The Synoptic Climatology of Winter and Summer Air Pollution in Glasgow, Scotland

doi:10.21203/rs.3.rs-5031381/v1

An air mass type approach was employed to investigate the dependence of winter and summer pollutant concentrations at two sites in central Glasgow on weather conditions over the period from 2005 to 2020. A two-stage synoptic typing process of principal component analysis (PCA) and cluster analysis was used to establish the air mass types. Daily means of seven surface meteorological variables calculated from hourly observations were entered into P-mode PCA. Three (two) principal components explained 72.2% (60.2%) of the variance in winter (summer). Days with similar weather conditions were grouped to produce six clusters (air mass types) for both winter and summer. There were statistically significant (p < 0.05) differences in daily mean O₃, PM₁₀, NO_x, NO₂ and CO between the six air mass types in both seasons. Apart from PM₁₀, variations in pollutant concentration with air mass type were more pronounced in winter than in summer. In winter, the highest mean NO_x, NO₂ and PM₁₀ concentrations occurred under the continental air mass, which is usually associated with high pressure, calm conditions limiting dispersion and the transport of pollutants from Continental Europe on a southeasterly wind. Furthermore, most breaches of air quality guidelines occurred in winter for NO₂ and PM₁₀ occurred under that air mass type. Pollutant concentrations are generally lower in summer (apart from O₃). Poor summer air quality was found to occur disproportionately under stable anticyclonic tropical and tropical continental air mass types.

Britain was the first country to undergo the Industrial Revolution, with urbanisation and the growth of heavy industry leading to a marked rise in air pollution. By the mid-19th century, smoke levels in Glasgow were nearly as high as in London (Stradling and Thorsheim, 1999). Glasgow is known for its industrial legacy of coal mining, iron and steel, and shipbuilding. The city’s historical dependence on soft coal for fuel led to high levels of black smoke and poor air quality.

Whilst smaller in area than Edinburgh, Glasgow has the highest population of any Scottish city (635,130 in 2021). It is currently affected by a multitude of pollutant sources from combustion, industrial processes, energy generation and transportation. In particular, the M8 urban motorway slices through the inner suburbs of Glasgow. In 2004, Glasgow City Council proposed an Air Quality Action Plan, which aimed to reduce nitrogen dioxide (NO₂) and particulate matter (PM₁₀) concentrations in the city centre. Glasgow City Council (2008, p 12) attributed 73% of PM₁₀ emissions to road traffic, especially diesel engines. With road traffic also the largest contributor to the city’s NO₂, a Low Emission Zone (LEZ) was introduced in the city centre over a five-year period from December 2018 (Lee et al., 2019). The LEZ initially applied to buses only. Since 1st June 2023, all types of vehicles entering the LEZ must meet emission standards. Non-compliant vehicles receive a £60 penalty fine for entering the LEZ (as of June 2024). Poor air quality is the greatest environmental threat to human health (WHO, 2021). Lee et al. (2009, p 421) found “convincing evidence for an association between long-term exposures to PM₁₀ and NO₂ and respiratory hospital admissions in…Greater Glasgow.” Across Scotland as a whole, 2500 premature deaths annually have been attributed to high pollutant concentrations (IQAir, 2023).

Two factors determine pollutant concentrations at a particular location: the nature of the emissions and the state of the atmosphere (Oke, 1987). Air stability and wind play a major role in the horizontal and vertical dispersion and transportation of pollutants. The poorest air quality tends to be associated with a stable atmosphere when temperature inversions can occur and suppress vertical mixing.

Previous studies have utilised a synoptic typing approach to group multiple meteorological variables into air mass types and identify synoptic-scale situations that are favourable for poor and good air quality. Temperature and moisture are the most important variables to analyse in order to identify the correct air mass type (Kalkstein and Corrigan, 1986). Cloud cover, mean sea-level pressure (MLP), wind speed and wind direction are also important variables that differentiate between the different air mass types. Principal component analysis (PCA) and cluster analyses are often used to identify the air mass types. Applications of the synoptic typing approach in air quality studies include the work of Shahgedanova et al. (1998) in Moscow; and Cheng and Lam (1999) in Hong Kong.

Hodgson and Phillips (2021) investigated the relationship between air mass types and daily mean pollutant concentrations in Birmingham, UK. This study considered ozone (O₃), PM₁₀, nitrogen oxides (NO), NO_x, NO₂ and carbon monoxide (CO) over 15 winters and 15 summers from 2000 to 2015. High pollutant concentrations were associated with anticyclonic types. The tropical (polar) continental air mass type was most likely to produce extremely high concentrations in summer (winter). Apart from O₃, pollutant concentrations were higher in winter than in summer. The passage of the tropical continental airmass over mainland Europe’s industrial cities further elevates the concentrations (King and Dorling, 1997). Local air quality breaches are thus related more broadly to pollutant transport across national boundaries. McGregor and Bamzelis (1995) found that cyclonic westerlies favoured the best air quality in Birmingham because such flows have fewer pollution sources and the higher wind speeds result in greater dispersion.

Buchanan et al. (2002) researched the influence of weather types on the concentrations of black smoke and PM₁₀ in Edinburgh (Scotland’s capital city) from 1981 to 1996. They found elevated pollutant levels to be associated with anticyclonic, southerly and south easterly weather types.

