Variability of Saharan Cyclone Tracks, I: Climatology Study


 Objectively, Saharan cyclones have been detected for the period from 1967 to 2019 using mean sea level pressure (SLP); their tracks have been specified from nearest neighbor cyclonic positions and classified into long/short tracks depending on the area of influence of the cyclones. Additionally, the detected long tracks have been objectively classified into five main routes directed generally eastward, northeastward and northward, accounting for approximately 41.6%, 19.7% and 30.4% of the total long tracks, respectively. Mainly for long tracks, three cyclogenesis areas, where more than 99% of cyclones are generated, were identified, with more than 61% generated in the Atlas region. Moreover, four far cyclolysis areas were identified, where approximately 74% of these cyclones terminated, with more than 66% of them terminating in the eastern study region. Furthermore, statistical analysis indicated that Saharan cyclones are commonly generated in the spring and summer, with ~35.3% and 46.3%, respectively. However, the highest numbers occur in spring in the northern Saharan and in summer in the southern Saharan, with ~49.1% and 57.7%, respectively. Temporally, the monthly distribution indicates that most of the cyclones moving along the five main routes are generated in warm months, namely, May to August. Approximately 85% of these cyclones have a lifespan of three days, while only 1% span more than five days.


1-Introduction
The Saharan region is classi ed as a dominant global source of aeolian and mineral dust (Prospero et al. 2002;Washington et al. 2003), and large dust plumes are transported to the surrounding regions and beyond (see, e.g., Guerzoni and Chester 1996;Heintzenberg 2009;Muller et al. 2009). The atmospheric transport of mineral dust may play an important role in climate forcing by altering the radiation balance in the atmosphere through the processes of the scattering and absorption of radiation (Tegen et al., 1997;Haywood and Boucher, 2000;Harrison et al., 2001;Sokolik et al.,2001). Consequently, the climate of the Saharan region plays an important role in the dynamic characteristics of the global climate system through, for example, the radiative properties of the region and as a source of mineral dust. Based on simulations of climate change scenarios (Schubert et al. 1998; Leckebusch and Ulbrich 2004), a northward shift in cyclone tracks was found with warming climate state. Moreover, the intraseasonal and interannual variabilities in Saharan cyclogenesis have in uenced the poleward transport of energy (Alpert et al. 1990a&b) and initiation of dust storms (Egger et al. 1995), which in turn have in uenced the features of the Saharan climate. Identifying the regional distribution of the generation, growth, translation, and decay of high-and lowpressure systems is of central importance for characterizing extratropical climates (Wernli and Schwierz 2006) and investigate climate change simulations from a synoptic perspective (e.g., Hall et al. 1994; Sinclair and Watterson 1999; Fyfe 2003; Raible and Blender 2004).
Specifying cyclone tracks can aid in determining the locations in uenced by dust storms (Trigo et al. 2002), and variations in the mean tracks, caused either by anthropogenic factors or by long-term natural variability (Blender and Schubert 2000), have strong in uences on regional climate. For example, southern Mediterranean cyclones appear as secondary lows of the large North African depression (Romero et al. 1999; Jans`a et al. 2001; Almazroui et al. 2017).
Over land during the warm season, strong sensible heating plays an important role in the genesis and preservation of cyclones, and the southward extension of cyclones deepens the upper troughs in the Lee Atlas (Campins et al. 2010). However, in the cold season (Thorncroft and Flocas 1997) show that the southward movement of the polar jet stream and its interaction with the subtropical jet can cause surface cyclogenesis in the Saharan area. In addition to the previously established factors that in uence Saharan cyclogenesis, Lee cyclogenesis likely constitutes a main initiation mechanism and plays an important role in the growth, development and variability of Saharan lows (Egger et al. 1995).
This study speci es the generated Saharan cyclones and their tracks over the whole Saharan region and throughout the year considering all seasons, not only a speci ed season or area (Prezerakos 1985 In this study, we determine the main sources, tracks and termination regions of Saharan cyclones and describe the internal structure of Saharan cyclones by studying the span time, characteristics of the sources and termination regions, and monthly/seasonal variabilities. Saharan cyclones are classi ed into long/short tracks, which represents the rst step in studying regional climate or regional climate variability.
This paper is organized as follows. Section 2 describes the data and methodology. Section 3 presents and discusses the results of the statistical characteristics of the detected Saharan cyclones. The summary and conclusions are given in the nal section.

