Stratospheric ozone over the United States in summer linked to observations of convection and temperature via chlorine and bromine catalysis

Photochemical Framework for Catalytic Ozone Loss in the Lower Stratosphere.

Studies of catalytic ozone loss in the lower stratosphere at high latitudes established the network of catalytic reactions linking inorganic chlorine to the rate of ozone loss in the lower stratosphere. Simultaneous in situ aircraft observations of ClO, BrO, ClOOCl, ClONO₂, HCl, OH, HO₂, NO₂, particle surface area, H₂O, and O₃ in the transition through the boundary of the Arctic vortex (38⇓⇓⇓–42) showed explicitly the loss of ozone as well as the distinct anticorrelation between the concentration of the rate-limiting radical ClO and the ozone concentration. It is the chlorine monoxide radical, ClO, in combination with the rate-limiting step ClO + ClO + M → ClOOCl + M in the catalytic cycle first introduced by Molina and Molina (34) and the catalytic cycle rate limited by ClO + BrO → Cl + Br + O₂ first introduced by McElroy et al. (35) that constitute the reaction mechanisms capable of removing ozone over the polar regions in winter at the observed rates (36, 41, 42).

A distinguishing feature of the regime within the polar jet, which defines the boundary of the Arctic vortex in winter, is that temperatures within the vortex are lower by ∼6 K to 7 K than outside the vortex. This modest suppression in temperature is adequate to trigger the heterogeneous catalytic conversion of Cl_y to Cl₂ and HOCl at H₂O mixing ratios of 4.5 parts per million by volume (ppmv) on simple, ubiquitous sulfate aerosols (8, 45⇓⇓–48) via the three reactions displayed in the upper left of Fig. 2. The Cl₂ and HOCl products of Cl_y heterogeneous catalysis on sulfate aerosols (33) photodissociate to produce Cl atoms that react with O₃ to produce ClO. Hereafter, we refer to this series of reactions as the conversion of Cl_y to ClO.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

An example of the dependence of heterogeneous catalytic conversion of inorganic chlorine (Cl_y ≈ HCl + ClONO₂) on temperature, water vapor, and sulfate loading is displayed in a manner that distinguishes rapid conversion of Cl_y to free radical form in the shaded region (with the threshold defined as 10% chlorine activation in the first diurnal period) from the unshaded region for which there is virtually no Cl_y to ClO conversion. These domains establish the photochemical framework for the analysis of convective addition of water, sulfate addition by volcanic injection or overt sulfate addition for SRM, or combinations thereof. The broad blue line dividing the perturbed and unperturbed domains corresponds to a sulfate reactive surface area of 2 μm²/cm³; the green line represents a shift in sulfate reactive surface area to 20 μm²/cm³.

Examination of conditions in the Arctic lower stratosphere coupled with extensive results from laboratory experiments and modeling (45⇓⇓–48) have set in place the temperature−water vapor−sulfate coordinate system defining the regime of rapid heterogeneous conversion of Cl_y to ClO (2). Fig. 2 displays a schematic illustrating the temperature−water vapor threshold between the domain in which conversion of inorganic chlorine to its catalytically active forms becomes significant (shaded region) and the temperature−water vapor domain that leaves inorganic chlorine bound in its reservoir species (unshaded region) (45⇓⇓–48). Probabilities (γ) associated with the heterogeneous reactions considered here are sensitive to aerosol composition (45⇓⇓–48). In particular, reactions involving HCl are governed by its uptake and solubility, which are strongly dependent on both the sulfuric acid weight percent of the aerosol and temperature. As the sulfuric acid weight percent decreases, the solubility of HCl increases. The sulfuric acid weight percent is itself a function of relative humidity. With shifts to colder temperatures and/or higher water vapor mixing ratios leading to more dilute sulfate within the aerosol, the reaction probabilities for the conversion of Cl_y to ClO increase exponentially. Therefore, wherever the specific conditions of temperature and water vapor are satisfied, the heterogeneous catalytic conversion of Cl_y to ClO can occur on the simple, ubiquitous binary aerosol, and ozone loss can result.

