Ozone Variability over Egypt

The variability of ozone over Egypt has been studied in this work. Higher values of coefficient of variation (COV) at eight stations occur at winter while lowest values occur at summer. The COV is function of latitude in annual, winter and spring where it decreases gradually from the north to south of Egypt. The trend analysis of ozone over the eight stations indicates negative trends over northern stations for both annual and seasonal time series with the greatest one at winter. The horizontal distribution of ozone trend values is negative over all Egypt in winter while it positive over middle and south Egypt in the other seasons. The long-term variability of the behavior of the annual ozone shows positive trend values in ozone are the dominant features during the period 19791989 at the most stations. Negative trend values in ozone are the dominant features during the period 19902014 at all stations. The Mann–Kendall test confirms that there is an abrupt change towards decreasing of ozone occurs in 1981, 1984, 2012, 1999, 2000, 1998 and 2010, while a change towards increasing of ozone appears in 2006, 2014, 1993 and 1989.


Introduction
Global total column ozone amounts (which are dominated by ozone in the stratosphere) declined by a few percent between roughly the 1970s and the start of the current century (WMO, 2011), Total ozone at any location on the globe is defined as the sum of all the ozone in the atmosphere directly above that location (David.W. et al, 2014), although its small proportion in terms of the total atmospheric composition, ozone plays an important role in global weather and climate as well as surface ecology.Ozone prevents most harmful UV radiation from reaching the biosphere, Due to this high energy, UV-B radiation can have several harmful impacts on human beings ( DNA damage, skin cancers, corneal damage, cataracts, immune suppression, aging of the skin, and erythema), on ecosystems, and on materials ( UNEP, 1998( UNEP, , 2003)).So we have to protect ozone layer from depletion, the study variation and trend of total column ozone is very important to detected its behavior many early studies has performed to the trend of ozone depletion, one of the ways to study the time series behavior is Abrupt change, which can be generally defined as occurring when some part of the climate system passes a threshold or tipping point resulting in a rapid change that produces a new state lasting decades or longer (Alley et al., 2003).In this case "rapid" refers to timelines of a few years to decades.
Both satellite and ground based data have documented a process of high stratospheric ozone depletion in the Antarctic environment during spring months (Farman et al., 1985;Stolarski et al., 1986).While this effect cannot be expected to apply in other places to the same extent, studies that are more recent have indicated that ozone depletion is not confined to Antarctica, but that it has global features (Atkinson et al., 1989;Bojkov et al., 1990;Stolarski et al., 1991Stolarski et al., , 1992)).There is some evidence showing that the decreased ozone levels have caused increases of ultraviolet radiation at the surface (Kerr and McElroy, 1993).Analyses of total ozone have focused on trend analysis (Reinsel et al., 1994;Bojkov and Fioletov, 1995) due to the increasing concern about dramatically decreasing stratospheric ozone.However, a full description of the variability of ozone should also include means, seasonal variation and extremes.This information should be more useful in understanding ozone dynamics.Evidence of global stratospheric ozone loss has been largely based on satellite observations."Over the northern hemisphere middle and high latitudes, data from Dobson stations have been used (Krzyscin, 1994).
To measure long term trends in total ozone, an accurate instrument is required.The Brewer type spectrophotometer has proven to be one of the world's most accurate ozonemeasuring devices (WMO, 1994).It should be remembered that ground-based instruments give point measurements, while satellite data cover a certain area.The objective of this work is to study the variability and trend analysis of total column ozone over Egypt.