Despite having the third largest population of any British city after London and Birmingham, there has been minimal research into relationships between weather and air quality in Glasgow. Moreover, the synoptic typing approach has not been applied to the city to investigate these associations. In the light of this, the aim of this paper is to assess how air mass type affects pollutant concentrations in Glasgow. The objectives are as follows:

to derive winter and summer air mass types;
to establish whether there is a relationship between air mass type and pollutant concentrations; and
to determine whether days with poor air quality are disproportionately associated with a certain air mass type.

This paper is structured as follows. Section two describes the collection and analyses of data. Results are described in section three and then discussed in section four. Conclusions are drawn in section five.

Seven meteorological variables were obtained from the Centre of Environmental Data Analysis, a repository of UK Met Office data, for Bishopton (18 km west of Glasgow City Centre; altitude = 59 m): MSLP, air temperature (TEMP), wind speed, wind direction, relative humidity (RH), cloud cover (CLOUD) and rainfall (RAIN). This is the closest station to the city centre (figure 1) that has values available for all seven variables from 2005 to 2020. The distance between Bishopton and Glasgow City Centre is acceptable here because air masses typically cover areas greater than hundreds of square kilometres (Kalkstein and Corrigan, 1986). Hourly readings were abstracted for the winter (DJF) and summer (JJA) months. Apart from the wind direction, daily means were then calculated by averaging the 24 hourly readings. Across the 16 winters and summers, there were 81 days with all or some of their hourly observations missing. Winter had more days (5.1%) than summer (1.0%) with missing values.

Daily mean concentrations (µgm^-3) of O₃, PM₁₀, NO_x, NO₂ and CO in winter and summer were abstracted from DEFRA’s air quality archive for two urban background sites in Glasgow City Centre: Glasgow Centre and Glasgow Townhead (figure 1). Two monitoring sites were used because one site did not record all five pollutants over the complete 15-year period. Glasgow Centre was used from 1st January, 2005 to 14th August, 2012. Glasgow Townhead was used from 7th October, 2013 to 31st December, 2020; CO was not measured at this site. Thus, no readings were available for part of summer 2012, and the whole of the winter of 2012-3 and the summer of 2013. When averaged across all available pollutants, the percentage of missing days at Glasgow Centre and Glasgow Townhead were 11.4% and 7.3% respectively. Given that the records from the two sites do not overlap temporally, it is impossible to combine them into one homogeneous time series. This means that separate analyses were performed for each site.

The Glasgow Centre station (altitude = 5 m) is in St Enoch Square (figure 2). This is a pedestrianised area of the city centre; it is an open area that is surrounded on three sides by shops and offices. The nearest road is 10 m from the monitoring station; the busy commercial road of Argyle Street is approximately 20 m to the north of the site. The Glasgow Townhead station (altitude = 40 m) is located to the east of the city centre in a 1960’s redevelopment area between Kennedy Street and St Mungo’s Avenue. The site is surrounded by gardens, and a mixture of low-rise and high-rise housing (figure 3). The M8 motorway runs from west to east approximately 360 m north of the monitoring station. The distance between the Glasgow Centre and Glasgow Townhead sites is 1.2 km.

Principal component analysis and cluster analysis were used to derive the winter and summer air mass types. Given that wind direction is a circular variable, it is not possible to calculate mean daily direction from the hourly observations. Consequently, zonal (U) and meridional (V) wind speeds were calculated from the hourly wind speed and direction observations, which were then averaged at the daily timescale. The skew of the rainfall distribution was reduced by taking the common logarithm. The seven variables entered into the PCA were thus MSLP, TEMP, U, V, RH, CLOUD and Log₁₀RAIN. The associations between the variables differ seasonally (e.g., high pressure tends to lead to colder winters but warmer summers), so separate PCAs were run for winter and summer. PCA is a data reduction technique, which replaces the original variables with a set of linearly independent principal components (PCs). The first few PCs explain most of the variance in the data set. The PCA was run in P mode because this mode identifies groups of variables that are highly correlated over time. The input variables are measured in different units and have different variances. This meant that the PCA was performed on the correlation matrix because this standardises the variance and ensures that each variable has the same chance of loading strongly on to PC 1. Using the co-variance matrix would have been flawed because it would merely reveal which variables had the largest variance rather than the real association between the variables. It is easier to interpret the PCA when each variable is strongly related to only one PC. This is called simple structure. Rotating the PC axes sometimes improves the degree of simple structure. During the winter, the rotated component matrix (Varimax orthogonal rotation) maximised simple structure and accordingly was used. The unrotated matrix was used in summer. PCs with eigenvalues greater than one were retained for further analysis because these explain more variance than one variable (Kaiser, 1960).

The PC scores from the significant PCs (eigenvalues > 1) were then entered into Ward’s clustering method (Ward, 1963). The purpose of clustering is to group together those days with similar weather conditions and air mass type. The squared Euclidean distance was used as the similarity measure.