2-Data And Methodology
Using the objective method developed by (Hannachi et al. 2011;Almazroui et al. 2015), generated cyclones and their tracks were objectively identi ed in the Saharan region, i.e., the area between longitudes 10° W-32.5° E and latitudes 17.5° N-32.5° N. The cyclone characteristics were identi ed based on the mean sea level pressure (SLP) data from the National Center for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) from 1967 to 2019. The data have a 6-hour temporal resolution and are based on a regular 2.5°×2.5° latitude-longitude grid (Kistler et al. 2001; Kalnay et al. 1996). The identi ed cyclones were used to classify the cyclogenesis, cyclolysis and affected areas in the study region.
As an initial step, the original low spatial resolutions of SLP data were transformed to a ner grid (0.5° × 0.5°) based on an established method (Pinto et al. 2005;Hannachi et al. 2011;Almazroui et al. 2015) to increase the availability of off-grid-identi ed cyclone centers and tracks. Then, brie y, the method described on (Hannachi et al. 2011;Almazroui et al. 2015), a cyclone center at a grid point was de ned considering the following conditions: The SLP value is smaller than or equal to the value at each of the eight neighboring grid points.
The pressure difference between the considered grid center and each of the neighboring eight grid points ranges from 0.8 to 3.8 hPa, as noted by Ziv et al. (2013).
The SLP value is less than 1008.8 hPa, which is adequate to detect the initial stages of the development of Saharan cyclones and less than the value used by Bartholy et al. (2006) of 1012 hPa.
Furthermore, based on previous ndings (Hannachi et al. 2011;Almazroui et al. 2015), the cyclone tracks were determined as follows: Around the rst point representing a detected cyclone center, a box containing 20 ne grid points in all directions is used to search for a new track location, i.e., the next low-pressure center.
After specifying the location of the next center, the previous step is repeated for this new location while considering the results at the previous point in the track.
If no cyclone center is detected in the subsequent three time steps or the new center is beyond 50° E or 50° N, the track is terminated, and a new track search begins using a considered cyclone center in the Saharan region.
Some important classi cation steps were applied to the detected cyclonic tracks to specify the main area in uenced by the Saharan cyclones.
1-Long/short classi cation: The detected cyclonic tracks were classi ed into either short or long-track groups. The short tracks correspond to cyclones that affected local areas and stayed within the areas where they were generated; i.e., the tracks have latitudinal ranges less than or equal to 5° and longitudinal ranges less than or equal to 10°. Only 5° of latitude is used to prevent the cyclone moving into a new climate region, because a short latitudinal distance can shift a cyclone from the Saharan region to the Mediterranean region, i.e. a different climate region; however, this shift does not appear if 10° of longitude is used. However, long tracks correspond to cyclones that greatly affect their generation areas and potentially other areas.
2-Cyclogenesis areas: In this study, only the main cyclogenesis areas are considered, i.e., the areas where most cyclones are generated.
3-Cyclolysis areas: In this study, only the main cyclolysis areas are considered, i.e. areas highly affected by the nal positions (grid point) of cyclone tracks. 4-Affected areas: This term has not been previously de ned but indicates areas that are highly affected at least once by cyclones.
To specify the main routes of long tracks, the route classi cation method used by (Hannachi et al. 2011;Almazroui et al. 2015; Almazroui and Awad 2016) was applied; brie y, the following two steps were used for classifying long tracks into the ve main routes. STEP 1: Specifying the key routes i) Each cyclone track was divided into 6-hour segments to identify the track line between consecutive time steps over the cyclone Then, the slope, or angle from the north direction, of each segment was The slope angles ranges (i.e., the difference between the maximum and minimum angles of all slopes for each track were distributed into intervals, each spanning 20°. ii) The interval that contained the maximum number of segments for all tracks was identi ed, and all tracks were then reassessed to determine which tracks belong to the maximum frequency These tracks were grouped into Filter A. iii) Tracks belonging to Filter A were regrouped, and through identi cation, these tracks were found to have at least 70% of the same segment slopes (called the A1 subgroup). The latitude and longitude information (based on the percentage of similar segment was divided to create a 0.5 o lat/long grid, and for each crossing longitude/latitude, the latitudes/longitude of the available track segments were detected. Averaging all the corresponding latitude/longitude values provided all points for cluster A1. From the entire set of tracks, the tracks similar to those in cluster A1 with a correlation of at least ≥ 70% and a latitudinal/longitudinal distance of ≤ 6° were assigned to cluster A1 and excluded from the next stage. iv) Task iii was repeated to obtain cluster A2 with a target of 60% Tracks for cluster A2 were excluded from the next stage. v) Task iii was repeated to obtain cluster A3 with a target of 50% Tracks for cluster A3 were excluded from the next stage. vi) Tasks i-v were repeated with the next highest segment frequencies to obtain Filter B and cluster B1 (C1), cluster B2 (C2), and cluster B3 (C3).
vii) The nine clusters were classi ed based on similarity, which was determined by their intercorrelations based on tracks that overlap by at least ≥ 80% and a latitudinal/longitudinal distance range of ≤ 2.5°.
Finally, only ve high-frequency key routes were identi ed to classify the long tracks. STEP 2: Specify the main routes: All tracks were classi ed into one of the ve key routes based on the following two criteria: (a) the correlation between the track and the key route was ≥ 75%, and (b) the average latitudinal/longitudinal distance of each track was less than 400/600 km from the key route.