Data
In Egypt there are four stations providing ground-based total ozone measurements.
Table 1 illustrates the name, location, the instrument type used to measure total ozone and the period of measurements for each station.Occasionally, there are missing data (about 1% of the total data) which need to be dealt with in some fashion for calculation of the climatology.Linear interpolation is used to estimate missing values in order to make the time series complete, which enables a computation of monthly and yearly means and a time series analysis of the daily data.We also extract another four time series of TCO for the previous four stations from reanalysis ERA-Intrem total column ozone data (http://apps.ecmwf.int/datasets/data/interim-full-moda).To cover and represent the different regions of Egypt we also obtain the time series of TCO for the stations Alex  1.It is found that there is a very good agreement between observed and reanalysis data, where the correlation coefficient is 0.91 and the root mean square error is 6.1, this good agreement allowing us to use this data in our study.Figure 2 shows the study area and the location of eight stations, four of them are the ground-based stations while the other four stations we used the reanalysis ozone data.The reanalysis daily and monthly data for total amount of ozone and other meteorological parameters are obtained from (http://apps.ecmwf.int/datasets/data/interimfull-moda).ERA-Interim reanalysis data is the latest global atmospheric reanalysis data set of the European Center Weather Forecast (ECMWF) with period from 1979 till 2014 with resolution 1 o x1 o is used for obtain total column ozone from meteorological parameters.

Methodology
Bartlett test con be used to examine the homogeneity of the data when Gaussian distribution of values is considered.The method is simply accomplished by dividing the considered time series into K equal subperiod (k ≥2).The sample variance ( ) Where the summations range over the n values of the Series, Σi in the subperiod K.The estimated ratio between the maximum and minimum values ( ) A coefficient of variation (COV) for each individual station has been determined as follows: Where, SD is the standard deviation and μ is the temporal mean for N years.
Different methods have been applied to study the trend and fluctuations of ozone over the studied area.These methods are; (a) Linear regression by using Least Square method (Panofsky and Brier ,1963), (b) Gaussian low-pass filter (Mitchell et al 1966), (c) Binomial low-pass filter (Mitchell et al. ,1966 andTyson et al.,1975).Also the nonparametric Mann-Kendall (M-K) rank correlation test (Sneyers, 1990;Schonwiese and Rapp, 1997;Hasanean, 2004) has been used to detect any possible trend in ozone series, and to test whether or not such trends are statistically significant.To visualize the decadal and inter-decadal fluctuations or "persistence" in the behavior of the KSA ozone, cumulative seasonal means method is used (Pavia and Graef 2002).The advantage of this is to reveal time varying structures in time series.The cumulative seasonal means time series can be defined as; Where, i x is the total amount of ozone and N is the number of years of data used.

Abrupt change procedur:
"To statistically identify the abrupt climatic changes in the annual mean TCO time series, the nonparametric Mann-Kendall (MK) rank statistic is employed (Sneyers 1975(Sneyers , 1990)).This is a suitable method for testing any abrupt change in mean values (Goossens and Berger 1986; XiaoHui and XiangDe 2007).The MK method can be considered in detecting the beginning of any change in a time series (Goossens and Berger 1986).The intersection between the two resulting trend lines in the region ±1.96 of the standardized statistic, at the 5 % significant level, is considered to be an evidence of an abrupt change in the time series.Estimation of MK rank statistic is done by considering only the relative values of all the terms in the time series i X under analysis and replacing the i X values with their ranks i Y , arranged in increasing order.
Where the mean A significant trend at the threshold 1 α (e.g., 0.05) is observed when α threshold is determined by the means of the standard normal distribution table, such that ). ( If the values of are significant, then they can be considered whether the trend is increasing or decreasing, by looking at how However, when a series shows a significant trend, it is necessary to locate the start of the abrupt change by means of a sequential analysis.In this case, the analysis is performed with the same time series but with reversed sequence.To identify an abrupt change in the time series, this analysis is performed with the same SAT time series but with the reverse sequence, i.e., number Where N is the total number of elements in the series.Therefore, the values of for the reversed series can be calculated; this is similar to as in Equation 7. In the absence of any trend in the series, the graphical representation of u and u′ in terms of i generally gives curves, which overlap several times.

