GROMOS-C (GRound-based Ozone MOnitoring System for Campaigns) is an ozone MWR measuring the ozone emission line at 110.836 GHz at Ny-Ålesund, Svalbard (79∘ N, 12∘ E), which is described in detail in . It was built by the Institute of Applied Physics (IAP) at the University of Bern. GROMOS-C is very compact, meaning it can be transported and operated at remote field sites under extreme climate conditions. It further can switch the frequency of the local oscillator and measure the 115 GHz carbon monoxide (CO) emission line. The system noise temperature of the instrument is about 1080 K. The optics setup of GROMOS-C has two rotating mirrors such that observations in all four cardinal directions are possible. Therefore, GROMOS-C observes in the four cardinal directions (north, east, south, and west) under an elevation angle of 22∘ with a sampling time of 4 s. Ozone VMR profiles are retrieved from the ozone spectra with a temporal averaging of 2 h leveraging Atmospheric Radiative Transfer Simulator version-2 ARTS2; and Qpack2 software according to the optimal estimation algorithm . An a priori ozone profile is required for optimal estimation and is taken from an MLS climatology of the years 2004–2013. The sensitive altitude range for this instrument extends from 23 to 70 km. The vertical resolution of ozone profiles is 10–12 km in the stratosphere and increases up to 20 km in the mesosphere as estimated from the width of the averaging kernels. The averaging kernels (AVKs) of GROMOS-C together with its measurement response, errors, and ozone profiles are shown in Appendix A (Fig. ). In the lower stratosphere, the errors are below 0.3 ppmv and reach above the stratopause values up to 0.4 ppmv. More details about the uncertainty and the AVKs can be found in .

MIAWARA-C (MIddle Atmospheric WAter vapor RAdiometer for Campaigns) is a ground-based MWR measuring the pressure-broadened rotational emission line of water vapor at the frequency of 22 GHz. It was also built by the University of Bern and is located at Ny-Ålesund, Svalbard (79∘ N, 12∘ E). The MIAWARA-C front end is an uncooled heterodyne receiver with a system noise temperature of 150 K. The antenna is followed by a dual-polarization receiver. The incident radiation is split into vertical and horizontal polarization by an orthomode transducer (OMT) located immediately after the feedhorn. The two polarized signals are processed in the two identical receiver chains and separately analyzed in a fast Fourier transform (FFT) spectrometer model Acqiris AC240. The spectrometer has a 400 MHz bandwidth and a spectral resolution of 30.5 kHz. The standard measurement cycle of MIAWARA-C is to measure sky east, reference east, sky west, and reference west for about 15 s each. Every 15 min the ambient load is measured for about 2 s, and the sky at 60∘ elevation is measured for about 15 s. A tipping curve is performed to determine the sky temperature at 60∘ elevation. The difference spectra in the east and west directions and the two polarizations are then calibrated separately with the hot and cold measurements close in time. Similar to GROMOS-C, MIAWARA-C retrieval is also performed with ARTS2 and QPACK software according to the optimal estimation algorithm . An a priori measurement is taken from an MLS climatology of the years 2004–2008. From the measured spectra we retrieve water vapor profiles that cover an altitude range extending from 37 to 75 km with a vertical resolution of 12–19 km and with a time resolution of 2–4 h depending on the opacity of the troposphere. For MIAWARA-C, retrievals are performed with a constant time resolution and with a constant noise of 0.014 K. The AVKs of MIAWARA-C, together with its measurement response, errors, and water vapor profiles, are shown in Appendix A (Fig. ). In the upper stratosphere, the errors are 0.5 ppmv and increase from 0.5 to 1.5 ppmv in the mesosphere. A detailed design of the instrument and description of the retrieval algorithm can be found in and .

