2.1 Data and Method
Sea level pressure (SLP) and geopotential height (500hPa) data for this study was obtained from NCEP-NCAR Reanalysis (Kalnay 1996). This dataset yields 6-hourly data which can be converted to a daily mean to correspond with the single daily weather type in the catalogue. The classification domain in this study is defined as the intersections of the parallels 35oN, 40oN and 45o N with the meridians 7.5oE, 12.5oE and 17.5oE and is centred over the Tyrrhenian Sea (central point is 40oN, 12.5oE). Values of SLP (p1-p9) were taken for the nine-point grid (Fig. 1) and used to calculate the eight circulation parameters following the equations below. The scaling factors that appear in the formulae are dependent on geographical latitude.
P = 0.0625[(P1 + P3 + P7 + P9) + 2(P2 + P4 + P6 + P8) + 4P5]
W = 0.25[(P7 + 2P8 + P9) - (P1 + 2P2 + P3)]
S = 0.653[0.25(P3 + 2P6 + P9) − 0.25(P1 + 2P4 + P7)]
D = arctan(W/S)
F = (W2 + S2)1/2
ZW = 1.056[(P7 + 2P8 + P9) - (P4 + 2P5 + P6)] – 0.951[(P4 + 2P5 + P6) - (P1 + 2P2 + P3)]
ZS = 1.305[0.25(P3 + 2P6 + P9) − 0.25(P2 + 2P5 + P8) – 0.25(P2 + 2P5 + P8) + 0.25(P1 + 2P4 + P7)]
Z = ZW + ZS
Where,
P = Surface pressure (hPa).
W = Zonal component of geostrophic (surface) wind calculated as the pressure gradient between 35o and 45oN.
S = Meridional component of geostrophic (surface) wind calculated as the pressure gradient between 7.5oE and 17.5oN.
D = Wind direction (azimuth degrees).
F = Wind speed (m/s).
ZW = Zonal vorticity component.
ZS = Meridional vorticity component.
Z = Total vorticity.
A daily weather type is then assigned following a set of guidelines (see Jenkinson and Collison 1977). These determine if the weather type is pure directional (i.e. N, NE, E, SE, S, SW, W, NW or N), cyclonic or anticyclonic (C, A) or a hybrid (CNE, CE, CSE, CS, CSW, CW, CNW, CN. ANE, AE, ASE, AS, ASW, AW, ANW, AN) and are as follows:
- The flow direction (D) uses eight compass points.
- If |Z| is less than F then an advective or pure directional type exists, determined by D (rule one).
- If |Z| is greater than 2F, then the weather type will be cyclonic, if Z > 0, or anticyclonic if Z<0.
- If F is less than |Z| and 2F, a hybrid type exists, vorticity is determined by rule three, and direction by rule one. Hybrid types are CNE, CE, CSE, CS, CSW, CW, CNW, CN, ANE, AE, ASE, AS, ASW, AW, ANW and AN.
- If F and |Z| are less than 6, an undetermined (U) weather type is reported.
Atmospheric circulation at height in the mid-latitudes is much less complex than at surface. It reflects the position of axis of the polar front and whether this exhibits a zonal or meridional configuration. Key patterns that dominate the region are positive and negative. Geopotential height data for the 500hPa surface was used to construct the upper air classification (JC500). Given the equivalent latitude it is appropriate to follow the same methodology developed by Miro et al. (2020) for the Iberian Peninsula. First, the centre of the grid is displaced five degrees to the west to capture the fact that structures at this level will move from west to east and it is an approaching upper air pattern that has an influence on lower atmospheric conditions. At this height, isobaric curvature tends to increase more than flow strength and thus the threshold for discriminating between pure cyclonic/anticyclonic and hybrid classifications can be increased from 2 (as in the original surface method) to 6. The threshold between a trough pattern (cyclonic) and pure advection is lowered from 1 to 1/3, which enables weak (but influential) troughs to be captured. In contrast, the threshold separating a ridge from pure advection is raised from 1 to 4/3 allowing strong winds, so anticyclonic curvature is considered as advection in stable conditions. Consequently the rules for classifying the upper air types (JC500) are as follows:
1. If Z > 0 and |Z| < (1/3)*F or Z < 0 and |Z| < (4/3)*F types are considered as pure advection (N, NE, E, SE, S, SW, W, NW).
