TOMCAT/SLIMCAT (hereafter SLIMCAT) is an offline three-dimensional chemical transport model (CTM). It is forced by the ECMWF ERA5 winds and temperatures, which are interpolated to the model grid with a horizontal resolution of 2.8° × 2.8° and a total of 32 levels from the surface to ∼ 60 km (Feng et al., 2021). The model contains a detailed description of stratospheric chemistry, including heterogeneous reactions on sulfate aerosol and PSC surfaces. It has been widely used in studies of the dynamical transport and chemical reactions in the stratosphere (Dhomse et al., 2011; Chipperfield et al., 2017; Weber et al., 2021) and has been shown to accurately simulate key stratospheric chemical species (Feng et al., 2021; Li et al., 2022). The standard version of the model uses a simplified thermodynamic equilibrium PSC scheme, which includes the parameterization of liquid sulfate aerosols (LA, including binary sulfate aerosols and supercooled ternary solution, STS), solid nitric acid trihydrate (NAT), and ice. The model calculates the equilibrium vapor pressure of H₂SO₄ using the expression from Ayers et al. (1980), and when the H₂SO₄ concentration exceeds the equilibrium vapour pressure, sulfate aerosol is considered to be present. Notably, the stratospheric H₂SO₄ in the model is not provided by initial conditions but is instead calculated based on an assumed number density (10 cm⁻³) of sulfate aerosol, with the stratospheric aerosol surface area density (SAD) derived from ftp://iacftp.ethz.ch/pub_read/luo/CMIP6 (last access: 18 November 2025; Arfeuille et al., 2013; Dhomse et al., 2015; Feng et al., 2021). This scheme also calculates the concentrations, mass fractions, and solubilities of H₂SO₄, HNO₃, HCl, HBr, HOCl, and HOBr in liquid aerosols (Carslaw et al., 1995a, b).

The presence of NAT and ice is determined based on HNO₃, H₂O, and temperature using expressions from Hanson and Mauersberger (1988). NAT is considered to be present when the concentration of HNO₃ exceeds 10 times its equilibrium vapour pressure (Grooß et al., 2018). The model assumes that NAT particles exist in two modal radii (0.5 and 6.5 µm), and the SAD required for heterogeneous reactions is derived from small-radius NAT particles calculated by the condensable amount of HNO₃, with a limit of 1 cm⁻³ for the number density of small particles. The remaining HNO₃ condenses into large-radius NAT particles (Davies et al., 2002; Feng et al., 2005). Ice is assumed to form when H₂O concentration exceeds its equilibrium vapour pressure, and the model assumes a number density of 10 cm⁻³ for ice particles, with their SAD calculated from the condensable amount of H₂O. The model also uses simplified denitrification and dehydration schemes, with large-radius NAT particles and ice particles assumed to settle at velocities of 1100 and 1500 m d⁻¹, respectively (Feng et al., 2011), to redistribute HNO₃ and H₂O.

In this study, we use SLIMCAT to simulate the period 1979–2022. In addition, fixed ozone-depleting substances (ODS) for the year 2000 and climatologies of sulfate aerosol SAD and solar flux were used to exclude the effects of variations in sulfate aerosol SAD, solar flux, and anthropogenic ODS emissions. Based on the Arctic winter and early spring HNO₃ and H₂O derived from SLIMCAT, we diagnosed offline the equilibrium temperature of NAT (TNAT) according to Hanson and Mauersberger (1988), which is the warmest temperature at which PSC particles can theoretically start to form and persist (Tritscher et al., 2021; Leroux and Noel, 2024). We consider PSCs to be present when the temperature at the grid point falls below TNAT and calculate the PSC coverage areas. These areas, based on TNAT, are the theoretical maximum areas of PSC. Li et al. (2024) showed that the PSC area derived from SLIMCAT is in good agreement with CALIPSO observations in terms of seasonal evolution, interannual variability, and spatial distribution. This strengthens confidence in the performance of the SLIMCAT model in simulating PSCs.