3-Results And Discussion
By applying the above method, approximately 5851 tracks were detected, and they were divided into 3221 (or 55.1% of the whole tracks) long tracks and 2630 (or 44.9%) short tracks; i.e., most of the generated cyclones had long tracks.
The seasonal distribution of the detected tracks, as shown in Table 1, indicates that approximately 296 tracks were detected in winter, with more than 55% being long tracks; in spring, approximately 1722 tracks (or 29.43 of the total tracks) were detected, and more than 66% were long tracks. Furthermore, 2912 tracks (or 49.77% of the total detected tracks) were observed in summer, with more than 51% being long tracks, and 921 tracks were identi ed in autumn, with 46.69% long tracks and 53.31% short tracks. This seasonal distribution suggested that most of the cyclone tracks were generated in summer and spring, with ratios of 49.  The annual track distribution (Figure 1 In contrast, the months from April to September had the highest track numbers, with more tracks than the monthly average; the month with the most tracks was July, with 531 tracks, or approximately two times the monthly average (269 tracks), then coming May with 482 tracks. Of these months, September had the fewest tracks at 275, or approximately the monthly average. The results from (Trigo et al. 1999) were inconsistent with the current results, where the maximum number of cyclones occurred in May and was not high in July; this difference may be explained by the fact that their results depended on cyclones generated in only a small area, namely, the Atlas region.

3-1 Cyclogenesis areas
The spatial distribution of the Saharan cyclone regions, as shown in Figure 2-a, includes three main generation areas: the Atlas area (the area between 10° W-10° E and 15° N to 32.5° N, labeled "ATLAS" in the gure), the South Saharan area (the area between 10° E-37.5° E and 15° N to 22.5° N, labeled "STHSH" in the gure), and the North Saharan area (the area between 10° E-37.5° E and 22.5° N to 35.0°N , labeled "NTHSH" in the gure). These cyclogenesis areas were classi ed as main Saharan dust sources (Prospero et al. 2002;Israelevich et al. 2002) or as main high-frequency cyclone regions (Wernli and Schwierz 2006).
Approximately 61.87% of the long Saharan cyclone tracks (or 1993 cyclones) were generated in the Atlas area, and approximately 31.08% (or 1001 cyclones) and 7.05% (or 227 cyclones) were generated in the southern and northern Saharan areas, respectively ( Table 2)  The seasonal distribution of long tracks (Table 2) indicated the corresponding cyclones generated in the Atlas area accounted for 84 (or 4.21%), 717 (or 35.98%), 949 (47.62%) and 243 (or 12.19%) of all cyclones in the winter, spring, summer and autumn seasons, respectively. These seasonal distributions suggest that most of the cyclones in this area are generated in summer and spring (or 83.6% of the cyclones generated in the two seasons), while the lowest number of cyclones is generated in winter. Of the south-Saharan cyclones (Table 2), the distribution of the generated tracks indicates that 78.02% of cyclones occurred in spring and summer, while 5.00% of cyclones occurred in winter, an increase over the corresponding value in the Atlas area. Of the north-Saharan cyclones, as shown in Table 2, the distribution of the generated tracks suggests that 57.71% of cyclones occurred in spring, and only 7.05% occurred in autumn. Additionally, the ratio of the number of cyclones generated in the northern Sahara area in winter reached more than 56% of that in summer (twenty-nine winter cyclones compared to 51 summer cyclones).