2. When |Z| > 6F, the type is cyclonic C if Z > 0 or anticyclonic A when Z < 0).
3. If (1/3)*F < |Z| < 6F and Z > 0 (cyclonic) a trough above is defined (N, NE, E, SE, S, SW, W, NW).
4. If (4/3)*F < |Z| < 6F and Z < 0 (anticyclonic) a ridge above is defined (N, NE, E, SE, S, SW, W, NW).
5. If F < 6 and |Z| < 6 the type is unclassified (U).
2.2 Results of the surface synoptic classification (JCsfc)
A 1% sample of days is taken from the resulting 27,028-day catalogue and cross referenced with the respective 1200 UTC and 1800 UTC SLP charts it is clear that the technique captures the characteristics of surface circulation effectively. The marked seasonality of Central Mediterranean atmospheric circulation is reflected by differences in the intra-annual occurrence of the principal JCsfc types. Seasonal circulation change is a consequence of the poleward shift in the axis of the polar jet in the summer and a reduction in flow strength due to the decrease in the hemispheric temperature gradient. As the polar jet migrates south in winter months there is a strong effect on circulation patterns at surface and a higher variability in weather type occurrence.
Days classified as either ‘U’, ‘C’ or ‘A’ types dominate the surface catalogue (Table 1). Unclassified ‘U’ types, representing little pressure gradient and SLP close to the annual mean, are most common and are a main persistent feature of summer circulation (53% of all days). It is often the case in summer months that long runs of ‘U’ types have an incidence of single ‘C’ type or ‘A’ type days, yet the vorticity index, z, (upon which the type is determined) indicates these patterns are often very close to the threshold of ‘6’ (-6)[1] and thus represent only very weak positive (or negative) vorticity. In the summer period more intense cyclonicity[2], for instance, occurs for periods of 2–3 days once every three years on average.
Table 1 Frequency (percentage days of occurrence) of JCsfc during the full all year classified period (1948-2021) and the winter (DJF) season but also the sample period (2001-2021) all year and the winter season (DJF) used in the air pollution analysis. Only types with more than one percentage of occurrences are listed.
|
Days
|
C
|
U
|
W
|
A
|
CW
|
E
|
CE
|
AW
|
SW
|
CNE
|
1948-2021
|
27028
|
33.27
|
33.65
|
4.43
|
13.05
|
1.71
|
2.49
|
2.25
|
1.71
|
|
|
1948-2021
Winter
|
6677
|
47.40
|
17.39
|
8.21
|
4.63
|
3.31
|
3.82
|
3.47
|
1.27
|
1.48
|
1.51
|
2001-2021
|
7668
|
37.14
|
35.08
|
4.11
|
8.69
|
1.81
|
2.26
|
3.12
|
1.17
|
|
|
2001-2021
Winter
|
1895
|
49.5
|
17.41
|
7.92
|
4.43
|
3.17
|
2.85
|
4.01
|
|
1.11
|
1.69
|
Classified ‘C’ types are also frequent in the JCsfc series at over 33% of days and exhibit considerable annual variability from 19% of days in 1948 to 44% in 2018. ‘C’ types include all surface cyclonic flow and therefore comprise patterns with gentle cyclonic curvature of the surface isobars which can be a response to summer heat, inducing surface ‘thermal low’ circulation, or shallow upper air troughs (Fig. 2a) and more active storm events (Fig. 2b). ‘C’ patterns in the area arise due to travelling depressions from the Atlantic (Fig. 3a) however it is more common that cyclonicity develops within the region itself (Fig. 3b). The Gulf of Genoa, west of Italy, is an important zone of cyclogenesis (low pressure formation) with these systems then travelling eastwards. The Alps and other coastal topography are the principal factor for cyclones forming here due to orographic lift. Furthermore, these barriers can also hinder eastward moving upper level troughs encouraging cyclones to strengthen (Trigo et al 2002). Cyclogenesis is more prevalent in late autumn and winter by the strong baroclinic conditions arising due to the enhanced maritime-continental temperature gradient at this time of year. ‘C’ types are most common in the winter (47.75%) and autumn months (35.68%).
Anticyclonic types are infrequent compared to their occurrence in northern European catalogues. There is a distinct seasonality with many more incidences in spring (27% including all hybrid types) than other times of the year (winter is less than 8%). ‘W’ and ‘E’ are the most common directional weather types but are noticeably less so in the summer under the slack pressure gradients. More than half of the 27 weather types together occur on less than 3% of all days although there is more representation by the less frequent types in the winter months.