Mann-Whitney test for step trend
Given the data vector ) ........, , , ( , partition X so that is the rank of the observations.The null hypothesis H0 is accepted if quantiles of the standard normal distribution corresponding to the given significance level α for the test.Homogeneity of the total column ozone (TCO) of the annual and seasonal time series over Egypt stations has been detected by means of Bartlet test (described in section 3).Table 2 shows Bartlet test (short-cut) result for annual and seasonal ozone of land based stations and reanalysis data respectively.The mean annual and seasonal of TCO seems to be homogeneous for all stations regarding the ratio (

Coefficient of variation (COV)
The variability in total column ozone at a single location is strongly influenced by the movement of air from one region to another.Thus, total column ozone averages over the entire globe, or over large regions, often show less variability than total column ozone at a specific location (Bodeker et al., 2001).

Trend analysis
The annual ozone time series for the eight stations under study have been investigated to determine their trends.The studies of trends have been performed by means of both a simple and sophisticated tools.These methods are: 1) Least square method of first order (linear regression), 2) Moving filters by using Gaussian low-pass filter and binomial low-pass filter and 3) Mann-Kendall rank statistical test.Discussion of the important aspects of the long-term variations and trends are given below for annual ozone time series.

Trend by least square method
Table 4 illustrates the values of trend by least square method for the annual and The trend of the reanalysis TCO time series of the eight stations is positive for the autumn season.Also, the trend of the southern three stations (Hurgada, Aswan and Ewinat) is positive for the annual and seasonal TCO time series.

Fluctuation of annual Ozone
Trends of annual ozone of the concerned stations by means of Gaussian and binomial low-pass filter (dotted curve, thin curve) has been investigated for ground based (observed) and reanalysis data (Figures 4 and 5 ).Also, there is a high consistence between the periods of increase and decrease in observed and reanalysis data of Cairo especially for the increase period 1979-1983.Two periods of high consistence between observed and reanalysis data appears at Aswan, the first is from 2006 to 2012 while the second is from 1991 to 2000.

Mann-Kendall rank statistical test
The values of M-K trend test (u) were valued according to Sneyers (1990).Table 5 shows the M-K statistics for the annual, winter, spring, summer and autumn measured and reanalysis time series of the considered stations.With respective to the measured four time series, positive trends are observed over the four stations for the annual and seasonal time series except for the winter time series of Cairo.

Cumulative annual means
To visualize the decadal and inter-decadal fluctuations or "persistence" in the behavior of the Egypt ozone, cumulative annual means method is used (Pavia and Graef 2002).The advantage of this is its ability to reveal time varying structures in time series.
In this section, the long-term variability of the behavior of the annual ozone is analyzed with regard to the time variations of annual ozone.Cumulative annual means will be used to visualize the decadal and inter-decadal fluctuations in the behavior of the annual ozone, because they advantageously reveal time varying structures in time-series, which are not readily recognizable in raw data.Moreover, the cumulative means have a smoothing effect, similar to low-pass filters (Lozowski et al., 1989).Persistent phases of alternating increase or decrease of the ozone, which vary in length, are recognizable in the time series of the annual ozone.

Abrupt change analysis
The concept of abrupt change is by now well established.For example Rogers  1998, 1999, 2002 and 2004 respectively.The second one is the increase period of 2007-2014, in which the mean ozone is 2.48 DU higher than the average of the whole period.
In the increase period, the positive departures account for more than 52%, in which the highest ozone for 17 year and the ozone in 2009 is 6.76 DU higher than that of the whole period.The Manne Whitney test detected two abrupt changes in the ozone time series of Cairo the first one is around 1979 at the 0.05 significance level while the second one is around 1993 at the 0.08 significance level, Figure 11b shows the departure curves of the ozone in Cairo station.It is shown that the departure curve fluctuates significantly, with the characteristics of two decreasing and one increasing abrupt changes since the 1968s.
The period of more ozone, 1980-1992, in which the mean ozone is 2.9 DU more than that of the whole period.The first period of less ozone from 1968 to 1977, in which the mean ozone is 4.49 DU less than that of the whole period.The second period of less ozone is from 1994 to 2014, in which the mean ozone is 0.29 DU less than that of the whole period.

Summary and conclusions
The variability of ozone over Egypt has been studied in this work.

