The hydroxyl radical (OH) is the major oxidizing agent in the atmosphere. Chemical reactions with OH initialize the removal of carbon monoxide (CO), methane (CH₄), and volatile organic compounds (VOC). The OH required for these oxidation processes is produced everywhere in the sunlit atmosphere by UV photo dissociation of ozone (O₃) and subsequent reaction of the emerging O(¹D) with water vapor:

O₃ + hn(<340nm)®O(¹D)+ O₂ (R1)

O(¹D) + H₂O®OH + OH (R2)

Reactions with OH initialize chain reactions, which eventually lead to a removal of the oxidized species from the atmosphere. Since Levy [1971] first postulated this basic concept of tropospheric photochemistry, numerous estimates of the globally averaged OH concentration have been obtained from the budget of species that are exclusively removed by reaction with OH and for which the source strengths are well known (e.g., ¹⁴CO, CH₃CCl₃). The most recent of these studies yielded concentrations close to 1x10⁶ OH/cm³ [Prinn et al., 1995; Krol et al., 1998; Spivakovsky et al., 2000].

Why should we measure OH locally?

The global mean OH controls the atmospheric lifetimes of trace gases like CH₄ and CO, which are relatively well mixed on a global scale. It can also be viewed as a measure for the self-cleansing capacity of the atmosphere. However, global OH is not a representative quantity on a regional or local scale, for example, polluted or forested areas. In fact, due to the short lifetime resulting from its high chemical reactivity, OH is highly variable in space and time.

The natural variability of OH and its response to atmospheric parameters can be studied only by local measurements at high time resolution. Variations of OH mainly reflect the influence of chemical processes that control the transformation of other important gases and eventually initiate the production of tropospheric ozone or particles. Accordingly, local OH measurements constitute a valuable tool for the direct observation of chemical processes in the atmosphere and for testing our understanding of photochemistry models for different meteorological situations and trace gas burdens.

The challenge of measuring OH

Since the importance of atmospheric OH was recognized in the early 1970s, many intensive efforts (primarily in the U.S. and Germany) were undertaken to measure its concentration on a local scale [see reviews by Crosley, 1994, 1995]. Many of the early OH instruments suffered from insufficient sensitivities and interferences from other gases [Crosley and Hoell, 1986; Beck et al., 1987]. This is not surprising in view of the high demands of measuring atmospheric OH:

1. Its high reactivity in the gas phase and on surfaces places severe practical constraints on instruments that use sampling inlets.
2. The extremely small concentration of typically less than 10⁷ OH/cm³ requires a very high detection sensitivity.
3. High specificity is needed to avoid interferences with trace gases that are much more abundant than OH.
4. An OH concentration standard is not easily available for instruments that require calibration.

In the stratosphere, early measurements were successful [e.g., Anderson, 1976; Heaps and McGee, 1985; Stimple and Anderson, 1988] but lacked simultaneous measurements of other trace gases required to test photochemical theories. Comprehensive sets of simultaneous measurements of OH and trace gases were first obtained in the planetary boundary layer. Although the temporal and spatial resolution of the first field measurements of OH were relatively low, they allowed first tests of tropospheric photochemistry models [Perner et al., 1987; Poppe et al., 1994].

It took roughly twenty years from the first attempts to measure tropospheric OH until, in the first half of the 1990s, experimental techniques achieved the desired reliability, sensitivity, and temporal-spatial resolution to enable systematic measurements of OH in different parts of the world, on the ground, on aircraft, and on ships. While many of the experimental problems of OH detection now seem to be solved, even the most recent OH instruments are technically complex and their deployment in the field is still not trivial.

Current measurement methods

Different techniques have been used for the successful detection of atmospheric OH. These can be grouped into two categories: direct methods, which use optical spectroscopy for OH detection, and indirect methods, which convert OH chemically into another species that can be detected with high sensitivity and specificity. An overview of most of the recent techniques can be found in Journal of the Atmospheric Sciences [52, 3297-3441, 1995].