In this study, two different methods are used to calculate the daily PSC coverage areas of CALIPSO and SLIMCAT, respectively. For SLIMCAT, the PSC coverage area is computed by summing the areas of model grid cells containing PSCs, hereafter referred to as the “Grid method”. Since CALIPSO observations of PSCs along its orbit do not cover all model grid cells, the CALIPSO PSC area is calculated using a statistical method referred to Pitts et al. (2018) (hereafter, the “P18 method”), which divides 50–90° N into 10 latitude bands, and the PSC area is estimated as the sum of the occurrence frequency of PSC in the 10 latitude bands multiplied by the area of each band. Li et al. (2024) compared the differences in SLIMCAT PSC areas calculated using the “Grid method” and the “P18 method”. They indicated that the areas derived from both methods are similar. However, when the “P18 method” is used, the daily variability in the SLIMCAT PSC area increases slightly. To avoid this issue, the “P18 method” is used to calculate the PSC coverage areas of CALIPSO, while the “Grid method” is applied to SLIMCAT.

The PSC volume is then derived by vertically integrating the PSC area from CALIPSO, MIPAS, and SLIMCAT. For MIPAS, the PSC area is calculated using the “P18 method” applied to all PSC detections up to 6 km below the cloud top height (Spang et al., 2018).

Figure 1a shows the interannual variation of Arctic PSC volume anomalies from CALIPSO and MIPAS observations and SLIMCAT simulations. Due to CALIPSO's higher detection threshold, the observed PSC volume is significantly smaller than the volumes of MIPAS and SLIMCAT (Li et al., 2024). Nevertheless, all three datasets exhibit remarkably consistent interannual variability in PSC volumes, indicating that the SLIMCAT model realistically captures observed interannual variation in Arctic PSC. Figure 1b and c (left) show the relationship between the Arctic PSC volume and the QBO index of December. Due to the limited number of MIPAS samples, we only perform linear regression on the PSC volume and QBO index for CALIPSO and SLIMCAT. Both CALIPSO and MIPAS observations and SLIMCAT simulations show that the PSC volume increases with increasing QBO index. However, the CALIPSO PSC volume does not show a statistically significant correlation with the QBO index (p = 0.38), with the limited sample size of satellite observations likely contributing to this result. In contrast, the SLIMCAT PSC volume shows greater variation with the QBO index, and the regression line passes the significance test (p = 0.01), likely due to the larger sample size. In addition to the correlation, the probability distribution of the PSC volume also shows a distinct difference between the WQBO and EQBO phases. During the EQBO phase, a large portion of the PSC volume is near zero. For example, of the 14 SLIMCAT simulations, 5 show values below 1×106 km³. In contrast, during the WQBO phase, all SLIMCAT simulations exceed 1×106 km³. The probability distribution functions (PDF) of the PSC volume during the WQBO and EQBO phases also support the above conclusions (right panel of Fig. 1b and c). The results show that the PDFs of PSC volume during both WQBO and EQBO phases exhibit skewed distributions. Compared to the WQBO, the PSC volume during the EQBO phase shifts to smaller values, suggesting that the tropical QBO can influence Arctic PSC. Note that large PSC volumes can occur during the EQBO phase, because the Arctic stratospheric vortex is influenced not only by the QBO but also by other factors, such as ENSO (Brönnimann et al., 2004; Garfinkel and Hartmann, 2008; Zhang et al., 2022).

To investigate the relationship between the QBO and PSCs, composite analyses of PSCs are performed (WQBO minus EQBO). It is important to note that before performing the composite analysis on all variables, the linear trends of the variables were removed. However, our results indicate that the detrending has a minimal impact on the composite results. Figure 2 shows the differences in the Arctic PSC area during winter between the WQBO and EQBO phases derived from the CALIPSO observations and the SLIMCAT simulations. The CALIPSO observations show that during the WQBO phase, the PSC area in January and February on 400–600 K isentropic levels is significantly larger than during the EQBO phase, whereas there is a non-significant negative area anomaly in mid-December. The differences in PSC area derived from SLIMCAT for the same periods are shown in Fig. 2b. Although the differences in the SLIMCAT PSC area are larger than those observed by CALIPSO, the simulations successfully reproduce the temporal and spatial distribution of the anomalies, such as the negative area anomaly in mid-December and the maximum positive anomaly on ∼ 500 K in January. The greater differences in SLIMCAT PSC area between the WQBO and EQBO phases primarily result from SLIMCAT simulating larger PSC areas than CALIPSO observations, likely due to CALIPSO's higher detection threshold (Li et al., 2024). We then analyze a longer dataset derived from SLIMCAT simulations from 1979–2022 (Fig. 2c). With increasing sample size, significant regions markedly increase. Furthermore, the negative area anomaly in mid-December shifts to a positive anomaly, although these positive anomalies are not significant, possibly due to the influence of negative anomalies in the 2006–2021 period. In contrast, the significance of the mean anomalies in December increases, as can be seen in the right panel of Fig. 2c. Furthermore, the maximum positive anomaly appears in January on ∼ 500 K isentropic level, and the height of the maximum anomaly gradually decreases from December to February, which is consistent with the climatological characteristics of the PSC area (Pitts et al., 2018; Li et al., 2024). These results confirm that the December QBO phase can significantly influence the PSC area in winter, with consistently larger areas observed during the WQBO phases compared to the EQBO phase.