3-2 Cyclolysis areas
The generated cyclones terminate in four main areas, as shown in Figure 2-b: the North African area, Mediterranean area, Arabian Peninsula and Red Sea area and eastern area. Approximately 2393 cyclones (74.29% from total generated cyclones) terminated in these four regions, and approximately 828 cyclones (or 25.71%) were not classi ed in any one of these areas (Table 3). The North African area (the area from 10° E to 35° E and from 22.5° N to 32.5° N, labeled "NRTREG" in Figure 2-b) was in uenced by 383 cyclones (11.89% of the long cyclonic tracks, or 16.01% of the tracks classi ed as associated with cyclolysis regions), including 9 (or 2.29%), 149 (or 38.90%), 196 (or 51.17%) and 29 (or 7.57%) in winter, spring, summer and autumn, respectively, as shown in Table 3.
The Mediterranean area (the area from 10° E to 37.5° E and from 32.5° N to 45° N, labeled "MEDREG" in Figure 2-b) was in uenced by 419 cyclones (13.01% of the long tracks, or 17.51% of the tracks classi ed as associated with cyclolysis tracks), as shown in Table 3. This area was classi ed as a cyclone center by the NH (Raible et al. 2008) and was considered a generation area for eastern Mediterranean cyclones/rain (Alpert et al. 1990a&b;Kahanaet al. 2002;Rubin et al. 2007). Any discrepancies among results are potentially because the authors considered the existence of cyclones but did not consider cyclonic tracking. These cyclonic tracks include a maximum of 189 (or 45.11%) in spring (Table 3) and a minimum of 48 (11.46%) in winter ( Table 3). The maximum number in this region was found in spring, not in summer, as in the North African area, although winter was the season with the fewest cyclones in both areas.
The Arabian Peninsula and Red Sea area (the area from 35° E to 57.5° E and from 15° N to 32.5° N, labeled "ARSREG" in Figure 2-b) was in uenced by 580 cyclones (18.01% of the long-track cyclones, or 24.24% of the tracks classi ed as cyclolysis tracks), as shown in Table 3, with a maximum number of 261 cyclones in summer (or 45.0% of the area cyclones) and a minimum number of 13 cyclones in winter (or 2.24% of the area cyclones). Additionally, 306 cyclones formed in spring and autumn (or 32.41% and 20.34% in spring and autumn, respectively). This area was previously classi ed as a generation area for Mediterranean cyclones (Alpert et al. 1990a&b) because cyclones were detected in the area, but the tracks from the initial sources were not considered.
The Eastern area (the area from 37.5° E to 60° E and from 32.5° N to 50° N, labeled "ESTREG" in Figure 2b) was in uenced by 1011 cyclones (31.39% of the long-track cyclones, or 42.25% of the tracks classi ed as being associated with cyclolysis regions), as shown in Table 3. However, this area, especially around the Black and Caspian Seas, was considered a generation region for the Middle East (Alpert et al. 1990a&b) or an active cyclonic region linked with synoptic systems over Europe and the Mediterranean (Trigo et al. 2002). A highly seasonal distribution (Table 3) was found, with 423 cyclones (or 41.84% of the area cyclones) in summer and 407 cyclones (or 40.26% of the area cyclones) in spring; the minimum number of 67 cyclones was found in winter (6.63% of all cyclones in the area). This distribution indicated that more than 82% of the cyclones in the eastern area occur in spring and summer and only 11.3% occur in autumn.

3-3 Contributions of cyclogenesis in cyclolysis areas
The Atlas area was where 1261 (or 52.7%) of the cyclone tracks that reached the main cyclolysis areas (Table 4) formed, while the South Sahara and North Sahara areas were where 920 (or 38.4%) and 212 (or 8.9%) of these cyclones originated, respectively. In detail,

