Double cores of the Ozone Low in the vertical direction over the Asian continent in satellite data sets

Using four satellite data sets (TOMS/SBUV, OMI, MLS, and HALOE), we analyze the seasonal variations of the total column ozone (TCO) and its zonal deviation (TCO*), and reveal the vertical structure of the Ozone Low (OV) over the Asian continent. Our principal findings are: (1) The TCO over the Asian continent reaches its maximum in the spring and its minimum in the autumn. The Ozone Low exists from May to September. (2) The Ozone Low has two negative cores, located in the lower and the upper stratosphere. The lower core is near 30 hPa in the winter and 70 hPa in the other seasons. The upper core varies from 10 hPa to 1 hPa among the four seasons. (3) The position of the Ozone Low in the lower and the upper stratosphere over the Asian continent shows seasonal variability.


Introduction
The ozone layer is important not only to stratospheric and tropospheric climate (Andrews et al., 1987;Xie F et al., 2016;Zhang JK et al., 2018;Guo D et al., 2017a), but also to ecological systems (Fuhrer and Booker, 2003). Without the ozone layer, humans, animals, and plants could not live (Van der Leun et al., 1995). Molina and Rowland (Molina and Rowland, 1974) discovered that chlorofluorocarbons were thinning the ozone layer. Furthermore, great ozone loss was found over the South Pole (Farman et al., 1985) and in the Arctic (Newman et al., 1997). Therefore, ozone depletion attracted much attention. Meanwhile, a similar pattern of ozone depletion was found in the middle latitudes. Based on Total Ozone Mapping Spectrometer (TOMS) data, Zhou XJ et al. (1994) first found a noticeable low value center by comparing the TCO over the Tibetan Plateau (TP) with the eastern part of China at the same latitude. Zou H (1996), however, calculated the zonal deviations of global total ozone in different seasons of 1979-1991 using TOMS data, confirming the Ozone Low by zonal deviations. But these definitions of the Ozone Low are not the same and do not have a quantitative indicator. Bian JC et al. (2006) reported an unusual Ozone Low in the winter over the TP based on longer TOMS datasets. Recently, Guo D et al. (2015) found a double core structure of the Ozone Low using the Aura Microwave Limb Sounder (MLS) data. Besides analyses of the phenomenon, there are some studies on the Ozone Low mechanism (Guo D et al., 2017a, b ;Li ZK et al., 2017). In previous studies, the contribution of the dynamic process to the formation of the OV was associated with the South Asian High (SAH) (Liu Y et al., 2003;Li ZK et al., 2017;Tian WS et al., 2008;Qin H et al., 2018;Guo D et al., 2012). Moreover, many studies have suggested that dynamic effects dominate the Ozone Low formation while stressing that the chemical effect is also a possible factor. Fu C et al. (1997), Liu Y et al. (2010), and Guo D et al. (2015) emphasize the chemical factors. Tian WS et al. (2008) claimed that the atmospheric column loss, which is impacted by the terrain, can partly explain the results in column ozone loss. There are also other studies focused on detailed features of the Ozone Low (Ye ZJ and Xu YF, 2003;Zhou SW and Zhang RH, 2005;Moore and Semple, 2005;Tobo et al., 2008;Liu Y et al., 2009;Zhang JK et al., 2014). In addition, the Ozone Low is found not only over the TP, but also over the Iranian Plateau (IP) (Wang WG et al., 2006;Yan RC et al., 2011).
However, those studies on the Ozone Low do not have a unified criterion and show varying uncertainties in the different datasets. Furthermore, previous studies do not consider the seasonal change of the Ozone Low. To clarify these problems, we examine similarities and differences of the Ozone Low using four satellite observation and one observation datasets (described in the following section). Section 3 presents seasonal variation and vertical structure of the Ozone Low. Sections 4 and 5 present discussion and conclusions.