The cornerstone of our understanding of sulfate−halogen-induced reductions in ozone over midlatitudes of the Northern Hemisphere (NH) is built upon observed column ozone loss following the 1991 eruption of Mount Pinatubo (8, 10⇓–12). The impact of the volcanic eruption on ozone extended over a period of nearly 4 y following the eruption when column ozone concentrations over the NH decreased by a maximum of 5% in the latitude region 35°N to 60°N (10). Model analysis of the impact emphasized the central role of halogen radical catalytic loss of ozone, particularly the important role of bromine radicals in the lower stratosphere at elevated levels of sulfate aerosol loading (10). The addition of sulfate to the stratosphere by either volcanic injection or overt addition for SRM is indicated in Fig. 2 as the sulfate “shift” to the green line that serves to move the domain for rapid heterogeneous catalytic conversion of Cl_y to ClO.

The four catalytic cycles that must be taken into account in the assessment of ozone loss rates in the lower stratosphere include the most important rate-limiting steps under unperturbed as well as perturbed conditions, i.e., conditions of elevated water vapor or lower temperatures. In this analysis, unperturbed refers to background sulfate loading of 1 μm²/cm³ to 3 μm²/cm³ and a water vapor mixing ratio of 4.5 ppmv. The four dominant rate-limiting catalytic steps include (i) ClO + ClO + M → ClOOCl + M, (ii) BrO + ClO → Br + Cl + O₂, (iii) NO₂ + O → NO + O₂, and (iv) HO₂ + O₃ → OH + 2O₂. Under unperturbed conditions in the lower stratosphere between 10 km and 22 km, the catalytic loss of ozone is dominated by the HO₂ + O₃ → OH + 2O₂ rate-limiting step, as originally demonstrated by Wennberg et al. (49). The ClO + BrO cycle plays a significant role (∼15%), exceeding the ClO dimer catalytic cycle by more than an order of magnitude under unperturbed conditions. Above ∼22 km, the catalytic cycle rate limited by NO₂ + O → NO + O₂ becomes dominant. The rate-limiting catalytic species ClO, BrO, HO₂, and NO₂ thus constitute the baseline against which unperturbed conditions may be contrasted relative to perturbed cases involving temperature variability and the convective injection of water vapor.

NEXRAD Weather Radar Map of Storm-Top Height Geographic Distribution and Penetration Depth into the Stratosphere over the United States in Summer.

The NEXRAD weather radar network has markedly advanced our understanding of both the frequency and depth of tropopause-penetrating convection in the lower stratosphere over the United States in summer. Before the radar analysis methods developed by Homeyer (50) and applied by Solomon et al. (51) for mapping the 3D structure of convective penetration, elevated water vapor mixing ratios in the stratosphere were observed in situ during multiple summertime aircraft missions over the United States (1, 2, 52). These observations of both vapor-phase H₂O and the HDO isotopologue, obtained aboard NASA’s WB-57 and ER-2 aircraft, provide direct evidence of water vapor deposited by convection in the stratosphere. Maximum water vapor values observed in situ range from 8 ppmv to 18 ppmv for individual plumes typically sampled a day to a few days after convective injection. In support of the in situ observations, the NEXRAD weather radar data provide compelling statistics on the frequency, 3D structure, and high accuracy determination of the storm-top altitude of convection.

Solomon et al. (51) used radar analysis methods developed by Homeyer (50) and observations from the operational NEXRAD radar network to create a high-resolution, 3D, gridded radar reflectivity product for 2004 over the conterminous United States east of the Rocky Mountains. By combining the NEXRAD analysis with the lapse-rate tropopause height derived from the interim reanalysis of the European Centre for Medium-Range Weather Forecasts (ERA-Interim), they produced high-resolution maps of convection overshooting the tropopause level at 3-h intervals for the entire year. The ERA-Interim estimates of the tropopause altitude agree well with high vertical resolution observations from radiosondes (53). These ice-rich overshooting parcels lead to injection of water vapor into the stratosphere through mechanisms including turbulent mixing and gravity wave breaking (53, 54).