Absorption spectroscopy

Differential optical absorption spectroscopy (DOAS) uses the UV absorption lines of OH at 308 nm from transitions between rotational levels of the electronic ground-state (X²P, v''=0) and the first electronically excited state (A²S⁺, v'=0). Usually a laser beam is transmitted through a 1-10 km path in the open atmosphere and the absorption spectrum is detected. The application of Lambert-Beer's law directly yields the OH concentration from the measured absorption. The great advantage of this method is that the required absorption cross-section can be calculated from well known spectroscopic properties of the OH molecule to within better than 10%. Thus, no additional calibration is required [Mount, 1992; Dorn et al., 1995a, Hausmann et al., 1997].

Different configurations of the absorption light-path have been used, either as a double-pass over a long distance (10 km) [Mount, 1992], or as multi-pass where the light beam was folded more than 100 times between the mirrors of an open White cell (total light path length 1-4 km) [Armerding et al., 1994; Dorn et al., 1995b]. The latter configuration enables local OH measurements by absorption spectroscopy and is fairly mobile. Recently a folded-path system was used to measure OH onboard a ship [Brauers et al., submitted to Journal of Geophysical Research].

Detection limits of 8 • 10⁵ OH/cm³ (1s) were demonstrated in field experiments using a folded light path of 1800 m total length and a time resolution of 200 s [Brandenburger et al., 1998].

Laser-induced fluorescence (LIF)

All LIF instruments used for tropospheric OH detection apply the Fluorescence Assay with Gas Expansion (FAGE) concept which was pioneered by Hard, O'Brien and coworkers [Hard et al., 1984; Chan et al., 1990; Hard et al., 1995]. Ambient air is expanded though a nozzle into a low-pressure (1-5 mbar) fluorescence cell. OH radicals present in the resulting gas beam are excited using a narrow-band pulsed laser and are directly detected via their fluorescence. As in the case of DOAS, LIF takes advantage of the discrete UV line absorption spectrum (A²S⁺ <– X²P) for the selective excitation of OH [Stevens et al., 1994; Heal et al., 1995; Holland et al., 1995; Hard et al., 1995].

In contrast to DOAS, the LIF technique needs to be calibrated. The accuracy obtained with current techniques is about 20%. Even after years of development it is still a difficult task to generate known concentrations of OH at ambient conditions to calibrate the instruments in the field [Holland et al., 1998]. Typical detection limits for 1 minute measurement time are about (4-7) • 10⁵ OH/cm³ (2s) allowing a good time resolution at a sufficient sensitivity to investigate the natural variability of OH.

The LIF technique can be extended to the in situ measurement of HO₂ radicals [Hard et al., 1995]. Their detection is realized indirectly by adding an increment of NO at the inlet of the fluorescence chamber to convert a certain amount of HO₂ into OH.

HO₂ + NO®OH + NO₂

The efficiency of this titration reaction must be regularly controlled by calibration measurements.

Current LIF instruments are compact and robust enough to be used on aircraft and a number of measurements of OH radicals in the free troposphere were conducted in recent years [Brune, this issue; Wang, this issue].

LIF instruments used for measurements in the stratosphere use excitation at 282 nm into the first excited vibrational level of the upper electronic state (A²S⁺, (v'=1)) of OH [Wennberg et al., 1995]. The use of this technique for tropospheric OH is impeded by laser induced production of OH radicals in the detection chamber at 282 nm [Hard et al., 1995].

Chemical conversion techniques

In the past few years three indirect techniques have been applied for OH detection:

Radiocarbon technique

Ambient air is drawn into a UV-transparent flow reactor, to which a small quantity of ¹⁴CO is added. A fraction of ¹⁴CO is oxidized to ¹⁴CO₂ which is then collected, purified, and analyzed for ¹⁴C by gas-proportional counting. The amount of ¹⁴CO₂ is related to the OH concentration through the CO+OH reaction constant and the residence time in the reactor [Felton et al., 1990]. While straightforward in principle, this technique is experimentally challenging and needs long counting times for low OH concentrations.