3-4 Affected regions
The most affected areas, i.e., the area in uenced at least one time by any tracks, were identi ed based on long tracks, and four regions were highly in uenced by Saharan cyclones (Figure 2-c): the South Sahara (labeled STHEF in the gure), North Sahara (labeled NRTEF in the gure), Mediterranean (labeled METEF in the gure) and Arabian Peninsula (labeled ARPEF in the gure) regions.
The STHEF region (the area from 00° E to 35° E and from 15° N to 23° N in Figure 2-c) was affected by 2076 cyclones (or approximately 64.45% of long-track cyclones), as shown in Table 5, and these cyclones were divided into 61 cyclones (or 2.94% of the cyclones that affected this region), 615 cyclones (or 29.62%), 1100 cyclones (or 52.99%) and 300 cyclones (or 14.45%) in winter, spring, summer and autumn, respectively. The NRTEF region (the area from 00° E to17.5° E and from 24° N to 32.5° N in Figure 2-c) was affected by 1409 cyclones (or 43.74% of long-track cyclones), as shown in Table 5; this was the second most affected region by Saharan cyclones after the STHEF region. The seasonal effect in this region was highest in spring, with 611 cyclones, or 43.36% of the cyclones that affected the region, followed by 529 cyclones (or 37.54%) in summer.
The MEDEF region (the area from 20° E to 40° E and from 30° N to 37.5° N in Figure 2-c) was affected by 1187 cyclones (or 36.85% of the long-track cyclones), as shown in Table 5. The highest numbers of cyclones occurred in spring and summer, with 485 cyclones (or 40.86% of the cyclones that affected the region) and 446 cyclones (or 37.57%), respectively. However, the MEDEF region was weakly affected in winter and autumn, with 103 (or 8.68%) and 153 (or 12.89%) cyclones, respectively.
The ARPEF region (the area from 40° E to 50° E and from 25° N to 32.5° N in Figure 2-c) was the least affected area (Table 5); notably, only 632 cyclones (or 19.62% of the long-track cyclones) were observed in this area. Most of the cyclones were generated in summer and spring, with 274 cyclones (or 43.35%) and 252 cyclones (or 39.84%) in these seasons, respectively; thus, approximately 83.19% of all cyclones that affected the area occurred in these seasons.

3-5-1 General description of main routes
By applying the route conditions, more than 90% of the long tracks were classi ed into ve main routes (Table 6). The rst route (or the South Sahara route) passes directly eastward over Sahara and northeast over the Arabian Peninsula (AP) (Figure 3-a); this route encompassed 936 tracks, or 29.06% of all long tracks (Table 6). This route was noted for spring by (Alpert et al. 1990b).
The second route (or North Sahara route) passes directly eastward over northern Africa and the south Mediterranean, and some of the corresponding tracks are directed southeastward over the northern AP ( Figure 3-b). This route encompasses 373 tracks, or 11.58% of all long tracks (Table 6), and includes the area known as the Saharan depression or Khamasin depression (Tantawy 1964;Alpert and Ziv 1989) or the Saharan spring cyclones track (Alpert et al. 1990a&b).
The third route (or eastern Mediterranean route) passes generally northeast over Sahara and east/northeast over the Mediterranean and Middle East, and only a few tracks pass southeast over the Arabian Gulf (Figure 3-c). This route encompasses 499 tracks, or 15.49% of all long tracks, as shown in Table 6. This route is considered one of three main winter NH storm paths (Wernli and Schwierz 2006), or one of the four main passes that in uence Turkey (Karaca et al. 2000). Both the HMSO (1962) and Alpert and Ziv (1989) suggested that this route in spring contributes to approximately 43% of the North African cyclones that reach the eastern coast of the Mediterranean.
The fourth route (or the northeast Egypt route) passes northeastward over Egypt and the Middle East and southeast over the region around the Arabian Gulf (Figure 3-d). This route encompasses 136 tracks, or 4.22% of all long tracks (Table 6). This route was mentioned by (Alpert and Ziv 1989; Awad and Mashat 2014) as a track for African dust.
The fth route (or northern route) generally passes directly northward or near northward (Figure 3-e) and encompasses 980 tracks, or approximately 30.43% of all long tracks (