The decrease in the number of annual occurrences of ‘A’ type days (and concomitant increase in ‘C ‘ days) is the principal temporal trend to emerge from the catalogue (Fig. 4). The annual number of ‘C’ types has increased in the time period at a rate of 5.9 days per decade (R2 = 41.6%, F = 49.1 = p < 0.001) and there has been an even clearer decline (8.8 days per decade) in ‘A’ types (R2 = 60.4%, F = 105.25 p < 0.001). The vorticity component (z) demonstrates a shift towards cyclonicity (an increase in mean annual z) in this region in the second half of the twentieth and first two decades of the twenty-first century. It is notable that this increase happens in all seasons but not in the winter months. This is contrary to future modelled scenarios for circulation change (e.g. Otero et al. 2018) that suggest greater anticyclonicity across the Mediterranean region however the classified domain is at a smaller scale within the region, is a centre of cyclogenesis and therefore likely to be highly responsive to more localised disturbances. On the other hand, Otrero et al. (2018) do report that that the set of models participating in the Coupled Model Intercomparison Project Phase 5 (CMIP5) did underestimate the number of cyclonic days currently occurring in Southern Europe.
2.3 Results of the upper air synoptic classification (JC500)
A JC500 series is generated using the modified procedure of Miro et al. (2020). This is manually validated against a 1% sample of 500 hPa charts and it is seen to effectively capture the flow at this level. As expected it is typified by westerly zonal advection, which can be subdivided into anticyclonic, cyclonic or pure westerly flow (Table 2). The dominant westerly configuration is a trough pattern over the area. Although all possible JC types occur, hybrids and any type with easterly or southerly flow components are much less numerous (for instance, ‘SE’ types only occur on 20 days (0.07%) in the 27,028-day catalogue). There are significantly less ‘U’ types (6.23% compared to 33% at surface) at this height. Given the small number of hybrid weather types they can be grouped together as Ridge Westerly (RW) and Ridge Easterly (RE) or Trough Westerly (TW) and Trough Easterly (TE). Across the year the overall flow is relatively consistent although there is a slight increase in easterly flow in the winter.
Table 2
Frequency (percentage of days of occurrence) of JC500 during the full all year classified period (1948–2021) and the winter (DJF) season but also the sample period (2001–2021 all year and the winter season (DJF) used in the air pollution study. Only types with more than one percentage of occurrences are listed. .
|
Days
|
TW
|
TE
|
RW
|
RE
|
C
|
A
|
W
|
NW
|
SW
|
U
|
N
|
NE
|
1948–2021
|
27028
|
39.12
|
3.28
|
9.78
|
1.18
|
3.45
|
1.00
|
27.47
|
4.97
|
1.68
|
6.23
|
|
|
1948–2021
Winter
|
6677
|
37.90
|
5.62
|
10.14
|
2.31
|
4.51
|
1.38
|
20.42
|
7.39
|
1.18
|
5.24
|
1.56
|
1.17
|
2001-21
|
7668
|
41.42
|
3.91
|
10.07
|
1.51
|
3.71
|
1.03
|
27.06
|
5.05
|
1.75
|
6.07
|
|
|
2001-21
Winter
|
1895
|
41.88
|
5.66
|
10.21
|
2.22
|
4.49
|
1.32
|
19.46
|
8.14
|
1.37
|
5.02
|
2.49
|
1.27
|
Surface circulation is strongly influenced by upper air flow particularly on westerly JCsfc days which almost entirely (98%) reflect the dominant direction aloft (either TW, RW and W). Anticyclonic and easterly surface types are less associated with zonal patterns aloft. For example, under surface A types, the relatively uncommon ridge easterly flow is present on 15% of days.
The JC500 catalogue gives synoptic information on the many instances where surface patterns are unidentifiable (‘U’ types). Table 3 summarises the variability in upper atmospheric circulation that accompanies surface U types. Although all upper types can occur, some are over-represented compared to the overall frequency of incidence. ‘U’ is more associated with negative rather than positive vorticity pattens aloft but not exclusively. JC500 indicates different mechanisms can operate away from the surface which are implicated in the behaviour of a circulation-related environmental variable on these days. Surface C types, the second most common pattern in the JCsfc series, is strongly associated with dominant W and TW flow, but can also be accompanied by ridge patterns aloft (11% of occurrences) suggesting that many of these are shallow features. Unlike at surface there is no statistical trend in the annual or seasonal frequency of any of the principal weather types.
Table 3
Upper atmospheric circulation accompanying surface U types. The ratio between JC500 type is associated with surface ‘U’ to its frequency for all types with bold values representing above average occurrences.
|
A
|
C
|
NW
|
RE
|
RW
|
SW
|
TE
|
TW
|
U
|
W
|
% occurrence under JCsfc U types
|
1.14
|
1.96
|
3.60
|
1.76
|
12.20
|
1.90
|
1.92
|
31.19
|
8.54
|
34.48
|
Ratio occurrence under weather type: occurrence under all types
|
1.18
|
0.57
|
0.72
|
1.13
|
1.30
|
1.13
|
0.51
|
0.81
|
1.37
|
1.26
|