Liquid phase scrubbing

A potentially new indirect OH measurement method has emerged recently. The advantages are its experimental simplicity, portability, and low expense. The atmospheric OH signal is related to a compound produced by liquid phase reaction of scavenged OH with salicylic acid [Chen and Mopper, 2000]. The reaction product is separated and quantified by HPLC with fluorescence detection. Although still under development, early field studies seem to indicate the feasibility of this approach for OH measurements in clean air. Taking a sampling period of 45-90 minutes, the expected detection limit is estimated to be (3-6) • 10⁵ OH/cm³.

Selected Ion Chemical Ionization Mass Spectrometry (SICIMS)

Ion-assisted OH measurements are performed by pulling ambient air into a flow tube reactor. The central part of the airflow is sampled and OH present is almost quantitatively converted to H₂³⁴SO₄ by addition of an excess of isotopically labeled ³⁴SO₂ [Eisele and Tanner, 1991; Tanner et al., 1997]. Ionization is achieved by reaction of H₂³⁴SO₄ with in situ generated NO₃^–(HNO₃ ) ions further downstream of the reactor. The product ion H³⁴SO₄^– is specifically detected in a quadrupole mass spectrometer with high sensitivity allowing discrimination from the H³²SO₄^– signal originating from naturally occurring H₂³²SO₄.

In the past few years this technique was further developed and successfully used during several aircraft campaigns (ACE 1, PEM-Tropics A/B).

SICIMS has a very low detection limit of 2 • 10⁵ OH/cm³ (2s) with a fast time response of 30 s. As in the case of LIF, calibration to obtain the absolute concentrations is a challenge. The total systematic error of the ground-based instrument is about 30% [Tanner et al., 1997] whereas the accuracy of the aircraft instrument is 40-60% demonstrating the experimental difficulties to calibrate the instrument for different flight altitudes [Mauldin et al., 1997, 1998, 1999].

Intercomparison of OH instruments

Increasing confidence in the validity of OH field measurements has been gained from a number of intercomparison exercises. The results of the two most comprehensive campaigns, THOPE [Tropospheric OH Photochemistry Experiment, 1993], performed in the US (1993), and POPCORN [Photo-Oxidant Formation by Plant Emitted Compounds and OH Radicals in North-Eastern Germany, 1994], carried out in Germany [1994] are summarized in special issues of J. Geophys. Res. [102, 6169-6510, 1997] and J. Atmos. Chem. [31, 1-246, 1998], respectively.

During THOPE the OH concentrations observed with SICIMS and DOAS agreed within the instrumental uncertainties when clear days and low-NO_X conditions were selected for comparison [Mount et al., 1997]. On average, the SICIMS data were 20% lower than the absorption measurements, which can be explained by different air masses being sampled by the SICIMS point measurements and over the DOAS light path extending over 10 km length.

	Figure 1: Atmospheric absorption spectra measured using DOAS as a function of time of day (UT). Solid lines are reference absorption spectra of OH radicals fitted to the measurements (Adapted from Dorn et al., 1996).

During POPCORN DOAS LIF and folded-path DOAS instruments were compared. The absorption instrument provided the spectroscopic evidence (Figure 1) that, in fact, hydroxyl radicals were measured [Brandenburger et al., 1998]. Figure 2 shows a diurnal profile of OH measured by LIF and folded-path absorption. The agreement is, in most cases, excellent [Brauers et al., 1996; Hofzumahaus et al., 1998].

	Figure 2: Diurnal variation of OH measured using LIF (o) and DOAS () during POPCORN (Adapted from Hofzumahaus et al., 1998).

Verification of the fundamental properties of atmospheric OH by direct field measurements

To test our understanding of the chemical system controlling OH, it is necessary to have measurements of OH concentrations along with measurements of those parameters that are known to control OH. The new generation of instruments has yielded large OH data sets comprising measurements in different environments. They can be used to test fundamental properties of atmospheric chemistry.

Figure 2 illustrates the dependence of the OH concentration on the solar UV flux. Starting from nighttime concentrations below the detection limit (5x10⁵ cm^–3) OH reaches its maximum (typically in the range (2-10)x10⁶ cm^–3 in summer) at local noon when photolysis peaks.