The parametric independent samples one-way analysis of variance (ANOVA) was used to assess whether pollutant concentrations vary with air mass type. A higher Snedecor’s variance ratio (F) indicates a greater difference. The pollutants (apart from O₃) were logged to the base ten before conducting the ANOVA to remove the distributions’ pronounced positive skew. Whilst a statistically significant ANOVA result (p < 0.05) indicates that the air mass types differ as a whole, it does not mean that one type shows a significant difference from all the other types. This information came from the post hoc Tukey test, which identifies pairs of air masses with significantly different pollutant concentrations.

Standardised (Z) scores were used to identify days with poor air quality. Three thresholds were used: high (1.75 < Z < 2), very high (2 < Z < 3) and extremely high (Z > 3). For each pollutant, the Chi-Square test was used to ascertain whether days with high concentrations occur disproportionately under certain air mass types.

3.1 Principal component analysis

In winter, three PCs (eigenvalues > 1) account for 72.15% of the variance. PC 1 (explained variance = 30.8%) has a strong positive relationship with CLOUD, U and RH; and a moderate positive association with TEMP (table 1a). A westerly wind (U > 0) in winter brings warmer but cloudier air. Days with high positive PC 1 scores have strong westerly winds, high cloud cover and high RH, whereas high negative PC 1 score days have an easterly flow. PC 1 is thus a zonal wind component. PC 2 (explained variance = 21.8%) groups together MSLP and Log₁₀RAIN, which have a strong inverse relationship, as wetter conditions are associated with lower pressure.

Table 1 Principal component loadings for (a) winter and (b) summer

(a)

	PC 1	PC 2	PC 3
MSLP	-0.04	-0.93	0.04
TEMP	0.53	-0.03	0.73
U	0.74	0.14	0.09
V	0.27	-0.14	-0.85
RH	0.67	0.18	-0.18
CLOUD	0.81	0.08	0.05
Log₁₀RAIN	0.39	0.76	0.26

(b)

	PC 1	PC 2
MSLP	-0.71	0.25
TEMP	-0.28	0.73
U	0.51	0.51
V	0.19	0.64
RH	0.86	0.10
CLOUD	0.79	-0.06
Log₁₀RAIN	0.83	-0.03

There are only two significant PCs (eigenvalues > 1) in summer, which collectively explain 60.2% of the variance. PC 1 (explained variance = 41.9%) has a strong positive association with RH, Log₁₀RAIN and CLOUD; and a negative relationship with MSLP and TEMP (table 1b). PC 1 describes the two dominant weather conditions experienced during the summer: cool and wet with low pressure, or warm and dry with high pressure. PC 2 (explained variance = 18.3%) has a moderately strong relationship with TEMP and V, and a moderate positive association with U. PC 2 captures the co-variability between south-westerly (U > 0 and V > 0) winds and warmer temperatures.

3.2 Winter air mass types

The properties of the six air mass types are now described by identifying the common features of the surface synoptic charts on a sample of days in each cluster (figure 4 and table 2).

Table 2 Arithmetic means of the meteorological variables for the six winter air mass types for Glasgow from January 2005 to December 2020

Air mass type	Number of days	TEMP (°C)	MSLP (hPa)	U (Knots)	V (Knots)	RH (%)	CLOUD (Oktas)	RAIN (mm)
1	298	7.4	1004.1	2.4	-2.1	89.5	7.0	11.3
2	159	4.2	992.6	0.5	-2.8	86.3	5.8	9.2
3	200	3.5	1016.0	-1.5	-2.3	80.0	3.5	0.8
4	325	3.7	1008.0	1.4	1.0	92.2	7.1	5.0
5	237	0.9	1016.1	-0.2	0.8	85.2	4.8	0.3
6	124	7.7	1027.9	1.4	-2.4	88.4	7.1	0.2

Type 1: Returning Polar Maritime

The source region of this air mass type is Greenland and Canada. Despite its cold source, this air mass is the second warmest because there is a mild south-westerly wind immediately upwind of Britain. The long oceanic fetch means that this is the wettest air mass type.

Type 2: Arctic Maritime

This air mass originates from the North Pole and the Arctic Ocean. Its shorter passage over the sea results in cooler conditions than the polar maritime type. This air mass has the lowest mean pressure and so produces high levels of precipitation, with snow and hail showers common over northern Scotland.

Type 3: Polar Continental

This air mass originates from Eastern Europe and Russia. It is cold and dry. Thermal and moisture properties vary depending on the length of its sea passage. It tends to be associated with relatively high pressure.

Type 4: Tropical Maritime

This is the most commonly occurring air mass type. It originates around the Azores, with south-westerly winds bringing mild and moist air to Scotland. Overcast and drizzly conditions are common, especially near western coasts. This air mass type brings moderate amounts of rainfall.

Type 5: Continental

This is the coldest air mass type (mean temperature = 0.9^oC). The source region is eastern Europe and Russia. This air mass type tends to be cooler than polar continental (type three) because the path is largely over cold Continental Europe. Its short sea track over the southern North Sea means that little warming takes place.

Type 6: Stable Anticyclonic Westerly

This air mass has the highest mean pressure of the six clusters. High pressure centred to the south-west of Ireland results in mainly dry conditions. The west to northwesterly flow around the northern periphery of the high brings in air from the Atlantic resulting in mild conditions.

3.3 Summer air mass types

The mean properties of the six air mass types are given in table 3.