Data
We considered four satellite observations and one observation data. They are briefly described below.
The Ozone Monitoring Instrument data version 3 (OMI) was launched on NASA's Earth Observing System's Aura satellite. OMI measures backscattered hyperspectral radiation in the range of 270-500 nm in three channels (Levelt et al., 2006). We use OMI data for total column ozone (TCO). The resolution of the TCO is 1° × 1° (latitude × longitude) and temporal coverage is from January 2005 to December 2014.
The Microwave Limb Sounder on the Aura satellite data v4.2 (MLS) instrument was launched on the Aura satellite. It looks through the atmospheric limb along the orbital track and obtains a vertical ozone (O 3 ) profile by scanning the field of view up and down (Waters et al., 1999(Waters et al., , 2004(Waters et al., , 2006. Because MLS detects microwave emissions, this instrument measures ozone profiles during both the daytime (ascending) and nighttime (descending). Moreover, the number of observations in the day and night times are almost the same. Much rigorous and detailed verification work has been done to evaluate the MLS data (Yan XL et al., 2015;Shi CH et al., 2017). We use MLS satellite data v4.2 (MLS) both for the total column ozone and for the vertical profiles of ozone (O 3 ). The resolution of the TCO is 1° × 1° (latitude × longitude) and temporal coverage is from January 2005 to December 2014 for the total column ozone. The total column ozone in MLS represents the ozone concentration in the stratosphere. But for the vertical profiles of ozone, the horizontal resolution of the ozone is 2° × 2° and temporal coverage is from January 2005 to December 2014. There are 55 layers of the ozone in the vertical direction. In addition, the ozone unit is parts per million by volume (ppmv). More publications on MLS can be obtained from the website of the Jet Propulsion Laboratory.
The Halogen Occultation Experiment data version 19 (HALOE) was launched on the Upper Atmosphere Research Satellite (Russell Ⅲ et al., 1993(Russell Ⅲ et al., , 1994. The experiment uses solar occultation to meas-ure vertical profiles of ozone (O 3 ) and other tracers. HALOE tracks the top edge of the Sun. There are no systematic differences in the tracking procedures and no significant biases in the record between sunrise and sunset (Nazaryan et al., 2005). The horizontal resolution of the ozone in this article is 5° × 5° and temporal coverage is from January 1992 to December 2005. There are 271 layers of the ozone in the vertical direction.

Methods
Here [F] is zonal mean of F. According to this definition, the zonal deviation (TCO * ) of the total column ozone (TCO) and the zonal deviation ( ) of the ozone concentration profile (O 3 ) will be investigated. The smaller the negative values of TCO * and , the stronger the Ozone Low. In this work, the TP area is defined as 25°N to 40°N, 80°E to 100°E. The integrated column Z (units: DU) can be calculated as follows (Bian JC et al., 2011): Here, P is the pressure (units: hPa) and M is the mixing ratio of volume (units: ppmv).

Seasonal Variability of the TCO
The TCO on the Asian continent shows seasonal variability in different periods for different datasets (Figure 1). A trough of the TCO can be found over the TP in all four seasons. Over the TP, an Ozone Low is identified in spring (March, April, and May, short as MAM) in the OMI and TOMS satellite observations. The value is about 290 DU in the TOMS and 280 DU in the OMI. Obviously, the Ozone Low becomes lower in the summer (June, July, and August, short as JJA), where a value of 270 DU is both observed in the TOMS and OMI. And the scope of the Ozone Low is enlarged. In the autumn (September, October and November, short as SON), the Ozone Low drops to 255 DU in the OMI and TOMS observations. In the OMI, the scope of the Ozone Low is reduced in winter (December-February, short as DJF). The Ozone Low is about 260 DU over the TP. However, the OV disappears in the TOMS data. Over the TP, a very weak trough is identified in the winter. The TCO shows a quite different seasonal change in the MLS satellite observations; the Ozone Low appears only in the summer. There exists a closed center controlling the western side of the Asian continent. Additionally, we can find a low trough in Spring and Autumn on the Asian continent.
It is obvious that the TCO shows a seasonal variation over the TP area. To display clearly the seasonal changes in intensity of the Ozone Low, the multi-year-average of the TCO over the TP has been calculated. Figure 2 shows that the lowest value of the TCO occurs in October and the highest in April in the TOMS data and in March in the OMI data. Thus, we can conclude that the TCO over the TP reaches its maximum in the spring and its minimum in the autumn. Error bars indicate ±2 standard deviations. The standard deviation reaches the maximum in February in the OMI and MLS and in March in the TOMS, which means that there is a great un-certainty of the TCO in the spring and winter.

Seasonal Variability of the TCO *
TCO * also shows a seasonal variation of pattern and intensity in different datasets (Figure 3). In the spring, there exists a negative value center about -25 DU near the TP in the TOMS and OMI. A negative value center about -20 DU can be found on the northwest of the TP in the MLS. There exist two negative value centers in the summer in the TOMS and OMI. One is about -25 DU over the TP, the other is about -20 DU over the Iranian Plateaus (IP). In the MLS, two negative value centers can be also found, which are to the north of those in the other datasets. One is over the Caspian Sea with the center value about -25 DU, the other is about -20 DU over the TP. In the autumn, a negative value center about -15 DU can be observed over the TP in the TOMS and OMI. The IP is still dominated by negative TCO * , but the negative value center disappears. The TCO * centers can be identified over the TP in the TOMS and OMI only in winter, with the central value about -10 DU. By contrast, TCO * shows a quite different pattern in the MLS in the autumn and winter. There are no obvious negative value centers on the Asian continent. Furthermore, a positive value center can be detected over the TP in the winter.
The TP area average of the TCO * is shown in Figure 4. Error bars indicate ±2 standard deviations. Compared to the multi-year-average of the TCO (Figure 2), the TCO * shows (Figure 4) totally different results. In the datasets, the TP area average of the TCO * is lowest in May and highest in January. It also reflects that the Ozone Low is strong in summer and weak in winter over the TP accord-    ing to the latitudinal distribution of ozone. If all the data obey the standard normal distribution, when error bars of all data for each data set are under the zero line in a month, there is at least a 95% possibility of the Ozone Low occurrence in that month. Hence, ac-cording to that condition, we find that the Ozone Low exists from May to September.
In summary, we conclude that the TCO over the TP reaches its maximum in the spring and its minimum in the autumn. The Ozone Low is strong in summer over the TP. There are obvious negative values of the TCO * over the TP from May to September.