	Figure 3: Correlation plot of all LIF OH data versus the photolysis frequency of ozone, J(O1D). (Adapted from Holland et al., 1998).

It is clearly demonstrated in Figure 3 that the photolysis of ozone, yielding O(¹D), indeed constitutes the dominant parameter controlling the diurnal variation of OH (see contribution by W. Brune, this issue). Here all OH measurements made during POPCORN are correlated with concurrently measured ozone photolysis frequencies, J(O¹D) [Holland et al., 1998; Ehhalt, 1999; Ehhalt and Rohrer, 2000]. The resulting correlation coefficient r² = 0.83 indicates that more than 80% of the total variation in OH can be explained by the variation of J(O¹D).

	Figure 4: Dependence of the measured OH concentration on NO2 during the POPCORN field campaign. To make this behavior visible, the OH data were first normalized with respect to J(O1D) and then plotted versus equal log(NO2)-intervals of 0.1. Full curve corresponds to the model-calculated dependence. [Adapted from Ehhalt, 1999].

The non-linear dependence of OH on NO_X, which represents the next most important control factor, can also be extracted from recent field measurement data (Figure 4).

How well do local measurements of OH compare with models?

After experimental confirmation of the two major functional dependencies of OH we have to ask for the quantitative representation of atmospheric OH by models.

Summarizing the results of OH field measurement campaigns carried out in the planetary boundary layer we can state:

1. In clean air masses, good agreement within the combined uncertainties of models and measurements is obtained. On average, the differences observed are 10%-30%, which is very small. However, there are individual cases showing differences up to 50% and more [Mauldin et al., 1998; Ehhalt, 1999; Frost et al., 1999; Brauers et al., data from a recent ship campaign on the Atlantic Ocean, submitted to JGR]. However, all cases have in common that the models tend to calculate higher concentrations.

	Figure 5: Comparison of observed and calculated OH concentrations versus NOX during the 1993 Idaho Hill experiment (THOPE). The different model calculations account for different amounts of unmeasured biogenic hydrocarbons [Adapted from McKeen et al., 1997].

2. In more polluted air, particularly regions of strong biogenic emissions, the discrepancy is larger, indicating far less understanding of these environments (Figure 5). However, this does not necessarily imply that the models are not capable of describing the more complex chemistry in these environments. Incomplete chemical characterization of the air masses and hence of the chemical processes in the model is most likely the cause of a systematic underestimation of OH loss reactions by models [McKeen et al., 1997].

Moreover, in the continental upper troposphere (above 8 km), a region which is considered clean, OH measurements using LIF exceeded modeled data by factors of 1.5 to 6. The analysis of the observations has suggested that this experimental finding again can be attributed to incomplete measurements of OH source molecules and the resulting lack of these processes in the model applied [Brune et al., 1998; Jaeglé et al., 1998].

Summary

Field measurements of OH have confirmed our understanding of elementary aspects of atmospheric chemistry.

Calculations of the global distribution of OH are consistent within a few percent with the current budgets of well-known tracers [Spivakovski et al., 2000]. On a local basis, however, the chemical system is currently understood to a much lesser extent.

During the past decade many successful field campaigns all over the atmosphere have demonstrated the availability of several different OH measurement techniques with sufficient sensitivity and time resolution. However, in view of many open questions, investigation of atmospheric OH photochemistry is necessary and will remain a further challenge for experimentalists and for modelers.

	Figure 6: Altitude profiles of measured (open circles) and modeled OH for 10 May 1996 during SUCCESS. Measurements and models are averaged into 0.5 km altitude bins. Models with (dash-dot line) and without (dashed line) acetone are compared. (Adapted from Brune et al., 1998)