Table 3 Arithmetic means of the meteorological variables for the summer winter air mass types for Glasgow from 2005 to 2020

Air mass type	Number of days	TEMP (°C)	MSLP (hPa)	U (Knots)	V (Knots)	RH (%)	CLOUD (Oktas)	RAIN (mm)
1	374	13.9	1007.1	1.6	0.2	89.6	7.1	9.1
2	319	13.1	1009.4	-0.1	-2.2	82.9	6.4	3.3
3	244	14.7	1016.8	0.7	-0.4	79.5	5.6	0.7
4	274	13.2	1016.1	-1.5	-2.7	74.5	4.3	0.2
5	156	16.5	1016.6	1.4	1.2	80.4	5.5	0.8
6	75	18.2	1023.9	-0.1	0.5	71.5	1.9	0.0

Type 1: Tropical Maritime

The most commonly occurring summer air mass type brings south-westerly winds from the Atlantic Ocean (figure 5). Frequent frontal depressions result in this being the wettest air mass type by a long way. The moderating effect of the ocean also results in relatively cool conditions.

Type 2: Arctic Maritime

This is the coldest air mass type (mean temperature = 13.1^oC) because its source region is the Arctic Ocean. Despite being the second wettest air mass type, the shorter sea track and lower temperatures mean that the arctic maritime air mass is much drier than the tropical maritime air mass.

Type 3: Returning Polar Maritime

This air mass originates from Northern Canada and Greenland. Due to its longer sea track over the North Atlantic, it is warmer than the Arctic Maritime air mass by an average of 1.6^oC. This cluster is relatively dry, with a mean daily rainfall total of 0.7mm.

Type 4: Polar Continental

This air mass usually originates from Scandinavia and has a comparable mean temperature to the Arctic Maritime air mass. This cluster leads to dry conditions, with a mean precipitation total of only 0.2mm.

Type 5: Stable Anticyclonic Tropical

High pressure centred over Britain or northern France leads to gentle westerly winds This air mass has relatively high cloud cover (mean = 5.5 oktas) but is usually dry (mean daily rainfall = 0.8mm).

Type 6: Tropical Continental

High pressure centred to the east of Great Britain leads to a south-easterly wind importing hot and dry air from Continental Europe or even North Africa. This is by far the warmest air mass. It also has a lower cloud cover, relative humidity and rainfall than the other five clusters. It is the least common cluster occurring on only 75 out of 1442 days.

3.4 Variability in pollutant concentration with air mass type

The F ratios for both winter and summer are all greater than one signifying that there is more variance between than within the airmass types. During the winter, the pollutants at Glasgow Centre and Glasgow Townhead all show statistically significant differences between air mass types at the 99.9% confidence level (table 4). In summer, some of the differences are only significant at the 0.05 level. This seasonal difference is most apparent for NO₂ at Glasgow Centre, where the winter and summer F ratios are 81.12 and 3.07 respectively.

Table 4: Results of the one-way analysis of variance (ANOVA) to compare pollutant concentrations between air mass types. The logarithm to the base ten of the pollutant concentration was used for the ANOVA.

Station	Pollutant	Winter F	Winter p	Summer F	Summer p
Glasgow Centre	O_₃	87.97	<0.001	11.64	<0.001
	PM₁₀	18.26	<0.001	31.88	<0.001
	NO_x	66.99	<0.001	2.38	0.037
	NO₂	81.12	<0.001	3.07	0.009
	CO	33.42	<0.001	3.12	0.009
Glasgow Townhead
	O_₃	81.81	<0.001	21.78	<0.001
	PM₁₀	18.15	<0.001	38.23	<0.001
	NO_x	59.25	<0.001	2.97	0.012
	NO₂	54.48	<0.001	3.26	0.006

The differences between the air mass types are displayed visually in figure 6. In winter, air mass type five (continental) has the highest average concentrations for all pollutants apart from O₃. During the summer, concentrations tend to be lower, with the differences between the clusters being less pronounced. However, air mass type six (tropical continental) tends to have slightly higher concentrations.

The results of the post hoc Tukey tests are presented in the difference matrices (figures 7 and 8). The presence of a pollutant in a grid square indicates that there is a statistically significant difference in concentrations between those two air mass types at the 0.05 level.

The pollutant that appears most frequently in the winter matrix for Glasgow Centre (figure 7a) is NO_x, which is significant in 13 out of 15 pairwise comparisons. At Glasgow Townhead (figure 7b), O₃appeared most frequently (12 out of 15). Considering both sites, NO_xhas the greatest number of statistically significant differences between pairs of air masses (24) followed closely by NO₂ and O₃ (both 23). The variation in NO₂ and NO_xcan be clearly seen in figure 6. During the winter, PM₁₀ has the lowest number of statistically significant differences (Glasgow Centre = 7; Townhead = 8). Air mass type five (continental) has the largest number of significant differences with other air mass types for both Glasgow Centre (22 entries) and Townhead (16 entries). Air mass type three (polar continental) has the lowest number of entries (15) at Glasgow Centre. For Townhead, air mass types four (tropical maritime) and six (stable anticyclonic westerly) both have the fewest number of entries (12).