3-5-2 Seasonal and monthly variations in the main routes and their lifespans
The total monthly distribution of classi ed tracks, as shown in Figure 4-a, displays two high numbers of tracks, one in August, with 485 tracks, or 16.59% of the classi ed tracks, and the other in May, with 431 tracks, or 14.74% of the classi ed tracks. Furthermore, the lowest numbers of tracks were observed in November, December and January, with 34, 16 and 34 long classi ed tracks, accounting for less than 1.2% of classi ed tracks in each month. This situation contrasts with the monthly distribution of Mediterranean cyclones (the water area northern Sahara), where for example Alpert and Ziv (1989) suggested that the low numbers of cyclones in warm months was explained by the dominance of the subtropical high in warm months.
In particular, along the South Sahara route, RT1 in Figure 4-b, no tracks were observed in December or January, and the highest number of tracks occurred in July, with 217, or 23.18% of the tracks along that route.
December, with one track, was the month with the fewest tracks along the North Sahara route, RT2 as shown in Figure 4-b. The highest number of tracks occurred in April, with 96 tracks, or 25.74% of the tracks along that route, as shown in Table 6. This high number was explained (Elfandy 1940;Alpert and Ziv 1989) by the thermal gradient between the cold African continent and the warm Mediterranean water.
Generally, the seasonal distribution of route tracks (Table 6) suggests that the lowest number of tracks occurs in winter, while spring had the highest numbers for North Sahara and the eastern Mediterranean and most tracks occurred in summer along other routes.
The lifespan distribution indicates that more than 66% of tracks along the South Sahara route have lifespans between two and three days, as shown by RT1 in Figure 4-c; only approximately 16% of tracks have lifespans longer than three days, and approximately 17% of tracks have lifespans of one day or less.
For the North Sahara route, RT2 in Figure 4-c, more than 63% of tracks have lifespans between two and four days, more than 20% of tracks have lifespans of one day or less, and more than 16% of tracks have lifespans between four and 7 days.
Approximately 41% of the tracks along the Eastern Mediterranean route, RT3 in Figure 4-c, have lifespans of one day or less, more than 46% of tracks have a lifespan between two and three days, and only approximately 12% of tracks have a lifespan between four and seven days.
Most of the tracks, approximately 76%, along the northeast Egypt route, RT4 in Figure 4-c, have lifespans between two and four days, 18% of tracks have a lifespan of a day or less, and only 5% of tracks have lifespans of 5 to 6 days.
On the northern route, RT5 in Figure 4-c, more than 79% of tracks have a lifespan of two days, and more than 20% of tracks have a lifespan between three and six days.

3-5-3 Cyclogenesis and cyclolysis regions for main routes
The main area of cyclogenesis for the South Sahara route, as shown in Figure 5

4-Summary And Conclusions
The climatology of the Saharan region cyclone tracks (routes) is objectively assessed (classi ed) and discussed in this paper. The study focuses on the cyclogenesis areas over the Saharan region (10° W-32.5° E and 17.5° N-32.5° N) over the whole year during the 1967-2019 period using 6-hour sea level pressure reanalysis data derived from the NCEP/NCAR dataset. Additionally, the cyclogenesis/cyclolysis areas of the detected Saharan cyclones are determined, and the contributions of main cyclogenesis areas to the main cyclolysis areas are statically assessed. Furthermore, the main routes of more than 90% of the long-track cyclones are classi ed and described. The main results are as follows.
1. Most of the detected tracks (55.1%) were long tracks, and they affected regions far from the main Saharan sources. The highest percentages of detected long tracks were found in spring and summer, and these tracks accounted for more than 81% of all long tracks. . The distribution of the cyclolysis areas indicated that more than 74% of the long tracks terminate in four cyclolysis regions. Additionally, more than 49% of the long-track Saharan cyclones (or more than 66% of cyclones from the cyclogenesis areas) reach the Arabian-Red Sea and eastern regions.
Moreover, approximately 42% and 39% of these cyclones were generated in summer and spring, respectively (or 81% for both seasons), and only 5.7% were generated in winter.
7. The contributions of the main cyclogenesis regions to cyclolysis areas indicated that the majority of cyclones from the main cyclogenesis regions reached the eastern region. Additionally, few cyclones from the Atlas region, the South Sahara region and the North Sahara region reached the Arabian-Red Sea, the Mediterranean region and the eastern region, respectively. Additionally, the contributions of the cyclogenesis regions indicate that most of the cyclones that reach the Arabian-Red Sea region originate in the South Sahara region, and most of the cyclones that reach the Mediterranean and North African regions originate in the Atlas region. Comparable numbers of cyclones from the Atlas (approximately 45.1%) and South Sahara (approximately 40.8%) regions reach the eastern region.
. Five main routes were classi ed from the detected long-track cyclones. Only approximately 9.22% of long tracks are not classi ed into one of these ve main routes.
9. The direction of the main routes indicated that approximately 40.64% of the long tracks pass eastward and over South or North Sahara. Moreover, approximately 19.71% of the long tracks pass northeastward over the Mediterranean or Egypt, and more than 30% of the long tracks pass directly northward or nearly northward across the Sahara region. 10. The monthly distribution of classi ed routes indicated that most long-track cyclones occur in August and May, with the fewest number occurring from November to January. Nevertheless, for the easterly routes (i.e., South and North Sahara routes), the highest numbers occurred in July for the South Sahara route and in April for the North Sahara route.
11. The lifespan of routes indicated that approximately 85% of tracks have a lifespan of approximately 3 days, and less than 13% of the cyclonic tracks have lifespans of more than three days to ve days; only approximately 1% of route tracks have lifespans longer than ve days.