We bin the CALIPSO PSC data into the SLIMCAT model grid, where the monthly mean CALIPSO PSC occurrence frequency is defined as the number of PSC observations in each month per grid point divided by the total number of observations. The SLIMCAT PSC occurrence frequency is defined as the number of PSC occurrences at a grid point within a month divided by the number of days in that month. Figure 3 presents the spatial distribution of the differences in PSC occurrence frequency on 500 K derived from the CALIPSO observations. During the WQBO phase, the Arctic PSC occurrence frequency shows an overall positive anomaly in winter (Fig. 3a), with a clear zonal asymmetry. The largest positive anomaly occurs over Greenland and the Barents Sea. The anomaly is largest in January, followed by December, and smallest in February. The spatial distribution of this anomaly may be related to the climatology of PSC occurrence frequency. According to Pitts et al. (2018), the monthly mean PSC occurrence frequency in the Arctic peaks in January, with a clear zonal asymmetry, and the highest frequency is centered over the Barents Sea, corresponding to the lowest geopotential height of the Arctic stratospheric vortex (Zhang et al., 2016). The zonal asymmetry in PSC occurrence frequency anomalies in Fig. 3 may also be influenced by QBO-induced shifts in the polar vortex position, which we discuss further below. In addition, due to the smaller sample size, especially at lower latitudes, banded areas of positive and negative anomalies similar to the CALIPSO satellite orbit are exhibited.

Figure 4 presents the spatial distribution of the differences in the PSC occurrence frequency between the WQBO and EQBO phases from 1979 to 2022, as simulated by SLIMCAT. Compared to the CALIPSO results (Fig. 3), the increased sample size from SLIMCAT simulations leads to a larger anomalous region being statistically significant. In addition, the differences in SLIMCAT PSC occurrence frequency between the WQBO and EQBO phases are greater than those in CALIPSO PSC. Similar to the CALIPSO results, SLIMCAT simulations show notable positive anomalies during the WQBO phase compared to the EQBO phase, which also shows a zonal asymmetric structure. Furthermore, the zonal asymmetry appears to be more pronounced in December and February than in January. The non-significant negative anomalies over northern Eurasia are observed in December and February (Fig. 4b and d), which is also present in the CALIPSO observations (Fig. 3b). Zhang et al. (2019) suggested that during the WQBO phase, the NH polar vortex tends to shift towards North America and away from Eurasia during winter, compared to the EQBO phase. This shift may affect the distribution of PSCs, resulting in a reduction over Eurasia. However, this negative anomaly does not occur in January, probably due to a less pronounced polar vortex shift associated with the QBO. Overall, during the WQBO phase, the Arctic winter shows a positive anomaly in PSC occurrence frequency, the centre of which is located between Greenland and the Barents Sea. Compared to the climatological maximum frequency of PSC occurrence (Li et al., 2024), this positive anomaly lies upstream (west) of the maximum occurrence frequency.

It is known that ENSO is another important factor influencing the Arctic stratospheric vortex in winter. During El Niño events, the Aleutian Low is deepened, planetary wave activity intensifies, and more planetary waves propagate into the mid- and high-latitudes of the NH stratosphere, thereby making the Arctic stratospheric polar vortex weaker and warmer than during La Niña events (Garfinkel and Hartmann, 2008; Salminen et al., 2020). Therefore, we should examine whether the conclusions that the QBO affects the PSC still hold after excluding the ENSO years. Figure 5 shows the differences in PSC occurrence frequency between the WQBO and EQBO phases, excluding years with strong ENSO. The results show similar patterns to Fig. 4, indicating that strong ENSO exclusion maintains the primary conclusions despite reducing the spatial significance extent due to reduced sample size.

Figure 6 shows the zonal mean differences in PSC occurrence frequency during the WQBO and EQBO phases derived from the SLIMCAT simulations. Consistent with the result from Zhang et al. (2021), which reports differences in the number of days with temperatures below 195 K, PSC occurrence frequencies are higher during the WQBO phase compared to the EQBO phase, especially in December and January. Furthermore, the seasonal variation of the differences in PSC occurrence frequency is consistent with the seasonal variation in PSC area differences shown in Fig. 2, with the altitude of maximum difference progressively decreasing from December to February. Overall, in Fig. 6a, the differences in PSC occurrence frequency peaks near 500 K, with positive anomalies exceeding 10 % north of 75° N.

