety of techniques and from a large number of platforms. The other is the suitability of the data for use in various scientific studies investigating processes controlling the distribution of water vapour or the role of water vapour in responding to or initiating climate change. Important data quality questions addressed in this chapter include:

What is the relationship among various measurement systems in the water vapour concentration that is derived?
Is there a standard to which all measurements can be compared?
Where there are significant differences in the measured water vapour concentration among the sensors and platforms, do we have an understanding of the reason for these differences?
Can upper tropospheric water vapour data obtained by satellites be adequately validated by existing in situ measurements using balloon-borne (radiosondes, frostpoint hygrometers), airborne (campaigns, MOZAIC), and LIDAR techniques?
Are the primary global data sets in the troposphere (TOVS) and stratosphere (HALOE) sufficiently well validated to form the basis for the analysis of water vapour in these regions?
From the longer-term (~8 years or longer) observations made in the stratosphere, is there consistency among the changes deduced from these measurements?

An extensive list of instruments is described and characterised in Chapter 1 of this report. Several data sets have been included despite relatively short measurement histories because of their usefulness in validating longer-term or more geographically extensive data sets. The approach taken here is to compare the measurements from a variety of techniques on a number of platforms and try to quantify the differences and where they exist; inconsistencies will be pointed out as well as areas of substantial agreement. Comparisons are carried out using several approaches. Extensive use is made of direct comparison of individual profiles. Where a sufficient number of direct comparisons between two instruments exist, differences are typically presented as average biases with 1s standard deviations. In a few cases, only individual profile comparisons are presented. Because of a limited number of profiles, particularly from in situ techniques, several different profile-matching criteria are used. In some cases, matches are improved by using dynamical tracers such as potential vorticity (PV) to better characterise the air mass in which the profiles were measured. Additionally, N₂O is used to group measurements with similar histories when applicable. Average profiles from a location or period of time are also compared in order to check that features such as the lower tropical stratosphere seasonal cycle are consistently represented in different data sets. Average profile comparisons are also useful when the period of observation for various platforms is not coincident. Especially in the troposphere where spatial and temporal variability is large, it is often necessary to combine in situ measurements so that they can be compared with satellite observations. Finally, an analysis of the spatial and temporal averages and the derived products 2¥CH₄+H₂O and tropical stratospheric entry value for water vapour are compared among different data sets to see how consistent these quantities are. Possible causes of long-term changes in water vapour are discussed in Chapter 3. However, estimates of changes in the stratosphere derived from a variety of moderately long data sets from as early as the 1950's are given here, and their consistency is assessed. For the longer-term data sets, the potential for changes with time is pointed out, for example, design changes of the frostpoint hygrometer or the use of multiple satellites for the TOVS series of upper troposphere water vapour measurements. These are discussed more completely in Chapter 1.

Plan of comparisons

Using the approaches described above, comparisons are made among aircraft-borne, balloon-borne, and ground-based sensors that obtain data in both the stratosphere and troposphere. Data from these sensors are also compared to data obtained with satellite instruments. An outstanding issue for the comparability of observations in the lower stratosphere has been the disagreement among various in situ measurements obtained by several balloon and aircraft platforms that are operated by United States agencies and universities. Unfortunately the discrepancies are not resolved in this report but several aspects of this disagreement are investigated and summarised in this chapter. Comparisons among European instruments using similar principles to the United States instruments are also reported with somewhat different results. Results of a laboratory comparison of the NOAA-CMDL frostpoint, NOAA-AL Lyman-a, and the Harvard Lyman-a techniques are reported but do not resolve the differences between these instruments. Three instruments flown on balloons (MIPAS, FIRS-2, and MkIV) make stratospheric measurements using infrared techniques. These are compared with in situ instruments as well as among themselves.

Most of the airborne and balloon-borne techniques have also been compared with various satellite-borne instruments. Comparisons are presented between these techniques and the HALOE, MLS, SAGE II, ATMOS and ILAS instruments. The ground-based microwave water vapour measurements (WVMS and WASPAM) provide another opportunity for comparison with the satellite systems, particularly in the middle and upper stratosphere and mesosphere.

In the upper troposphere, the most extensive data set in both time and space is the TOVS data (since 1979) obtained from the HIRS sensor on a number of operational meteorological satellites. More recently (1992-1998) the Microwave Limb Sounder (MLS) on the Upper Atmosphere Research Satellite (UARS) has also obtained data at several levels in the upper troposphere. These two data sets are compared with one another and with a newly available set of in situ measurements in the upper troposphere from observations on commercial passenger aircraft (MOZAIC). The MOZAIC data are also compared with data from several sensors flown on upper tropospheric aircraft. Although radiosondes have a number of limitations in their measurement capability of water vapour in the upper troposphere (see Chapter 1), they are widely used and will likely still play a role in the future in validation of remotely sensed tropospheric water vapour. A useful correction to the data for a particular type of sensor (Humicap A) on the widely used Vaisala RS80 radiosonde is presented. LIDAR for measuring tropospheric water vapour profiles is an emerging technique that is compared to several in situ measurements.

Comparisons are also made among different satellite systems. In the stratosphere, the water vapour measurements from the HALOE and SAGE II instruments are the longest near global data sets. Because of problems due to interference from aerosol and ozone at the wavelength from which water vapour is retrieved, the version of the SAGE II data available for this assessment has several drawbacks that are discussed more fully later in this chapter. For this reason, only limited attention is paid to assessing the data quality of the current data set. ATMOS is compared with the MAS, HALOE and MLS measurements in the stratosphere and the mesosphere. The MLS and HALOE instruments, both flying on UARS, provide a number of comparison opportunities during the approximately 1.5 year life of the MLS 183 GHz stratospheric water vapour channel. Although the ILAS sensor on the ADEOS satellite only obtained data for 9 months before the satellite failed, several comparisons were carried out with balloon-borne packages and HALOE. The ILAS sensor is used as a transfer for comparisons between several instruments when direct comparisons were not possible. The currently operating HALOE and POAM III instruments provide stratospheric comparisons during the periods when HALOE coverage extends to high latitudes. In the troposphere, the TOVS and MLS provide overlapping measurements. In the upper troposphere, the stronger water vapour signal allows SAGE II to make measurements of good enough quality that they provide some opportunity for comparison. Finally, the tropospheric MLS and TOVS data are compared. A summary of differences found between all the instruments compared is presented in section 2.6. Conclusions and recommendations are also given in the final sub-section of this chapter.