There are fewer significant differences between pairs of air masses in summer compared to winter. In summer, PM₁₀ has the largest number of statistically significant pairs (11) at Glasgow Centre (figure 8a). At Townhead (figure 8b), PM₁₀and O₃ both have the largest number (9). There are far more significant pairs for NO₂ and NO_x combined at Centre (8) than Townhead (1). CO only appears once in the Centre matrix. Air mass type six (tropical continental) has the largest number of significant differences at both Centre (15) and Townhead (9) stations. Arctic Maritime (air mass type number 2) shows the smallest number of significant differences with other air mass types (7) at Glasgow Central; arctic maritime and polar continental (air type number 4) jointly have the smallest number of significant differences at Townhead.

3.5 Comparing the occurrence of days with extreme pollutant concentrations across air mass types

The thresholds for extreme pollution days are displayed in Table 5. The Chi-Square test was used to test the null hypothesis that the occurrence of high concentrations does not vary with the air mass type. For winter, the differences are all significant at the 99% confidence level (table 6). Chi-Square tests were not performed in the summer. This is because the lower frequency of days with high concentrations violated the 20% rule for expected counts. Therefore, exceedance of the thresholds is more likely in winter than in summer. O₃ is an exception to this rule and exceeded the thresholds on six days in both seasons at Glasgow Centre and Townhead.

Table 5: Daily mean thresholds for exceptionally high, very high and high levels of pollutant concentration determined using standardised (Z) scores. All units are μgm^-3.

Pollutant	Exceptionally High Z score > 3	Very High 2 < Z Score ≤ 3	High 1.75 < Z Score ≤ 2
O₃ Centre	X > 85	68 < X ≤ 85	64 < X ≤ 68
PM₁₀ Centre	X >52	41 < X ≤ 52	39 < X ≤ 41
NO_x Centre	X > 298	222 < X ≤ 298	203 < X ≤ 222
NO₂ Centre	X > 95	75 < X ≤ 95	70 < X ≤ 75
CO Centre	X > 0.74	0.58 < X ≤ 0.74	0.54 < X ≤ 0.58
O₃Townhead	X > 94	77 < X ≤ 94	72 < X ≤ 77
PM₁₀ Townhead	X > 31	24 < X ≤ 31	23 < X ≤ 24
NO_x Townhead	X > 163	121 < X ≤ 163	111 < X ≤ 121
NO₂ Townhead	X > 64	51 < X ≤ 64	48 < X ≤ 51

Table 6: Chi-Square test results comparing the frequency of days with high concentrations in winter. Chi-Square tests were not performed for O₃ because more than 20% of the cells have expected frequencies of less than five.

Station	Pollutant	χ²	p
	O₃	-	-
	PM₁₀	16.3	<0.001
Glasgow	NO_x	67.4	<0.001
Centre	NO₂	33.2	<0.001
	CO	37.9	<0.001
	O₃	-	-
Glasgow	PM₁₀	17.8	0.001
Townhead	NO_x	27.6	<0.001
	NO₂	62.5	<0.001

The World Health Organisation’s (WHO) recommended guideline (2021) for daily mean NO₂ is 25μgm^-3. This was exceeded on 479 and 391 days at Glasgow Centre and Townhead respectively in the winter compared to only 77 and 10 days during the summer. For PM₁₀, the guideline is 45μgm^-3. This was exceeded on 73 and five days at Centre and Townhead respectively during the summer. For winter PM₁₀, 32 and four PM₁₀ breaches occurred at Centre and Townhead respectively. Across the 15 years, the WHO’s O₃ guideline was exceeded twice during the summer but never in the winter.

During the winter, the largest number of extreme days (‘high’, ‘very high’ and ‘exceptionally high’ days combined) occurred under air mass type five (figure 9ab), which is continental. This is particularly the case at Glasgow Centre. The pollutant with the greatest number of extreme days under this air mass type was CO at Glasgow Centre (46) and NO₂ at Townhead (38). Summer had very few extreme days, with only O₃ and PM₁₀ reaching the ‘high’ threshold at both sites (figure 9cd). These extreme summer days occurred during polar maritime, stable anticyclonic tropical and tropical continental air mass types.

3.6 Air mass persistence and pollutant concentrations

As poor air quality in winter is disproportionately caused by air mass type five (continental), it is instructive to assess whether pollutant concentrations are related to the duration of this air mass. In other words, are concentrations higher on a continental day when it is preceded by one or more days with the same air mass type? One-way independent samples analysis of variance yielded highly insignificant results across all pollutants at both sites (p > 0.5). All F ratios were less than one. For instance, mean NO₂ concentrations vary only slightly with sequence length. At Glasgow Centre, the inter-quartile range is relatively stable, with median values between 50 and 70µgm^-3regardless of the persistence of the air mass.

4.1 Principal component analysis

Whilst principal components have not already been derived for Glasgow, it is possible to compare this paper’s findings with PCs in other British cities. In winter, PC 1 is a zonal wind component (table 1), which encapsulates the fluctuation in weather conditions between strong westerlies (mild and damp) and easterlies (cooler and clearer). This mode of variability was only PC 2 in Hodgson and Phillips’ (2021) Birmingham’s study. This indicates that the zonal wind speed is more strongly correlated with more weather variables in Glasgow compared to inland Birmingham. Glasgow’s winter PC 2 captures the logical co-variability between MSLP and rainfall. Whilst warmer winters are associated with higher rainfall at Glasgow, the opposite is the case during the summer. Summer PC 1 describes the Atlantic Ocean’s influence over the city’s weather, with a greater Atlantic influence (stronger zonal wind) resulting in lower pressure, more rainfall and cooler temperatures. Summer PC 2 shows the strong co-variability between warmer temperatures and a southerly wind.