(
29 o E, 31 o N), Ismail (32 o E, 30 o N), Dahab (34 o E, 28 o N) and Ewinat (28 o E, 22 o N).The comparison between TCO of ground based station and its corresponding reanalysis data has been verified over Cairo and represented in Fig.
Study of the data homogeneity is important in climatology.Homogeneity is established differently in different climatic elements.The values of climatic elements could be the method of estimating daily and monthly averages.Artificial lakes and reservoirs and other man-made changes of local environment produce sources of inhomogeneity in historical record of climatic data.

Preprints
(www.preprints.org)| NOT PEER-REVIEWED | Posted: 13 February 2017 doi:10.20944/preprints201702.0040.v1seasonalozone of Egypt stations.It showed that there is a positive trend of annual and seasonal ozone at four ground-based stations.It is clear that the only negative trend occurs only in winter over Cairo (-0.15 DU/year), also the minimum annual trend value occurs at Cairo (0.03DU/year).With respect to the reanalysis TCO time series of the eight station, it is found that the maximum annual and seasonal trends occurs at Hurgada and Matrouh, it may be due to the location of these two costal stations( Beig et al. 2007).

Figure 4
Figure 4 illustrate the results obtained from the two methods for the observed values of TCO.Careful examination of the obtained results shows that during the period 2004 to 2013 an obvious increase of TCO occurred in Matrouh and Aswan, the increase in mean annual TCO reached to 4 DU.A pronounced decrease of TCO at Matrouh appears from 1998 to 2002, after that TCO have its average value until 2005.From 2006 to 2013 an obvious increase of TCO occurs at Matrouh, the maximum increase (4 DU) appears at the years 2009 and 2010.For Cairo station a decrease of TCO has been found from 1969 to 1977 with a maximum reached to 4 DU during 1970 to 1973, followed by an obvious increase of TCO from 1978 to 1984.During the period from 1985 to 2014, the values of TCO fluctuate around the mean.For Hurgada a decrease of TCO is found from the beginning until 2005, this period followed by a persistence of TCO around the mean until 2009.During the last period TCO fluctuate around its mean with considered amplitude.The behavior of TCO time series in Aswan can be divides into two obvious periods, the first one is from 1990 to 2000 is dominated by a decrease of TCO.The second period (2004-2012) is characterized by a pronounced increase of TCO.

Figure 5
Figure 5 illustrate the results obtained from the two methods (Gaussian and binomial low-pass filter) for the reanalysis values of TCO at eight stations.Generally, the time series of the 8 stations have nearly the same behavior, where we found that the

Figure 7
Figure 7 illustrate the cumulative annual mean (CAM) time series and the averaged CAM of the measured annual values of ozone of the stations Cairo, Matrouh, Hurgada and Aswan.Results from Cairo station show that CAM time series can be divided into three periods.The first period (1968-1971) illustrates a positive trend, the second period shows a negative trend, while the last period (1981-2014) shows a fluctuation of the trend around the CAM average.The cumulative annual means for Matrouh and Hurgda have the same behavior throughout their observational period.They show a negative trend in TCO during the first period (1998-2005) followed by a continuous positive trend until 2014.The maximum average CAM for these stations occurs in Matrouh (297.2DU).The cumulative annual mean for Aswan throughout its observational period shows a negative ozone trend during the first period (1984-1989) followed by a positive trend until 1994.A negative trend is detected at Aswan for the period 1994-2004 followed by a positive trend until 2014.The average CAM of this station is about 271 DU.

Figure 8
Figure 8 illustrate the cumulative annual mean (CAM) time series and the averaged CAM of the reanalysis annual values of ozone of the eight stations.It is clear that