The Microwave Limb Sounder (MLS) is one of the payloads onboard NASA's Earth Observing System (EOS)-Aura satellite, which was launched in 2004 . The satellite is in a sun-synchronous orbital altitude of 705 km with a period of 1.7 h and 98∘ inclination. MLS scans the atmospheric limb in the direction of orbital motion, which gives almost pole-to-pole coverage (82∘ S to 82∘ N), leading to retrieved profiles at the same latitude every orbit, with a spacing of a 1.5∘ great circle angle along the suborbital track. Ozone is retrieved from the band of 240 GHz, and water vapor is retrieved from the 183 GHz line. Temperature is derived from radiances measured from the 118 and 240 GHz channels with a vertical resolution between 3 and 6 km . The estimated temperature single-profile precision is 0.5–1.2 K from 100 to 0.001 hPa. MLS provides ozone profiles (version 5) from 12 to 80 km altitude with a vertical resolution of 2.5–6 km and water vapor profiles (version 5) from 10 to 90 km with a vertical resolution of 3.5 km from 316 to 4.64 hPa and 15 km above 0.1 hPa . The estimated ozone single-profile precision varies from 0.2 to 0.4 ppmv from the middle stratosphere to the lower mesosphere. For water vapor, the estimated precision is 0.2–0.3 ppmv in most of the stratosphere and increases to 0.7–0.8 ppmv in the middle mesosphere. It passes at Ny-Ålesund twice a day at around 04:00 and 10:00 UTC. Profiles for comparison are extracted if the location is within ±1.2∘ latitude and ±6∘ longitude of either Ny-Ålesund or the defined virtual conjugate latitude station.

The Modern-Era Retrospective Analysis for Research and Applications, version 2 MERRA-2 is the latest global atmospheric reanalysis produced by the NASA Global Modeling and Assimilation Office (GMAO) from 1980 to the present. MERRA-2 assimilates observation types not available to its predecessor, MERRA, and includes updates to the Goddard Earth Observing System (GEOS) model and analysis scheme so as to provide a viable ongoing climate analysis beyond MERRA's terminus. MERRA-2 provides a regularly gridded, homogeneous record of the global atmosphere and incorporates additional aspects of the climate system including several improvements to the trace gas constituents, land surface representation, and cryospheric processes.

In MERRA-2, methods of analysis, model uncertainties, and observations cause uncertainties . provides a comprehensive assessment of the MERRA-2 ozone product that relatively clearly shows the vertical distribution of ozone and water vapor in the stratosphere and has the best agreement with stratospheric ozone observations compared to other reanalysis products. identified that ozone in MERRA-2 data were expected to have higher uncertainties in regions of high variability, such as winter high latitudes. The uncertainties are related to the chemistry model ozone bias , which is altitude dependent in MERRA-2. Similar behavior was also found in other GEOS chemistry models . MERRA-2 stratospheric water vapor is also biased compared to independent observations from Aura-MLS and the Atmospheric Chemistry Experiment-Fourier Transform Spectrometer (ACE-FTS). In this study, we use the ozone and water vapor with 72 model levels from the surface up to 0.01 hPa and a horizontal resolution of 0.5∘ × 0.625∘. The time resolution is 6 h. MERRA-2 products are accessible online through the NASA Goddard Earth Sciences Data Information Services Center (GES DISC).

Figure  shows the climatological ozone distribution as a function of pressure and time deduced from GROMOS-C, MERRA-2, and Aura-MLS in both hemispheres. Many features of ozone MWR measurement climatologies are broadly consistent with model and satellite data. Some exceptional maximum values larger than 8 ppmv in MERRA-2 data above 0.1 hPa are described in Sect. 3.