4.2 Air mass types

The clustering process produced six air mass types for both winter and summer (figures 4 and 5). Previous classifications have also divided the days into six air mass types such as Hodgson and Phillips’ (2021) comparative Birmingham study and McGregor et al.’s(1999) winter analysis in the same city. The most common air masses in winter are tropical maritime (325 days) and returning polar maritime (298 days) (table 2). Tropical maritime is also the most frequent air mass type during the summer (374 days) (table 3). This highlights the dominance of south-westerly winds in Britain’s wind climatology.

4.3 Pollutant variability with air mass type

The results of the analysis of variance reveal that pollutant concentrations show a statistically significant difference between the six air mass types in winter and summer. With the exception of PM₁₀ at Glasgow Centre and Glasgow Townhead, winter has higher F ratios than summer (table 4). This demonstrates that O₃, NO_x and NO₂ concentrations vary more between air mass types during the winter. In winter, the continental air mass has the largest number of significant differences with other air mass types at both air quality monitoring stations (figures 7 and 8). This air mass has the highest mean NO₂, NO_x and PM₁₀ concentrations (figure 6a and c). The continental air mass is often accompanied by high pressure, which leads to low wind speeds and greater atmospheric stability. The stagnant conditions and lack of advective mixing reduce the dispersal of pollutants (Hodgson and Phillips, 2021). Furthermore, poor ventilation and vertical mixing induce NO from vehicle emissions to be oxidised to NO₂. Previous severe smog episodes have occurred during the winter in Britain. For example, urban background concentrations of NO₂ in London in December 1991 were at the time the highest ever recorded in Britain since measurements began in the 1970s (Bower et al., 1994). The elevated NO₂ concentrations over a four-day period occurred during a prolonged period with cold and stable anticyclonic weather types. The London SO₂ smogs in the 1950s and 1960s (e.g., December 1952) also occurred under anticyclonic weather conditions.

Pollutant concentrations reflect both local emissions and long-range transport. With this in mind, the south-easterly wind typically experienced in a tropical continental air mass imports pollutants from emission sources in Continental Europe such as in France and Germany (Baker, 2010). Malcolm et al. (2000) used modelling to estimate that over 50% of the sulphate aerosol in Edinburgh originated from outside the UK. Air quality is usually poorer under continental air masses than maritime air masses that bring in clearer air from the Atlantic. This effect will be enhanced for cities close to Britain’s west coast like Glasgow because there is limited land, and thus pollutant sources, upwind before the eastern seaboard of North America.

Pollution from NO_x, which encompasses NO and NO₂, is often a problem in a continental air mass. The mean winter concentration of NO_x (204.0 µgm^-3) at Glasgow Centre station is more than double that experienced during returning polar maritime, arctic maritime and stable anticyclonic air mass types (winter air masses one, two and six). The three air masses with low NO_x concentrations are all accompanied by cyclonic westerlies. This agrees with the findings of McGregor and Bamzelis (1995), who found that cyclonic westerly weather types had significantly lower NO₂ concentrations in Birmingham. In winter, mean O₃ concentrations were highest during the returning polar maritime air mass at both sites in Glasgow (figure 6 a and c). This finding is consistent with Hodgson and Phillips’ (2021) findings for Birmingham. Increased solar radiation leads to higher O₃ concentrations. This is because it causes NO₂ dissociates into NO and O, which are the precursors of O₃ formation via photolysis. However, this is not the explanation for Glasgow’s higher O₃ concentrations because this air mass type has a high mean cloud cover (7.0 oktas).

The tropical continental air mass has the largest number of statistically significant differences with other air masses for Glasgow Centre and Townhead during the summer (figures 8a and b). For both stations, mean O₃, NO₂, NO_x and PM₁₀ concentrations were higher during a tropical continental day than any other air mass type (figure 6b and 6d). During this air mass type, high pressure is positioned to the east or south of Britain. Stable conditions with light winds result in O₃ being trapped near the surface (Pope et al., 2016). Furthermore, the south-easterly wind transports ozone and precursor gases from heavily polluted areas in Continental Europe.

This agrees with Baker’s (2010) work, who found that the highest O₃ concentrations in Birmingham between 1998 and 2001 were associated with air parcels that had blown north-westwards from mainland Europe. Tropical continental is the warmest of the six summer air mass types (mean temperature = 18.2^oC). The higher temperatures would enhance the NO to NO₂ conversion rate. The lowest summer mean O₃ concentration at both Glasgow sites occurred during the tropical maritime air mass. This air mass is characterised by south-westerly winds, which replaces O₃ rich air with cleaner Atlantic air. In Pope et al.’s (2016) investigation, days with a westerly wind were found to have below average O₃ concentrations across Britain, thus supporting our paper’s findings for Glasgow.