The above analyses reveal that the tropical QBO can significantly influence the Arctic PSC occurrence. During the WQBO phase, the Arctic PSCs have a larger coverage area and higher occurrence frequency, which may contribute to an increase in Arctic stratospheric ozone depletion. In addition, as mentioned in the introduction, the QBO may influence Arctic PSC formation through modulating temperature, H₂O, and HNO₃. However, it is unclear how QBO-induced changes in these factors affect PSC. The underlying mechanisms by which the QBO influences Arctic PSC are now discussed.

Figure 7 presents the climatological temperature and the composite differences in temperature between the WQBO and EQBO phases. The climatological temperature exhibits a clear zonal asymmetry, with the lowest temperatures biased towards the Eurasian continent, especially near the Barents Sea. This asymmetry aligns with the favored location of the recent Arctic vortex climatology (Zhang et al., 2016; Wang et al., 2022). Significant negative temperature anomalies are observed in most of the Arctic regions between the WQBO and EQBO phases, consistent with previous studies (Holton and Tan, 1980, 1982; Zhang et al., 2019). Furthermore, the temperature anomalies associated with the QBO show a distinct zonal asymmetry, with the centre of the anomalies biased towards North America. During the WQBO phase, the polar vortex is shifted towards North America, contributing to this asymmetry (Zhang et al., 2019). It is important to note that the asymmetry of the negative temperature anomaly is more pronounced in December and February, while it is relatively weaker in January.

However, the negative temperature anomalies do not coincide well with the positive PSC occurrence anomalies, with the negative temperature anomalies located to the west of the positive PSC occurrence anomalies (Fig. 4). This is because PSC formation requires the ambient temperature to be sufficiently lower than the existence temperature (about 195 K; Tritscher et al., 2021). Only the QBO-induced Arctic cooling, occurring over the climatological cold centre, could lead to increased PSC occurrence. Therefore, the positive PSC anomalies are located near the climatological cold centre and appear to the east of the negative temperature anomalies.

In addition to the temperature, sufficient concentrations of H₂O and HNO₃ are critical for PSC formation (Li et al., 2024). Figure 8 shows the differences in H₂O and HNO₃ concentrations between the WQBO and EQBO phases observed by MLS and simulated by SLIMCAT over the Arctic (60–82° N). Both the MLS observations (Fig. 8a) and the SLIMCAT simulation (Fig. 8c) show positive H₂O anomalies above the 450 K isentropic level, with the maximum anomaly occurring in January and approaching ∼ 0.3 ppmv on the 650 K isentropic level. The altitude of the maximum positive anomaly of H₂O decreases with time similar to the decrease in the maximum positive anomaly of the PSC area, which may be related to the downward shift of the coldest centre of the polar vortex with time (Coy et al., 1997; Lawrence et al., 2018). In addition, SLIMCAT well reproduces the negative anomaly HNO₃ on 400–500 K, while the negative anomaly is smaller in December and January than in February and March.

Stratospheric species are influenced by both dynamical and chemical processes. Due to the different chemical and physical properties of the different species, the factors affecting their concentrations also vary. Stratospheric H₂O has two primary sources: transport from the troposphere and the oxidation of methane (le Texier et al., 1988). Since stratospheric temperatures increase with altitude, H₂O produced by CH₄ oxidation reactions (primarily CH₄ + OH → CH₃ + H₂O) is more important in the older air in the upper stratosphere. In summer and autumn, H₂O produced by chemical reactions accounts for 50 % of the total stratospheric H₂O at the stratopause. In contrast, in the lower stratosphere, where temperatures are lower, CH₄ oxidation is weaker, and H₂O mainly originates from dynamical transport (Thölix et al., 2016). Therefore, chemical reactions are not the primary cause of the differences in H₂O between the WQBO and EQBO phases.