The geographic distribution of overshooting events is markedly nonuniform, with the great majority occurring east of the Rocky Mountains and west of the Mississippi River, as presented in Fig. 3A. The largest concentration of overshooting events occurs over the high plains stretching from Texas to Nebraska and Iowa. The ongoing analysis of a 10-y hourly NEXRAD dataset for May through August of 2004–2013 confirms the diurnal, annual, and geographical patterns found by Solomon et al. (51). A key contribution that the NEXRAD system provides is the ability to map the storm-top potential temperature as a function of geographic position, frequency, and month of occurrence. The multiyear analysis indicates that 38,158 storms reached at least 2 km above the tropopause over the central United States in May−August between 2004 and 2013, with about 50% of these extending above the 390 K potential temperature level. The depth and frequency of penetration has significant consequences, so we delineate here the quantitative specifics of the NEXRAD observations with high spatial resolution values of HCl that inform the altitude-dependent distribution of available inorganic chlorine.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

(A) The geographic distribution of deep stratospheric convective injection from the NEXRAD weather radar 3D mapping of storms in the summer (May–August 2004–2013) that penetrate more than 2 km above the local tropopause. (B) The vertical distribution of inferred HCl as a function of potential temperature in the stratosphere over the United States in summer for two latitude bins (blue and red points), where HCl is calculated using in situ O₃ data from the NASA SEAC⁴RS mission. In situ measurements acquired by the National Oceanic and Atmospheric Administration chemical ionization mass spectrometer (CIMS) HCl instrument during the NASA AVE mission in June 2005 are included for comparison with the inferred HCl. Data acquired between 30°N to 50°N and 80°W to 105°W are shown. Also plotted are bin-averaged measurements of HCl acquired by MLS as a function of potential temperature, calculated from simultaneous measurements of temperature at the 100- and 68-hPa pressure levels (green squares). The dashed green lines indicate the range 1 SD from the mean for each 20 K potential temperature bin. The MLS satellite data were selected to be between 30°N to 50°N and 80°W to 105°W for June–August from 2004 to 2016. (C) The total number of storms that penetrate more than 2 km into the stratosphere over the central United States in May–August (black) and June–August (red) from 2004 to 2013 as a function of potential temperature. The calculated HCl mixing ratio is superimposed on the NEXRAD observations of frequency and penetration height.

The vertical coordinate system most appropriate for the quantitative coupling of the NEXRAD observations to that of inorganic chlorine is potential temperature (the temperature of an air parcel compressed adiabatically to 1,000 hPa) because, in the absence of diabatic processes, air parcels in the stratosphere are transported along surfaces of constant potential temperature, such that long-lived trace species exhibit consistent correlations with one another. In particular, this is a characteristic shared by long-lived tracers that are either produced or removed by increasing UV radiation as a function of increasing altitude in the stratosphere, e.g., HCl vs. O₃ and Cl_y vs. N₂O. Data from multiple in situ measurement campaigns, as well as satellite retrievals, have been used to quantify the relationships among these species (42, 55, 56). In Fig. 3B, high-resolution vertical profiles of HCl mixing ratio (blue and red circles) were inferred from in situ measurements of O₃ using the linear relationship between Aura Microwave Limb Sounder (MLS) measurements of HCl and O₃ at 100 hPa and 68 hPa, such that HCl ≈ 7.0 × 10⁻⁴ × O₃ (units of parts per trillion by volume). MLS version 4.2 data from 2004 to 2016, subselected to be between 30°N to 50°N and 80°W to 105°W for June through August, were used to derive this conversion factor. The in situ O₃ data used to calculate HCl throughout the lower stratosphere over the United States in summer are from the NASA Studies of Emissions and Atmospheric Composition, Clouds, and Climate Coupling by Regional Surveys (SEAC⁴RS) mission, which took place over the United States in the summer of 2013. Also shown in Fig. 3B are in situ measurements of HCl from the NASA Aura Validation Experiment (AVE) campaign over the United States in June 2005 and bin-averaged satellite measurements of HCl from MLS over the United States in summer as a function of potential temperature. The dotted lines define 1 SD from the mean of the MLS HCl data. The in situ and satellite measurements of HCl support the more complete vertical profiles of inferred HCl that are used subsequently to compare with NEXRAD data. HCl comprises most of the inorganic chlorine in the lower stratosphere, and, accordingly, Fig. 3B demonstrates the rapid rise in available inorganic chlorine with increasing potential temperature (altitude).