References

1. Anderson, J.G., The absolute concentration of OH (X²P) in the earth's stratosphere, Geophys. Res. Lett. 3, 165-168, 1976.
2. Armerding, W., M. Spiekermann, and F. J. Comes, OH mulipass absorption: Absolute and in situ method for local monitoring of tropospheric hydroxyl radicals, J. Geophys. Res., 99, 1225-1239, 1994.
3. Beck, S.M. et al., Operational overview of NASA GTE/CITE 1. Airborne instrument intercomparisons: carbon monoxide, nitric oxide, and hydroxyl instrumentation, J. Geophys. Res., 92, 1977-1985, 1987.
4. Brandenburger, U., T. Brauers, H.-P. Dorn, M. Hausmann, and D. H. Ehhalt, In-situ measurement of tropospheric hydroxyl radicals by folded long-path laser absorption during the field campaign POPCORN in 1994, J. Atmos. Chem., 31, 181-204, 1998.
5. Brauers, T. et al., Intercomparison of tropospheric OH radical measurements by multiple folded long-path laser absorption and laser induced fluorescence, Geophys. Res. Lett., 23, 2545-2548, 1996.
6. Brune, W.H. et al., Airborne in-situ OH and HO₂ observations in the cloud-free troposphere and lower stratosphere during SUCCESS, Geophys. Res. Lett., 25, 1701-1704, 1998.
7. Chan, C.Y., et al., Third-generation FAGE instrument for tropospheric hydroxyl radical measurement, J. Geophys. Res., 95, 18569-18576, 1990
8. Chen, X., K. Mopper, Determination of tropospheric hydroxyl radical by liquidphase scrubbing and HPLC: Perliminary results, J. Atmos. Chem., 36, 81-105, 2000
9. Crosley, D.R. and J.M. Hoell, Future directions for H_xO_y detection, NASA Conference Publication 2448, 1986.
10. Crosley, D.R., Local measurement of tropospheric HO_x, NASA Conference Publication 3245 (1994).
11. Crosley, D.R., Laser fluorescence detection of atmospheric hydroxyl radicals. In Progress and Problems in Atmospheric Chemistry, Advanced Series in Physical Chemistry, Vol. 3, , World Scientific Publishing Co., Singapore, (ed.) J.R. Barker, pp 256-317, 1995.
12. Dorn, H.-P., R. Neuroth, and A. Hofzumahaus, Investigation of OH absorption cross sections of rotational transitions in the A²S⁺, v'=0 <-- X²P, v''=0 band under atmospheric conditions. Implications for tropospheric long-path absorption measurements, J. Geophys. Res., 100, 7397-7409, 1995a.
13. Dorn, H.-P., U. Brandenburger, T. Brauers, and M. Hausmann, A new in situ long-path absorption instrument for the measurement of tropospheric OH radicals, J. Atmos. Sci., 52, 3373-3380 1995b.
14. Dorn, H. P., U. Brandenburger, T. Brauers, M. Hausmann, and D. H. Ehhalt, In-situ detection of tropospheric OH radicals by folded long- path laser absorption. Results from the POPCORN field campaign in August 1994., Geophys. Res. Lett., 23, 2537-2540, 1996.
15. Ehhalt, D.H., Photooxidation of trace gases in the troposphere, Phys. Chem. Chem. Phys., 1, 5401-5408, 1999.
16. Ehhalt, D.H. and F. Rohrer, Dependence of the OH concentration on solar UV, J. Geophys. Res., 105, 3565-3571, 2000.
17. Eisele, F.L., and D. J. Tanner, Ion-assisted tropospheric OH measurements, J. Geophys. Res., 96, 9295, 1991.
18. Felton, C.C., J.C. Sheppard, and M. J. Campbell, The radiochemical hydroxyl radical measurement method, Environ. Sci. Techn., 24, 1841-1847, 1990.
19. Frost, G.J. et al., Photochemical modelling of OH levels during the First Aerosol Characterization Experiment (ACE1), J. Geophys. Res., 104, 16041-16052, 1999.
20. Hard, T.M., R. J. O'Brien, C. Y. Chan, and A. A. Mehrabzadeh, Tropospheric free radical determination by FAGE, Environ. Sci. Technol., 18, 768-777, 1984.
21. Hard, T.M., L. A. George, and R. J. O'Brien, FAGE Determination of Tropospheric HO and HO₂, J. Atmos. Sciences, 52, 3354-3372, 1995.
