Ozonesondes provide the vertical profile of ozone with high vertical resolution.
Ozonesondes are lightweight and compact balloon-borne instruments measuring the ozone concentration from the surface through the mid-stratosphere (about 10 hPa or 30 km). A stable miniature pump (see Fig. 1) drives ambient air in cells of inert material, filled with differing concentrations of potassium iodide (KI) solutions. The ozone measurement is based on a chemical reaction between O3 molecules in the air and those solutions. The chemical transformation generates an electric current proportional to the amount of ozone, explaining the name electrochemical concentration cell (ECC).
The ozonesonde is flown on a rubber weather balloon in tandem with a regular radiosonde (see Fig. 2) for data transmission. During operation the ozone partial pressure (from the ozonesonde) and the pressure-temperature-humidity (PTU) and wind data (all from the radiosonde) are recorded telemetrically by a ground station during an ascent and descent cycle over ~2.5 hours. With a 20-30s response time of the ozone cells and an ascent rate of about 6 m/s, the effective vertical resolution of the ozone signal lies nowadays around 150 m. Ozonesondes are virtually all-weather, i.e., unaffected by clouds and precipitation, in contrast to most spectroscopic techniques, and they are relatively inexpensive and easy to operate.
Regular measurements with ozonesondes started in the second half of the 1960s at a few sites: in 1966 at Resolute Bay (Canada), in 1967 at Hohenpeissenberg (Germany), in 1968 at Payerne (Switzerland), in 1969 at Uccle (Belgium), and in 1970 at Wallops Island (USA). These ozone sounding stations provide the longest time series of vertical ozone distribution. Currently, there are around 60 active ozonesonde stations around the world. Their data are submitted regularly to international data archives as the World Ozone and Ultraviolet Radiation Data Centre (WOUDC), the Network for the Detection of Atmospheric Composition Change (NDACC), and the Southern Hemisphere ADditional OZonesondes (SHADOZ).
A measurement with an ozonesonde, called an ozone sounding, provides not only the ozone concentration as a function of the altitude (or pressure), but also the temperature, the relative humidity, wind direction and wind speed. In the following Fig. 5 all the information of such a sounding is shown.
Because each sonde is a new instrument, it is prepared prior to launch. Thus, a standard set of procedures and an accepted common reference are required to ensure the comparability of data taken with different instruments, operating and data-processing protocols. These protocols are regularly evaluated by the panel for the Assessment of Standard Operating Procedures for Ozonesondes (ASOPOS), in which the RMI is represented.
Check out this video to learn more about how an ozonesonde is prepared and launched at Uccle:
Despite the growing importance (global coverage!) of vertical ozone profile retrievals with satellite instruments, ozonesondes are still the only technique able to measure the ozone concentrations from the surface all the way up to the middle stratosphere with very high (absolute) accuracy and vertical resolution. The long-term ozone sounding stations like Uccle also provide the longest time series of vertical ozone distribution. Therefore, they have many application areas in which they are crucial:
- the evaluation of the temporal variability (and trends) of the ozone concentrations from the surface up to the middle stratosphere, see e.g. Long-term ozone trends at Uccle.
- ozonesondes are the backbone for the satellite ozone retrieval validation: satellite algorithms are based on ozonesonde climatologies and in turn satellites are validated by the sondes. Sonde profiles are being used to detect drifts in limb-measuring ozone satellites. Our group is active with the Atmospheric Composition SAF.
- to study ozone processes: the high vertical resolution of ozonesondes is important here, as atmospheric transport occurs in thin, quasi-horizontal layers. For example, since 1992 the annual Match campaigns have followed ozone-depleted layers in the mid- and high-latitude stratosphere as they move in and around the Arctic vortex.
Starting in 1969, and with three launches a week, the Uccle ozonesonde dataset is one of longest and densest of the world. Moreover, as the only major change was the switch from Brewer-Mast to Electrochemical Concentration Cell (ECC) ozonesonde types in 1997, the Uccle time series is very homogenous. However, efforts have been continuously taken to guarantee the homogeneity between ascent and descent profiles, under changing environmental conditions (e.g. SO2), between the different ozonesonde types, under changing preparation procedures. Important factors affecting the data quality and homogeneity of the ozonesonde time series are
- the frequency response of the ozone sensor: De Muer and Malcorps (1984) concluded that the measured frequency response can be represented by a first-order process and a diffusion process in series and developed a method for deconvolution of the ozone profiles through a process of Fast Fourier transform.
- interference with SO2, causing a quantitative reduction in the ozone detected by the cells.
- different types of sensors: De Backer et al. (1998) developed a correction method (now called PRESTO, for PRESsure and Temperature dependent Total Ozone normalization) to correct for the transition from Brewer-Mast to ECC ozonesondes in 1997.
- (changes in) the pump efficiency decrease with decreasing pressure
- box temperature
- background current
- altitude measurements by the radiosondes
Two different homogenized time series of ozonesondes are available. The entire time series have been homogenized using the operational PRESsure and Temperature dependent total Ozone normalization (PRESTO) correction method. The full report about the homogenization procedure is available here. The ECC time series have also been homogenized according to the Ozonesonde Data Quality Assessment (O3S-DQA) principles, and a comparison between both corrected datasets is provided in Van Malderen et al. (2016).