Fig. 3C displays the NEXRAD observations of convective injection as a function of potential temperature, showing the total number of storms observed in the summer (May–August) between 2004 and 2013, and the number of those storms that occurred in June, July, and August. Of particular importance is the fact that more than 50% of the observed 38,158 storms between 2004 and 2013 that reached 2 km above the tropopause penetrated above the 390 K potential temperature level. This height corresponds to the altitude of rapid increase in available inorganic chlorine. The vertical distribution of HCl is shown with the NEXRAD observations in the same coordinate system of potential temperature, which provides for the direct comparison of (i) convective penetration height and frequency with (ii) HCl, a lower limit on available inorganic chlorine, Cl_y.

Subsequent to the high in situ water vapor observations reported by Anderson et al. (2), Schwartz et al. (57) analyzed satellite observations of water vapor from MLS and confirmed that the lower stratosphere over the United States in summer is, indeed, unusually wet, with measured values reaching as high as 18 ppmv. The MLS satellite data indicate that the highest stratospheric water vapor mixing ratios at the highest latitudes globally are over the central and eastern United States in summer. These results are particularly noteworthy, because the true magnitude and number of water vapor enhancements over the United States in summer is likely significantly greater than reported by the MLS satellite instrument. This discrepancy occurs because elevated H₂O from convective injection is localized in space and typically layered vertically, such that the spatial resolution of MLS (e.g., 3 km vertical × 200 km horizontal at 100 hPa for H₂O) may often exceed the dimensions of the perturbation or randomly transect it, leading to an underestimate of the mixing ratio. For example, a 2-km-deep layer of elevated water of 50-km horizontal extent, even with optimal alignment of the convected geometry along the north−south viewing track of MLS, will only fill 1/6 of the MLS sample volume, with the other 5/6 filled with background levels of H₂O; this results in a substantial underreporting of the actual H₂O mixing ratio present within the convected domain. Nevertheless, Schwartz et al. (57) established the crucial fact that the stratosphere over the central and eastern United States is unique globally with respect to significantly elevated water vapor.

Another key piece of evidence relating the observations of enhanced water vapor to their convective source comes from the atmospheric chemistry experiment-Fourier transform spectrometer (ACE-FTS) satellite observations of the concentration of the heavy water isotopologue HDO globally in the lower stratosphere (58) as well as in situ observations of HDO within regions of convective injection from NASA aircraft missions (1). The HDO to H₂O ratio is expressed in the usual isotopic formulation of δD that is reported in per mil units and defined as δD(‰) = (R_obs/R_SMOW − 1) × 1,000, where R = [HDO]/[H₂O], “SMOW” refers to standard mean ocean water, and R_SMOW = 3.12 × 10⁻⁴. Observations of δD are important because convective injection followed by sublimation is characterized by less negative values of δD in water vapor compared with air that has passed through the tropical tropopause, which has δD values of around −650‰. This number corresponds to a 65% depletion of HDO relative to SMOW. In situ aircraft measurements of convective outflow show δD values of −200‰ (1, 59). The ACE-FTS global observations of δD at 16.5 km from Randel et al. (58) show δD values of approximately −490‰ over North America in summer but virtually no enhancement of δD over the global mean at 16.5 km altitude in any other geographic domain, including the Asian monsoon region. These measurements provide direct evidence for the convective source of water vapor in the stratosphere over the United States in summer as well as for the unique occurrence of deep stratosphere-penetrating convection in the global context.

Observations of Temperatures in the Lower Stratosphere over the United States in Summer.

We use high spatial resolution, high-accuracy in situ temperature measurements acquired in the specific altitude, latitude, longitude, and season appropriate for calculations of localized ozone loss in the lower stratosphere over the central United States in summer. These temperature data were acquired aboard the NASA ER-2 high-altitude aircraft on flights in the stratosphere during August and September 2013 over the central United States during the NASA SEAC⁴RS mission. For the present analysis, which focuses specifically on the central United States east of the Rocky Mountains, we select temperature data in the latitude range from 30°N to 40°N and in longitude from 105°W to 80°W. The observed temperatures from the aircraft in situ measurements are displayed as the gray dots in Fig. 4. For comparison, Fig. 4 also shows the temperature profile from a 13-y record (2002–2014) of gridded, monthly average radio occultation (RO) observations from 30°N to 40°N for July and August, referred to hereafter as T_RO. Although not equal in spatial resolution to the in situ observations, the RO data provide an independent measure of the average temperature for this region and season, with an accuracy (absolute) of 0.1 K and spatial resolution of ∼0.5 km in the vertical (60, 61).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