( 1985 )
described the marked changes in sea level pressure patterns of the North Atlantic during the 1920s.Flohn (1986) expanded the concept of abrupt climate change to include both singular events and catastrophes such as the extreme El Nino of 1982-1983.Trenberth (1990) and Miller et al. (1994), among others, documented a rather abrupt shift of sea surface temperature and atmospheric circulation features in the Northern Hemisphere in the mid-1970s.The concept has been extensively used as the foundation of many different methods to detect sudden irregular changes or discontinuities in SAT (Bengtsson et al. 2004; Bromwich and Fogt 2004; Trenberth and Shea 2005).Figures 9, 10 and Table 6 show the results of the Mann-Kendall rank statistic for the annual mean TOC using both the forward and backward sequential statistic methods.They reveal that abrupt climatic changes have taken place.However, the abrupt changes occur in the time series of the reanalysis data of the eight stations (Figure 10), while it occurs only in Cairo and Aswan of the measured data.Decreasing trends in TOC are evident for six stations in 1981 and 1984 (Matrouh, Alexendria, Ismail, Cairo, Dahab and Hurgada).For Aswan station, the same decreasing trends are detected around the years 1989, 1998 and 2010 (Fig. 10 and Table 6), while increasing trends appears around the years 1987, 1993 and 2008.For Ewinat station, the abrupt change toward decreasing occurs twice, in 1998 and 2009 (Fig.10 and Table 6), while increasing trends in TOC are evident in 1987, 1993 and 2008.In the time series of measured TOC of Aswan the abrupt change toward decreasing occurs in 2010, while increasing trends occurs around the period 1998 to 2004.The results of Ewinat illustrate that abrupt change toward decreasing occurs around 1998 and 2009, while increasing trends appears in 1989, 1993 and 1989.Abrupt change toward increasing of TOC occurs around the year 2006 appears at six stations (Matrouh, Alexendria, Ismail, Cairo, Dahab and Hurgada), and around 2014 at three stations (Ismail, Dahab, Hurgada).A trend toward decreasing around 2012 occurs in Ismail, Cairo, and Dahab.From the above results one can conclude that a change towards decreasing TOC is obtained in 1981, 1984, 2012, 1999, 2000, 1998 and 2010 and a change towards increasing TOC is obtained in 2006, 2014, 1993 and 1989 (Table 6).The results of the Manne Whitney test for abrupt change of the time series of TOC for the four ground based stations are illustrated in Figure 11. Figure 11a shows the departure curves of the ozone in Matrouh, departure is the deference of climate variables for 17 years.The test detected abrupt change in the ozone time series of Matrouh around 2006 at the 0.01 significance level.Also there are two obvious periods in the study area for the past 17 years, the first one is the decrease period of 1989-2005, in which the negative departures account for more than 47%, and the abnormal decrease years are Whitney test detected abrupt change in the ozone time series of Hurgada around 2009 at the 0.05 significance level, Figure 11c shows the departure curves of the ozone in Hurgada.Departure is the deference of climate variables for 14 years.It is shown in Figure 11c there are two obvious periods in the study area for the past 14 years.The first one is the lower period from 2001 to 2008, in which the ozone is 1.38 DU lower than that of the whole period.The second one is the higher period from

Fig. 2 :Fig. 3 :
Fig. 2: The name and position of the selected stations over Egypt.

Fig. 6 :
Fig. 6: The horizontal distribution of the trend values of the seasonal and annual reanalysis TCO for the period 1979-2014.

Fig. 7 :Fig. 9 :
Fig. 7: The comulative annual mean (CAM) time series and the averaged CAM of measured TCO of the stations Cairo, Matrouh, Hurgada and Aswan

Fig. 10 :
Fig. 10: Mann-Kendall test for abrupt change of the annual reanalysis TCO of the eight stations.

Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 13 February 2017 doi:10.20944/preprints201702.0040.v1
The COV of annual, winter, spring, summer and autumn of TCO of the reanalysis (8 stations) and measured (4 stations) data is displayed in Table3.The higher values of COV at all stations occur at winter while the lowest values of COV occur at autumn.The COV of spring at all stations are greater than those in summer and annual values.The COV in annual and winter decreases gradually from the northen station to the southern station.So it can be said that the COV is a function of latitude in winter and annual values.The maximum value of COV in spring and autumn for the eight stations occurs at Matrouh, it reaches 3.58 and 1.98 respectively.Also as in winter, the values of COV in spring decrease gradually from the northern station to the southern station.In summer season the behaviour of COV at our stations is different from that in winter, spring and autumn seasons, these results are more obvious in Figure3.Figure3illustrates the horizontal distribution of COV for the annual and seasonal values of TCO.It is clear that the values of COV of winter, spring and annual decreases gradually from north to south of Egypt.The patterns of COV for summer and autumn are different from winter and spring, where we found that the summer values of COV decreases from west to east, it may be affected by the west ward oscillation of the Indian monsoon low.The lowest values of COV of autumn appears at the middle of south Egypt and also over the Red Sea, it may be associated with the northward oscillation of the inverted V-shep trough of Sudan low.The high variability over north Egypt in winter and spring is associated with the traveling main and secondary Mediterranean cyclones.