Vertical Structure of the Ozone Low
Over the TP, the ozone maximum concentration in volume relative mixing ratio is in the middle stratosphere in all four seasons (figure omitted). The peak intensity is higher in the spring and summer. But the ozone deviation has two negative cores, located in the lower and the upper stratosphere ( Figure 5).
The vertical structure of the over the TP shows a similar pattern in the MLS and HALOE, and the position and intensity of the extremum exhibit seasonal variability. In the lower stratosphere, the extreme value of the is close to 30 hPa in the winter but 70 hPa in the spring, summer, and autumn. The extremum is slightly smaller in the spring and summer. The extremum is negative. Thus, the lower core of the Ozone Low is stronger in the spring and summer. Moreover, there exist error bars of the in the lower stratosphere less than zero in the summer in all data. Therefore, it is certain that there exists an ozone negative   The positions of the upper cores vary from 10 hPa to 1 hPa in all four seasons in the MLS and HALOE data. The extreme value of the is larger in the spring and summer, which is on the contrary to the extremum in the lower stratosphere. Comparing the value of the upper core and the lower core, the extremum in the lower stratosphere is slightly bigger in the spring and summer, while in the autumn and winter, the extremum of the upper layer is slightly bigger.
In order to determine the horizontal pattern of the Ozone Low, we calculated the vertical integration (100-30 hPa and 10-1 hPa) of mean in the four seasons.
The vertical integration in the upper stratosphere from 10 hPa to 1 hPa in the MLS and HALOE data ( Figure 6) shows that there exists a negative value center about -0.5 DU over the TP in the spring. By summer, the negative value center has shifted slightly to the northwest, with the value center around -0.5 DU in the MLS. However, the negative value center disappears in HALOE. In the autumn, the ozone negative center moves in the two datasets to the eastern part of China at the same latitude with the TP. In the winter, the ozone negative center moves out of the Asian continent. Meanwhile, there are still negative areas over the TP, and the O * 3 value of can reach -0.5 DU, lower than that in the spring and summer. This can explain why the upper centers of the TP are relatively stronger in the autumn and winter over the TP in Figure 5 although the negative value center of the continent of Asia is not over the TP.
In contrast, the integration from 100 to 30 hPa shows some different characteristics (Figure 7). In the spring, negative values occupy the whole Asian area. In the MLS data, the negative center is located on the west of the TP. The center value reaches -12 DU. In the HALOE data, the negative center is located above the TP, and the center reaches -8 DU. In summer, the negative center is over the Caspian Sea, with the center value about -12 DU in the MLS and -8 DU in HALOE. In the autumn, in the MLS there is a large negative center around -4 DU on the west of the TP; in the HALOE, however, the negative center about -4 DU can be observed just over the IP. In the winter, no obvious ozone negative center appears on the Asian continent in either of the two datasets.

Discussion
The seasonal variability of the TCO * is quite different from the vertical integration (100-30 hPa and 10-1 hPa) of the mean in the  four seasons. The possible reason for this result may be the unique terrain of the TP. In addition, the Ozone Low is found not only over the TP in the lower and the upper stratosphere. The position of the Ozone Low on the Asian continent changes with the seasons, which needs further study.

Conclusions
By using four satellite data sets (the TOMS/SBUV, the OMI, the MLS and the HALOE), we have analyzed seasonal variations of the TCO and TCO * and revealed the vertical structure of the Ozone Low.
The results are described below: (1) The TCO over the TP reaches the maximum in the spring and reaches the minimum in the autumn. The Ozone Low exists from May to September.
(2) The ozone deviation over the TP has two negative cores, located in the lower and in the upper stratosphere. The lower core is near 30 hPa in the winter and 70hPa in the other seasons. The upper core varies from 10 to 1 hPa among the four seasons. The lower core of the Ozone Low in the lower stratosphere is stronger in the spring and summer. The opposite is true in the upper layer. (3) According to the ozone content of the upper stratosphere, the Ozone Low on the Asian continent exists except in the winter in MLS and in the spring and autumn in HALOE. But for the lower stratosphere, the Ozone Low on the Asian continent exists in all sea-sons except the winter. The position of the Ozone Low in the lower and the upper stratosphere on the Asian continent shows seasonal variability.