Overall, the ozone profile reveals a characteristic seasonal dependence at polar latitudes. In particular, the altitude of the maximum ozone VMR, as well as its temporal variability, exhibits a seasonality. The peak ozone VMR (approximately 6.5 ppmv) appears in the NH in the late spring, whereas autumn shows the lowest values throughout the course of the year. The maximum observed and reanalysis ozone VMR (approximately 5.5 ppmv) in the SH is 1.0 ppmv smaller than the NH maximum and occurs later in the hemispheric spring season. The primary driver of the hemispheric maximum in ozone VMR is related to circulation processes throughout the stratosphere, including those associated with the BDC, transporting ozone-rich air toward the poles in the winter–spring hemisphere. To be precise, this circulation moves the ozone-rich air from the tropical photochemical source region to high latitudes after the polar vortex broke down and essentially enables the intrusion of ozone-rich air from the mid-latitudes into the polar region and the replacement of the ozone-depleted air masses. Another essential feature is that GROMOS-C and Aura-MLS capture the tertiary ozone VMR maximum at the northern polar latitude in the early winter and in late spring at the southern polar latitude. However, the tertiary ozone maximum in GROMOS-C occurs at altitudes close to or even above the limit of 0.8 for the measurement response. MERRA-2 performs poorly when trying to capture the tertiary ozone VMR maximum in both polar latitudes. Due to the complexity of altered dynamics in the polar regions introducing extra uncertainties into numerical models and data assimilation systems, ozone VMRs exhibit dramatic variability (red shading in Fig. b, e) in the mesosphere.

From September to November in the Southern Hemisphere, both MERRA-2 and MLS effectively capture the presence of the ozone hole in the lower stratosphere. However, the climatology of ozone in the NH does not reflect a corresponding signature. The annual cycle of ozone at 46 hPa in both hemispheres further reveals this significant feature, as shown in Fig. a. MLS exhibits a greater magnitude of ozone depletion compared to MERRA-2. In Fig. 8b, the ozone levels at 3 hPa in the NH follow an annual cycle, with a peak occurring in March and a minimum in October. The SH ozone VMR maintains a constant level of 5 ppmv throughout the summer while experiencing two minimum values in February and October. Furthermore, the ozone VMR in MERRA-2 consistently remains lower than that in MLS except during autumn in the SH. In the mesosphere, there is relatively good agreement between GROMOS-C and MLS, while MERRA-2 exhibits higher variability (Fig. c). Both hemispheres have dramatically different seasonal variations and distribution in ozone due to differences in the stratospheric dynamics of the two hemispheres.

Figure  shows the climatology of water vapor vertical and temporal distribution from MIAWARA-C, MERRA-2, and Aura-MLS over both stations (79∘ S and 79∘ N) compiled from measurements collected between 2015 and 2021. The annual variation of water vapor with a maximum during hemispheric summer and a minimum during hemispheric winter is clearly visible. During the hemispheric winter, the middle-atmospheric water vapor maximum is shifted down to about 10 hPa, whereas it rises up to the lower mesosphere during the hemispheric summer. The characteristics of water vapor in both hemispheres depend strongly on the mesospheric pole-to-pole circulation, which is an upwelling with moist air transporting upwards for the hemispheric summer and a corresponding downward motion with dry air into the stratosphere during the winter . In addition, due to the relatively long photochemical lifetime of water vapor, more water vapor produced by methane oxidation accumulates in summer.

The annual cycle of monthly mean water vapor VMR at three separate pressure levels (3, 0.3, and 0.02 hPa) in both hemispheres can be seen in Fig. . Water vapor VMR appears at a maximum in October in both hemispheres in the stratosphere (3 hPa) due to the oxidation of methane. Due to photodissociation caused by solar Lyman-alpha radiation acting as a sink and the amount of water vapor being in equilibrium between different photochemical processes and vertical transport, water vapor is more smooth, with nearly constant mixing ratios from winter to spring in the NH. In the mesosphere (0.3 and 0.02 hPa), water vapor gradient variations can be found during hemispheric spring and summer. The gradient at 0.02 hPa is steeper than at 0.3 hPa from April to July in the NH (from November to January in the SH). Furthermore, the positive gradient is weaker, but the time of increase lasts longer and shows an extreme negative gradient in the hemispheric autumn. In both hemispheres, the seasonal behavior of water vapor is almost symmetric at 0.02 hPa and has a slight asymmetry at 0.3 hPa. At 0.02 hPa the maximum value of the water vapor mixing ratio persists for 1 month, and at 0.3 hPa the decrease in water vapor is not visible until September in the NH (March in the SH), but it then appears with a steep gradient.