22. Hausmann, M., U. Brandenburger, T. Brauers, and H.-P. Dorn, Detection of tropospheric OH radicals by long-path differential-optical-absorption spectroscopy: Experimental setup, accuracy, and precision, J. Geophys. Res., 102, 16011-16022, 1997.
23. Heal, M.R., D. E. Heard, M. J. Pilling and B. J. Whitaker, On the development and validation of FAGE for local measurement of tropospheric OH and HO₂, J. Atmos. Sci., 52, 3428-3441, 1995.
24. Heaps, W.S., T.J. McGee, Progress in stratospheric hydroxyl measurement by balloon-borne LIDAR, J. Geophys. Res., 90, 7913-7921 1985.
25. Hofzumahaus, A. et al., Intercomparison of tropospheric OH measurements by different laser techniques during the POPCORN campaign 1994, J. Atmos. Chem., 31, 227-246, 1998.
26. Holland, F., M. Hessling, and A. Hofzumahaus, In-situ measurement of tropospheric OH radicals by Laser-Induced Fluorescence. A description of the KFA instrument, J. Atmos. Sci., 52, 3393-3401, 1995.
27. Holland, F., et al., Highly time resolved measurements of OH during POPCORN using laser-induced fluorescence spectroscopy, J. Atmos. Chem., 31, 205-225, 1998
28. Jaegle, L. et al., Sources of HO_x and production of ozone in the upper troposhere over the United States, Geophys. Res. Lett., 25, 1709-1712, 1998.
29. Krol, M., P. Jan van Leeuwen, and J. Lelieveld, Global OH trend inferred from methylchloroform measurements, J. Geophys. Res., 103, 10,697-10,711, 1998.
30. Levy II, H, Normal atmosphere: Large radical and formaldehyde concentrations predicted, Science 173, 141-143 1971.
31. Mauldin, R.L. et al., New insights on OH: Measurements around and in clouds, Geophys. Res. Lett., 24, 3033-3036, 1997.
32. Mauldin, R.L. et al., OH measurements during the First Aerosol Characterization Experiment (ACE 1): Observations and model comparisons, J. Geophys. Res., 103, 16713-16729, 1998.
33. Mauldin, R.L. et al., Measurements of OH during PEM-Tropics A, J. Geophys. Res., 104, 5817-5827, 1999.
34. McKeen et al., Photochemical modeling of hydroxyl and its relationship to other species during the Tropospheric OH Photochemistry Experiment, J. Geophys. Res., 102, 6467-6493, 1997.
35. Mount, G.H., The measurement of tropospheric OH by long-path absorption. 1. Instrumentation, J. Geophys. Res., 97, 2427-2444, 1992.
36. Poppe, D. et al., Comparison of measured OH concentrations with model calculations, J. Geophys. Res., 99, 16633-16642, 1994.
37. Perner, D., et al., Measurement of Tropospheric OH concentrations: A comparison of field data with model predictions, J. Atmos. Chem., 5, 185-216, 1987.
38. Prinn, R.G. et al., Atmospheric trends and liftetime of CH₃CCl₃ and global OH concentrations, Science, 269, 187-192, 1995.
39. C.M Spivakowsky et al., Three dimensional climatological distribution of tropospheric OH: Update and evaluation, J. Geophys. Res., 105, 8931-8980, 2000.
40. Stevens, P.S., J. H. Mather, and W. H. Brune, Measurement of tropospheric OH and HO₂ by laser-induced fluorescence at low pressure, J. Geophys. Res., 99, 3543-3557, 1994.
41. Stimpfle, R.M. and J.G. Anderson, In-situ detection of OH in the lower stratosphere with a balloon borne high repetition rate laser system, Geophys. Res. Lett., 15, 1503-1506, 1988.
42. Tanner, D.J., A. Jefferson, and F. L. Eisele, Selected ion chemical ionization mass spectrometric measurement of OH, J. Geophys. Res., 102, 6415-6425, 1997.
43. Wennberg, P.O. et al., In situ measurements of OH and HO₂ in the upper troposphere and stratosphere, J. Atmos. Sci., 52, 3413-3420, 1995.