Hansen et al. (2013) provided an explanation for the positive H₂O anomalies during the WQBO phase, attributing these anomalies to dynamical transport processes. Specifically, they proposed that the vertical transport anomaly of the BD circulation induced by the QBO leads to an increase in H₂O. To explore the relationship between H₂O anomalies and BD circulation anomalies, Fig. 9 shows the time evolution of the H₂O anomalies and climatology and the vertical velocity anomalies of the BD circulation in the Arctic lower stratosphere. The H₂O concentration in the polar stratosphere increases with altitude, which is caused by the CH₄ oxidation (Grooß and Russell, 2005). If the BD circulation anomalies are the direct cause of the H₂O anomalies, the anomalous upward vertical velocity, corresponding to a weaker BD circulation, would transport less H₂O-rich air from the upper layers to the lower layers, resulting in a negative H₂O anomaly. Therefore, the BD circulation anomalies cannot account for the positive H₂O anomalies as seen in MLS and SLIMCAT.

Changes in tropopause temperature and permeability may also influence stratospheric H₂O. In the Arctic, the tropopause is typically located around the 320 K potential temperature level. However, we note that the positive H₂O anomalies observed and simulated in our study are mainly concentrated above the 450 K, with no significant positive H₂O anomaly signals detected on 320–450 K (Fig. 8a and c). Therefore, we consider that local processes at the tropopause are unlikely to be the primary drivers of the H₂O anomalies above the 450 K.

Another potential cause for the anomalies in Arctic stratospheric chemical species could be the stratospheric polar vortex. This isolates the polar air from that outside the vortex. Khosrawi et al. (2016) found a clear inverse correlation between H₂O and temperature in the Arctic lower stratosphere, suggesting a high H₂O concentration in the Arctic lower stratosphere under a strong polar vortex. Pan et al. (2002) found that decreased amounts of tracers can be transported outside the vortex during strong polar vortex conditions compared to weak polar vortex conditions. In the lower stratosphere, H₂O concentrations are higher in the polar regions than in the mid-latitudes at the same altitude (Randel et al., 2001; Jiang et al., 2015), which is related to the downwelling branch of the BD circulation in the polar regions and the upwelling branch in the tropics (Morrey and Harwood, 1998). Therefore, a strong polar vortex may prevent the transport of high-moisture air at high latitudes to mid-latitudes, leading to increased H₂O concentrations within the vortex. Here, the zonal winds averaged over 60–65° N on the 500 K isentropic level are used to define the strength of the Arctic polar vortex. The relationship between polar vortex strength and H₂O concentration across different latitude bands is shown in Fig. 10a and c. Given the overall increasing trend in stratospheric H₂O concentration driven by global warming and increased methane emissions (Nedoluha et al., 1998; Oltmans et al., 2000; Dessler et al., 2013; Smalley et al., 2017), the monthly mean H₂O has been detrended before analysis. Both the MLS observation and the SLIMCAT simulation reveal a strong positive correlation between the Arctic H₂O concentration and the polar vortex strength, with correlation coefficients of 0.84 for the observations and 0.74 for the simulation. In contrast, the H₂O concentration in the mid-latitudes decreases as the polar vortex strengthens. The relatively slow decrease of H₂O in the mid-latitudes can be explained by the larger volume of air at the same latitude range, which makes it less susceptible to the influence of polar H₂O transport. The opposite evolution of H₂O inside and outside the polar vortex suggests that a stronger vortex suppresses the transport of H₂O from the polar region to mid-latitudes. Therefore, a stronger polar vortex induced by the WQBO isolates high concentrations of H₂O within the polar region, resulting in higher H₂O concentrations compared to the EQBO.

Since the change in stratospheric HNO₃ from December to February is relatively small, we display the relationship between HNO₃ from MLS and SLIMCAT and polar vortex strength from September to February (Fig. 10b and d). Overall, HNO₃ concentrations in the mid- and high-latitudes of the NH stratosphere increase with polar vortex strength. The correlation between HNO₃ and polar vortex strength is stronger in the mid-latitudes (with coefficients of 0.44 for observations and 0.48 for simulations) and is statistically significant, while the correlation is weaker in the high-latitudes (coefficients of 0.18 for observations and simulations), with weaker significance. Notably, the relationship between HNO₃ concentrations and polar vortex strength in the high-latitude stratosphere shows two distinct trends around wind speeds of 20 m s⁻¹. When wind speeds are below 20 m s⁻¹, HNO₃ concentrations increase with wind speed, but above 20 m s⁻¹, HNO₃ concentrations decrease.