Table 5
also illustrates that both of negative and positive trend values of winter season is less than those in the other seasons.trend values of the seasonal and annual reanalysis TCO over Egypt for the period 1979-Preprints (www.preprints.org)| NOT PEER-REVIEWED |

Posted: 13 February 2017 doi:10.20944/preprints201702.0040.v1 2014
. It shows that in winter the negative trend values appears north of latitude 22 o N and in spring north of latitude 28 o N, while in summer and autumn it appears north of 30 o N.

Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 13 February 2017 doi:10.20944/preprints201702.0040.v1 2010
to 2014, in which the ozone is 1.83 DU higher than that of the whole period.The Manne Whitney test detected two abrupt changes in the ozone time series of Aswan the first one is around 2000 at the 0.004 significance level while the second one is around 2012 at the 0.176 significance level, Figure11dshows the departure curves of the ozone in Aswan station.It is shown that the departure curve fluctuates significantly, with the characteristics of two decreasing and one increasing abrupt changes since the 1984s.The period of more ozone, 2001-2011, in which the mean ozone is 3.5 DU more than that of the whole period.The first period of less ozone from 1984 to 1999, in which the mean ozone is 2.36 DU less than that of the whole period.The second period of less ozone is from 2013 to 2014, in which the mean ozone is 2.48 DU less than that of the whole period.
The studying of COV for annual, winter, spring, summer and autumn ozone indicate that higher values of COV at eight stations occur at winter while lowest values occur at summer.The COV is function of latitude in annual, winter and spring where it decreases gradually from the north to south of Egypt.The studying of the horizontal distribution of COV for the annual and seasonal values of TCO indicate that the patterns of COV for summer and autumn are different from winter and spring, where we found that the summer values of COV decreases from west to east, it may be affected by the west ward oscillation of the Indian monsoon low.The lowest values of COV of autumn appears at the middle of south Egypt and also over the Red Sea, it may be associated with the northward oscillation of the inverted V-shep trough of Sudan low.The high variability over north Egypt in winter and spring is associated with the traveling main and secondary Mediterranean cyclones.The trend analysis by least square method and Mann-kendall test for the seasonal and annual values of ozone over our stations were made.The trend analysis indicates negative trends over northern stations for both annual and seasonal time series with the greatest one at winter.The horizontal distribution of ozone trend values is negative over all Egypt in winter while it positive over middle and south Egypt in the other seasons.The long-term variability of the behavior of the annual ozone shows positive trend values in ozone are the dominant features during the period 1979-1989 at the most stations.Negative trend values in ozone are the dominant features during the period 1990-2014 at all stations.The study of the Manne Whitney test for abrupt change of the time series of TOC for the four ground based stations confirms that there is abrupt change in the ozone time series of Matrouh around 2006 at the 0.01 significance level.And there are two obvious periods; the first one is the decrease period of 1989-2005 in which the abnormal decrease years are 1998, 1999, 2002 and 2004 respectively.The second one is the increase period of 2007-2014, in which the mean ozone is 2.48 DU higher than the average of the whole period.The Manne Whitney test detected two abrupt changes in the ozone time series of Cairo the first one is around 1979 at the 0.05 significance level while the second one is around 1993 at the 0.08 significance level.The period of more ozone, 1980-1992, inwhich the mean ozone is 2.9 DU more than that of the whole period.The first period of less ozone from 1968 to 1977, in which the mean ozone is 4.49 DU less than that of the whole period.