In the two ground-based radiometer measurements from GROMOS-C and MIAWARA-C, the aforementioned seasonal behavior in their ozone and water vapor distributions display very similar patterns to Aura-MLS and MERRA-2. However, there are distinct differences between the datasets. To quantitatively assess the consistency of the ozone and water vapor climatologies from GROMOS-C and MIAWARA-C, the relative differences (RDs) of ozone and water vapor VMR between the ground-based MWRs and the two other data sets are calculated using the following expression: 1RD=(φ)radiometer-(φ)dataset(φ)dataset⋅100%, where φ represents either ozone or water vapor VMR. Before evaluating the differences in climatology between MWRs and MERRA-2 and Aura-MLS, each MERRA-2 and Aura-MLS profile is convolved with the averaging kernels of MWRs. The convolution is performed according to the following expression: 2xconv=xa-Axa-x, where xconv is the convolved profile, xa is the a priori profile, A is the averaging kernel matrix, and x is the high-resolution profile of MERRA-2 and Aura-MLS.

Figure  shows the relative differences of ozone and water vapor climatologies from GROMOS-C and MIAWARA-C with respect to average convolved MERRA-2 and Aura-MLS. In Fig. a, the largest negative RD is larger than 50 % above 0.2 hPa in winter and spring because of the high bias of model ozone chemistry in the mesosphere in MERRA-2 . GROMOS-C shows relatively good agreement with MERRA-2 in the lower and middle stratosphere (50–5 hPa), with RDs smaller than ±5 %, but includes a low RD with magnitudes greater than 10 % in the upper stratosphere in autumn. GROMOS-C and Aura-MLS agree well, with RDs mainly within ±7 % throughout the middle and upper stratosphere, and exceed positive RD 10 % in the lower mesosphere (1–0.1 hPa) in March and October (Fig. b).

The RD between MIAWARA-C and MERRA-2 is larger than 50 % throughout the late autumn to spring months above 0.2 hPa in Fig. c. Simultaneously, MERRA-2 underestimates water vapor VMR at all altitudes due to the lack of assimilated observation to constrain the water vapor reanalyses in the polar mesosphere and also in part due to the methane oxidation parameterization being disabled in the GEOS Composition Forecast model . The MIAWARA-C agrees with MERRA-2 within 7 % in the stratosphere and lower mesosphere (about 100–0.3 hPa). MIAWARA-C and Aura-MLS show relatively good agreement, with RDs within 5 % throughout the stratosphere and lower mesosphere (approximately 100–0.02 hPa) in Fig. d. Furthermore, MIAWARA-C exhibits a negative RD within 8 % above the lower mesosphere (around 0.1–0.01 hPa). Additionally, we found a wet bias of 5 %–7 % between 100 and 1 hPa for MIAWARA-C measurements in each season. The wet bias becomes noticeable between July and November, primarily attributed to variations in instrument performance and, to a certain extent, the dynamic effects causing an elevation in water vapor levels within the stratosphere (as seen in Fig. a).

Note that the RD between ground-based MWRs and Aura-MLS is in part the Aura-MLS ozone and water vapor profile sampling (as mentioned in Sect. 2.3) and the measurement geometry, leading to seasonal variations in the polar ozone and water vapor distribution. Furthermore, the diurnal cycle in the ozone and water vapor has not been explicitly accounted for in the GROMOS-C and MIAWARA-C measurements. Neglecting the diurnal cycle potentially contributes to positive RDs between MWR measurement and other data sets in the upper stratosphere and lower mesosphere. Overall, GROMOS-C and MIAWARA-C are valuable to monitor the distribution of stratospheric ozone and mesospheric water vapor at the polar latitudes, respectively, which gives us more details to investigate their long-term variability, sources, and trend.