4.4 Variation between air quality monitoring stations

The Glasgow Centre station has higher pollutant concentrations than the city’s Townhead site in both winter and summer. This is explained by the fact that the Centre station is only 10 m from a road and so is more exposed to vehicular emissions, whereas the Townhead site is surrounded by gardens and housing. Baldwin et al. (2015) investigated the impact that distance from a road has on pollutant concentrations in Detroit, USA. They found that NO₂, NO_x and PM₁₀ concentrations reduced by half with distances of 89-129 m downwind from a road. Whilst the Townhead site is closer to the M8 motorway, this has a negligible effect on pollutant concentrations. This is because the M8 is 360 m to the north of the Townhead station meaning that the motorway is downwind of the air quality monitoring site on most days.

4.5 Comparing winter and summer

Apart from O₃, all pollutants record higher concentrations in winter than in summer (figure 6). Ozone concentrations are higher in summer because of greater insolation and warmer temperatures. In general, days with higher O₃ concentrations across winter and summer air mass types occur in conjunction with lower NO₂ and NO_xconcentrations. Cichowicz et al. (2017) found that PM₁₀ and CO also tend to be high when O₃ is low. This is evident to some extent in Glasgow, as winter air mass types with high NO₂ and NO_xhad lower O₃concentrations (figure 6). The pronounced seasonal difference for pollutants other than O₃can be attributed partially to surface temperature inversions, which are more frequent during the winter. There is also more vehicular traffic during the winter, which contributes significantly to higher pollutant concentrations.

4.6 Severe pollution days

High pollutant concentrations are much more likely to occur at the Glasgow Centre than Glasgow Townhead site. These high values occur disproportionately during the winter. Most days in winter with high NO₂, NO_x and PM₁₀ concentrations occur under the continental air mass type (figure 9a; air mass type five). Nearly 17% of days with this air mass have high NO₂ and NO_x concentrations. Tropical maritime (winter air mass type four) and polar continental (winter air mass type three) account for the second and third highest number of high days respectively. Poor air quality in a polar continental air mass during winter is logical. This is because this air mass often brings cold and calm anticyclonic conditions. Light winds limit dispersion and pollutants often become trapped beneath a temperature inversion. The polar continental air mass was also found to be an important contributor to poor air quality in Birmingham (McGregor et al., 1999; Hodgson and Phillips, 2021). The occurrence of tropical maritime amongst the most polluting air mass types is surprising at first sight. However, this is partly explained by the fact that this is the most common winter air mass type (325 days). Only eight and ten of these 325 days have high NO₂ and NO_x concentrations respectively, which is around 3% of days. This means that a polar continental air mass in winter is more than five times likely to yield high NO₂ and NO_xconcentrations when compared to a tropical maritime air mass. There will inevitably be some variability within an airmass type, and the tropical maritime days with high concentrations are likely to have higher pressure and lighter wind speeds. This was confirmed by calculating correlation co-efficients with these variables.

Air quality is better during the summer. There are no days at either Glasgow Central and Glasgow Townhead when NO_x and NO₂ concentrations are more than 1.75 standard deviations above their respective means (Z > 1.75) for winter and summer combined. Even for O₃ and PM₁₀, fewer than five days of three air mass types had high concentrations (Z > 1.75): polar maritime, stable anticyclonic tropical and tropical continental. Hodgson and Phillips (2021) also found that the latter two air masses accounted for many highly polluted summer days in Birmingham. Buchanan et al. (2002) had similar findings in their work on Edinburgh, in which high PM₁₀ concentrations were associated with anticyclonic conditions and south-easterly winds (hence with back trajectories from Continental Europe).

4.7 Health implications

NO₂ and PM₁₀ are the two pollutants that are the most likely to breach WHO guidelines for daily mean concentrations. NO₂ exceeded the guideline on 18.5% of days across the 15 winters and summers. Nearly 91% of these exceedances occurred during the winter. Across the whole study period, only 2.4% of days exceeded the PM₁₀ guideline. This means that PM₁₀ is a less serious problem in Glasgow than NO₂. The O₃ guideline was only breached twice, both of which occurred during the summer. These findings coincide with the priorities of Glasgow City Council’s Air Quality Action Plan (2008) that specifically targets NO₂ and PM₁₀.

Cheng and Lam (2000) proposed the idea of an air quality warning system. There is currently a weather-health warning system in place in Philadelphia, US, which forecasts air mass types responsible for extreme heat. With a growing amount of research in the field of weather and air pollution in Britain’s urban areas, there is the potential to predict the arrival of anticyclonic conditions and tropical continental air masses that are typically associated with high pollutant concentrations. This system would allow vulnerable, elderly and sick people to be warned beforehand of poor air quality. These warning systems need to take into account that today’s air quality is also dependent on previous emissions and weather conditions. The UK Met Office currently issues a daily air quality index (low, moderate, high and very high). These forecasts are produced using a regional model on a 12km grid resolution (Met Office, 2024). The forecast predicts the background level of air quality at the regional scale as opposed to the higher concentrations found close to roads and other emission sources.