HNO₃ is an important reservoir molecule for stratospheric reactive nitrogen (NO_y), produced by the oxidation of NO_x (NO₂ + OH + M → HNO₃ + M). However, under solar radiation, HNO₃ is destroyed through photolysis (HNO₃ + hv → OH + NO₂) and reactions with OH (HNO₃ + OH → H₂O + NO₃) (Keys et al., 1993), leading to lower concentrations in the Arctic stratosphere in summer. As solar radiation decreases in autumn, the stratospheric polar vortex begins to form, and the reduction in photolysis and the heterogeneous chemical reactions on background aerosols cause HNO₃ concentrations to increase gradually (Solomon and Keys, 1992; Keys et al., 1993). Therefore, as the strength of the polar vortex increases, HNO₃ concentrations increase in the mid- and high-latitude regions of the NH stratosphere. However, once wind speeds reach a certain level and the polar vortex becomes sufficiently strong and isolated, temperatures can become low enough for NAT to form. If a sufficient amount of NAT has formed, denitrification occurs within the polar vortex due to the NAT sedimentation, leading to a decrease in HNO₃ concentrations. This explains the observed decrease in HNO₃ concentrations at wind speeds above 20 m s⁻¹, as shown in Fig. 10b and d. During winter and early spring, Arctic temperatures are sufficiently low for denitrification to occur. Therefore, the lower temperatures during the WQBO phase result in lower HNO₃ concentrations compared to the EQBO phase (Fig. 8b and d).

Both H₂O and HNO₃ concentrations are influenced by PSC sedimentation. When the polar vortex is strong enough, both H₂O and HNO₃ concentrations decrease. It is important to note that different types of PSCs have different effects on stratospheric chemicals. NAT sedimentation primarily affects HNO₃ concentrations, while ice PSC sedimentation primarily affects H₂O concentrations. The contrasting thermal climatologies between the Arctic and Antarctic lead to fundamental differences in PSC composition, with ice PSCs constituting about 25 % of Antarctic PSCs but less than 5 % of Arctic PSCs (Pitts et al., 2018). As a result, ice PSC-related dehydration occurs annually in the Antarctic but rarely in the Arctic (Khaykin et al., 2013). Since NAT accounts for approximately 40 % of PSCs in the Antarctic and up to 60 % in the Arctic, significant denitrification can occur in both polar regions. Due to the dominance of NAT and the scarcity of ice particles in Arctic PSCs, the effects of the QBO on Arctic stratospheric HNO₃ and H₂O are markedly different.

As mentioned above, the WQBO phase tends to lower the Arctic temperatures (Fig. 7), reduce HNO₃ concentrations, and increase the H₂O concentrations (Fig. 8). The lower temperatures and higher H₂O concentrations are favourable for PSC formation, while the lower HNO₃ concentrations are unfavourable (Hanson and Mauersberger, 1988). However, their relative contributions remain unclear. To address this issue, we perform sensitivity analyses in which the temperature, H₂O, and HNO₃ are adjusted by adding or subtracting 50 % of the differences between the WQBO and EQBO phases. It is important to note that we did not adjust temperature, H₂O, and HNO₃ during the model runs, but instead made these adjustments during the offline diagnosis of the PSCs. Our adjustment strategy aims to reduce the PSC area during the WQBO phase and increase it during the EQBO phase, to investigate the main factors influencing Arctic PSCs by changing temperature, H₂O, and HNO₃. Table 2 provides the descriptions of the six sensitivity analyses.

Note that the temperature variation induced by the QBO accounts for most of the changes in the PSC areas (Fig. 11a–d). In particular, during the WQBO phase, when the temperature is subtracted by 50 % of the differences between the WQBO and EQBO phases, that means warmer than the original temperature, the PSC areas decrease and approach those during the EQBO phase. In contrast, during the EQBO phase, when the temperature is added by 50 % of the differences between the WQBO and EQBO phases, that means colder, the PSC area increases slightly. This suggests that PSC formation is more sensitive to temperature variations during the WQBO phase. Since Arctic temperatures are concentrated around the PSC formation threshold (Fig. 7), PSCs are highly sensitive to temperature changes, and even small changes in temperature will result in significant variations in PSC. QBO-induced changes in H₂O and HNO₃ lead to only small changes in PSC. During the WQBO phase, when the H₂O decreases by 50 % of the difference, the PSC area becomes smaller, although the magnitude of the change is relatively small (Fig. 11e–h). QBO-induced HNO₃ changes affect the PSC in late winter and early spring (Fig. 11k–l). During the WQBO phase, lower Arctic temperatures lead to stronger denitrification, resulting in less HNO₃ participating in PSC formation. These results show that it is important to simulate the NAT sedimentation process in the model accurately.