The time series of daily ozone for GROMOS-C, MERRA-2, and Aura-MLS at Ny-Ålesund, Svalbard (79∘ N, 12∘ E), extending from 2015 to 2021 are shown in Fig. . The ozone daily profiles measured with GROMOS-C cover a pressure range of 100–0.03 hPa, which corresponds to about 16–70 km. The horizontal upper and lower white lines indicate the bounds of the trustworthy pressure range where the measurement response is larger than 0.8, meaning that the measured spectrum contributes more than 80 % to the retrieved profile. The measurement data gaps are that GROMOS-C measured CO for about 2 months during winter 2017/2018 and the spectrometer had a hardware problem during winter 2016/2017 and summer 2019.

Figure  reveals the annual ozone cycle with higher ozone VMR in summer (about 6 ppmv) than in winter (about 4.5 ppmv) at about 5 hPa (≈ 35 km). The GROMOS-C ozone VMR time series together with MERRA-2 and Aura-MLS at about 5 hPa smoothed by a 30 d running mean is shown in Appendix B (Fig. a). We can see that all three datasets are able to capture the annual ozone variations well in the stratosphere. Stratospheric ozone largely follows the annual cycle of solar irradiation and is produced through the Chapman cycle; more specifically, ozone VMR is mainly dominated by photochemical production in the summer months in the Arctic middle atmosphere.

In late winter and spring, the stratospheric ozone's higher variability is largely associated with the stratospheric polar vortex. Figure  shows the ozone VMR starting to increase up to the maximum value of about 8 ppmv for some of the years in the stratosphere when the polar vortex is disturbed or weakened by the planetary waves leading to the formation of SSWs. It shows that the planetary wave activity results in meridional transport of the ozone-rich air from the subtropics towards the pole and significantly perturbed the distribution of ozone. For example, the polar vortex split and shifted away from Ny-Ålesund, and ozone VMR reached about 7 ppmv during the winter 2018/2019 SSW . In some years, the polar vortex is stable and strong over Ny-Ålesund, and ozone VMR sustains smaller values. At the end of the winter 2019/2020 season, the stratosphere featured an extremely strong and cold polar vortex, resulting in low stratospheric ozone in the polar regions . During late winter and early spring, stratospheric ozone decreases rapidly when the vortex passes over Ny-Ålesund.

and investigate the tertiary ozone maximum in the winter middle mesosphere at high latitudes based on the models and observations. In this study, GROMOS-C and Aura-MLS present the seasonal tertiary ozone layer at 0.03–0.02 hPa (about 70–75 km) in winter months, but the tertiary ozone VMRs have higher values of 15 %–20 % in MERRA-2 (as shown the red shading in Fig. b). The red shading originates from the uncertainties in the MERRA-2 product, which are expected to be magnified at high latitudes in winter and spring when the variability is increased compared to other seasons . The anomalous atmospheric dynamics, displaced or split polar vortex, and hemispherically asymmetric conditions during SSWs may cause complexity and additional uncertainties in the estimation of ozone flux and transport terms. Figure  shows the consistencies of GROMOS-C with both MERRA-2 and Aura-MLS datasets in the time series of ozone below 1 hPa. A clear annual cycle in the stratosphere is well captured by all datasets, and the higher variability of ozone in winter and spring seasons is clearly visible. Remarkably, the annual variation in MERRA-2 ozone is significantly different from the observed variation throughout the mesosphere above 1 hPa.

Figure  shows the time series of daily water vapor VMR from MIAWARA-C, MERRA-2, and Aura-MLS at Ny-Ålesund, Svalbard (79∘ N, 12∘ E), for the period 2015–2021. MIAWARA-C continuously measures water vapor profiles which cover a pressure range from 5–0.02 hPa corresponding to about 37–75 km. The horizontal upper and lower white lines again indicate the bounds of the trustworthy pressure range where the measurement response is larger than 0.8.