High spatial resolution in situ temperatures (gray dots) obtained during the NASA SEAC⁴RS mission over the United States in summer 2013. The distribution in temperature is, in part, due to gravity wave-driven temperature fluctuations. Bin-average profiles for these data are plotted in the red circles, and the smoothed mean profile, T_ave, is represented by the solid red line. Similarly, the profile of minimum temperatures, T_min, is shown in the cyan circles and line. T_mid (blue line) is defined as the temperature profile midway between T_ave and T_min. The July/August mean of RO temperatures for the same region (black dashed line) demonstrates agreement between the in situ and RO data sets of 2 K or less. Finally, the “standard atmosphere” temperature profile for July that has been used in the AER 2D model, averaged over a region extending from 33°N to 52°N is shown in the solid black line.

In the following modeling analysis, three different temperature profiles (Fig. 4) are used to evaluate the response of the rate-limiting steps for the catalytic loss of stratospheric ozone to temperature in both the presence and absence of convectively injected water vapor. The three temperature profiles are T_std, the AER 2D model “standard atmosphere” temperature (10, 23⇓⇓–26) profile, for the month of July; T_ave, a profile of the observed average temperatures with altitude, drawn from the SEAC⁴RS in situ aircraft measurements; and T_mid, a temperature profile that lies at the midpoint between T_ave and the envelope of observed minimum temperatures from the in situ aircraft measurements. Inspection of the temperature profiles presented in Fig. 4 reveals the similarity between T_ave and T_RO. The difference, ∼1 K to 2 K in the altitude region from 14 km to 20 km, results from the fact that we are using monthly and spatially averaged (5° × 5° × ∼1 km) RO data. However, the agreement between T_ave from the in situ aircraft observations and T_RO from the RO observations serves as an important cross-calibration between these two physically independent absolute temperature measurements.

As Fig. 4 makes clear, there is a significant range in observed temperatures in the lower stratosphere over the central United States in summer. Spatial and structural variability in combination with gravity waves that continuously traverse the lower stratosphere over the Great Plains in summer contribute to the observed temperature variability. These gravity waves are induced by strong convective events embedded in mesoscale convective systems, squall lines and tornadic storm structures, as well as the presence of the Rocky Mountains (62⇓–64). Also potentially contributing to the temperatures in the domain of elevated water vapor from convective injection is the radiative cooling to space at a rate of ∼0.05 K/d per 1 ppmv of additional water vapor (65, 66).

Given the remarkable nonlinearity of the heterogeneous catalytic processes that control Cl_y to ClO conversion, the temperature observations are critically important for determining the rate of catalytic loss of ozone; this is particularly true for excursions to low temperatures, given the extremely rapid heterogeneous catalytic conversion of Cl_y to ClO. The AER 2D model employs an empirically derived distribution about the selected temperature at each altitude, with a SD of ∼3 K. For example, with the peak temperature centered at 202 K, ∼7% of the data fall below 197 K (or above 207 K), and <1% are below 195 K (or above 209 K). This temperature distribution has always been intrinsic to the AER 2D model, and it is an important capability of the model to represent a range of temperatures under conditions appropriate to the lower stratosphere in summer over the Great Plains. The high spatial resolution, in situ aircraft observations verify the importance of the temperature distribution function of the AER 2D model.

Dynamics Defining Lower Stratospheric Flow Patterns over the United States in Summer.

The NAM creates a situation during July and August that is particularly conducive to the hydration of the lower stratosphere by extremely deep convection. Not only does it steer water vapor from the Gulf of Mexico across Texas and into the western Plains States in the lower atmosphere, it also generates an anticyclone in the upper troposphere and lower stratosphere that causes stratospheric air parcels to dwell markedly longer over North America than if they were advected by a purely zonal flow. This anticyclonic circulation is not stable, however, leading to regular ventilation. Thus, the mean residence time of air over the United States is on the order of a week, with some parcels residing significantly longer (67). The residence time sets the range of timescales for evaluating ozone loss over this region from one or more of these injections. Evidence that parcels are induced to circulate in a stratospheric anticyclone over North America during summer with convectively injected water vapor retention within the gyre is evident in tracer−tracer (7) and satellite data (68⇓–70).