The most evident feature of water vapor is its annual cycle with higher mixing ratios during local summertime and lower mixing ratios during local wintertime throughout the middle atmosphere. This seasonal behavior is mainly driven by the upward and downward branches of the mesospheric residual circulation. As shown in Appendix B (Fig. a), water vapor VMR has a maximum of about 7.5 ppmv in summer and a minimum of about 3.5 ppmv in winter at 0.1 hPa (approximately 60 km). Due to the air subsidence inside the polar vortex from autumn to winter, water vapor VMR reaches the maximum at 10 hPa.

In some years, such as 2020, water vapor exhibits a larger variability in late winter and spring. The variation of water vapor is mainly affected by the occurrence of a major SSW which interrupts the polar vortex; after that point, the vortex recovers, and the lower mesospheric water vapor content increases and is accompanied by a decrease in the stratosphere, corresponding to the water vapor vertical profile in the regions outside of the polar vortex . In general, in MIAWARA-C the annually varying mesospheric distribution of water vapor agrees well with reanalysis data and satellite observations. However, there are notable differences at the polar latitudes in the water vapor VMR compared to MERRA-2 throughout the stratosphere and the mesosphere. While MERRA-2 shows a tendency to lower water vapor VMR compared to MIAWARA-C and MLS observations, the seasonal variations of similar amplitude do show reasonably good agreement with the observations. This tendency is likely related to the lack of assimilated observations and known deficiencies in the representation of stratospheric transport in the reanalysis data.

Figures  and show time series of ozone and water vapor VMR from MERRA-2 and Aura-MLS at the conjugate latitude station (79∘ S, 12∘ E) for the 2015–2021 period, respectively. For comparison purposes, the conjugate latitude station results are lagged by 6 months relative to those for Ny-Ålesund. Both stations exhibit annual cycles of ozone, while the conjugate latitude station shows an ozone VMR maximum of about 6 ppmv in summer and a minimum of about 4 ppmv in winter at about 5 hPa, as shown in Appendix B (Fig. b).

Compared to the interannual ozone variability at Ny-Ålesund, the results from the conjugate latitude station are less variable throughout the spring in the stratosphere. Compared to the NH, the planetary wave activity is much weaker in the SH, where a minimum in the ozone mixing ratios prevails over the polar latitudes during the late winter and spring. There is a dominance of an isolated and stable polar vortex which inhibits the meridional ozone transport to the South Pole throughout the annual cycle, and the formation of polar stratospheric clouds promotes the production of chemically active chlorine and bromine, leading to catalytic ozone depletion in the lower stratosphere. This process is reflected well in the MLS observations and MERRA-2 data for September and October at the conjugate latitude station. Figure  presents the water vapor annual cycle with higher mixing ratios during local wintertime and lower mixing ratios during local summertime over the conjugate latitude station. Furthermore, polar stratospheric cloud particles can sediment with considerable velocities and irreversibly remove water and nitric acid, resulting in a substantial reduction in water vapor VMR at the lower stratosphere during the southern polar winter and early spring . After September, the water vapor VMR increases again as the polar stratospheric cloud influence reduces.

While MERRA-2 generally does an excellent job at reproducing the variability of stratospheric ozone and water vapor, it fails in the mesosphere as discussed in previous sections and also underestimates the vertical extent of the ozone hole, which appears to end at lower altitudes (larger pressures) in MERRA-2 than in MLS. This is also reflected in water vapor, where MERRA-2 fails to reproduce the vertical layering (e.g., July 2015 or July 2017) and underestimates the area of dehydration. Since the reanalysis is less constrained when it is assimilated , this behavior can be seen in MERRA-2 when it is compared to observations. It further demonstrates that ozone and water vapor in the NH will continue to be available from ground-based MWR observations; however, the detailed information about the SH winter as shown here for the “conjugate latitude” will be lost after the end of the last three limb sounders, such as Aura-MLS, which is still observing.