The highest mean concentrations were recorded for all pollutants in the winter (apart from O₃). This can be attributed both to greater emissions (e.g., vehicular traffic) and more favourable meteorological conditions (e.g., temperature inversions) during wintertime. For both winter and summer, the poorest air quality in Glasgow occurs under continental air masses (especially for NO₂, NO_x and PM₁₀). High pressure is usually positioned to the east or south of Britain in this air mass type, which demonstrates the importance of stable anticyclonic conditions in suppressing vertical mixing and dispersion. Furthermore, the south-easterly wind imports pollution from Continental Europe. By contrast, mean winter O₃concentrations were highest on days with a returning polar maritime air mass. During the study period, WHO guidelines were violated most frequently for NO₂ during the winter. NO₂ is a more serious problem for Glasgow than PM₁₀ and O₃.

In this study, the daily mean concentration of each air pollutant was related to mean daily weather variables. This means that diurnal variability was not considered. For example, pollutants that are sourced from road traffic emissions generally peak twice daily during the morning and evening rush hours. Ozone concentrations usually peak in the mid-afternoon. The use of daily averages conceals the worst health effects that occur at particular times of day. There were no air quality monitoring sites in Glasgow that had complete records over the period from 2005 to 2020. As data from two sites (Glasgow Central and Glasgow Townhead) were used over different time periods, caution needs to be exercised when comparing the mean pollutant concentrations for each air mass type between the two sites.

Further research should consider diurnal variability to assess air pollution-weather relationships at different times of day, which cannot be seen with daily averages. It is also intuitive to replicate this study’s analyses in spring and autumn to determine whether the relationships are stable to a change of season. The health implications of pollutant variations with air mass type could also be explored. The sequence of air mass types over time could be related to health indicators such as heart disease and respiratory admissions.

This study has successfully demonstrated the utility of the air mass approach to understanding the synoptic scale controls on air quality in Glasgow.

Funding

Self-funded

Conflicts of interest and Competing interests

None

Availability of data and material

Data sets used are freely available

Code availability

Not applicable

Authors’ contributions

The first author obtained the data and performed the analyses. The second author checked the first author’s analyses, edited her work and added his own insights.

Baker J (2010) A cluster analysis of long range air transport pathways and associated pollutant concentrations within the UK. Atmos Environ 44: 563-571. https://doi.org/10.1016/j.atmosenv.2009.10.030
Baldwin N, Gilani O, Raja S, Batterman S, Ganguly R, Hopke P, Berrocal V, Robins T, Hoogterp S. (2015) Factors affecting pollutant concentrations in the near-road environment. Atmos Environ 115: 223-235. http://doi.org. 10.1016/j.atmosenv.2015.05.024.
Bower JS, Broughton GFJ, Stedman JR, Williams ML. (1994) A winter NO2 smog episode in the UK. Atmos Environ 28: 461-475. http://doi.org.10.1016/1352-2310(94)90124-4.
Buchanan CM, Beverland IJ, Heal MR (2002) The influence of weather-type and long-range transport on airborne particle concentrations in Edinburgh, UK. Atmos Environ 36: 5343-5354. http://doi.org.10.1016/S1352-2310(02)00579-4.
Centre for Environmental Data Analysis (2023) The CEDA Archive. Available at: https://data.ceda.ac.uk. Accessed: 7 Mar 2023.
Cheng S, Lam KC (2000) Synoptic typing and its application to the assessment of climatic impact on concentrations of sulfur dioxide and nitrogen oxides in Hong Kong. Atmos Environ 34: 585-594. http://doi.org.10.1016/S1352-2310(99)00194-6.
Cichowicz R, Wielgosinski G, Fetter W (2017) Dispersion of atmospheric air pollution in summer and winter season. Environ Monit Assess 189, Article Number 605. http://doi.org.10.1007/s10661-017-6319-2
DEFRA (2023) UK AIR Data Archive. Available at: https://uk-air.defra.gov.uk/data/. Accessed: 7 Mar 2023
Glasgow City Council (2008) Draft Air Quality Action Plan. Available at: https://uk-air.defra.gov.uk/assets/documents/no2ten/Local_zone37_Glasgow_AQActionplan_1. pdf (Accessed: 8 May 2022). Hyperlink is no longer active.
Hodgson EC, Phillips ID (2021) Seasonal variations in the synoptic climatology of air pollution in Birmingham, UK. Theor Appl Climatol 146: 1421- 1439. http://doi.org.10.1007/s00704-021-03779-7.
IQAir (2023) Glasgow’s Air Quality Index (AQI) and United Kingdom Air Pollution. Available at: https://www.iqair.com/uk/scotland/glasgow (Accessed: 20 Jan 2023).
Kaiser HF (1960) The application of electronic computers to factor analysis. Educ Psychol Meas 20: 141-151. http://doi.org.10.1177/001316446002000116
Kalkstein LS, Corrigan, P (1986) A synoptic climatological approach for geographical analysis: assessment of sulfur-dioxide concentrations. Ann Am Assoc Geogr. 76: 381-395. http://doi.org.10.1111/j.1467-8306.1986.tb00126.x
King AM, Dorling S (1997) PM10 particulate matter - the significance of ambient levels. Atmos Environ31: 2379–2381. http://doi.org.10.1016/S1352-2310(97)00001-0.
Lee D, Ferguson C, Mitchell R. (2009) Air pollution and health in Scotland: a multicity study. Biostat 10: 409-423. http://doi.org/10.1093/biostatistics/kxp010.
Lee D, Robertson C, Ramsay C, Gillespie C, Napier G. (2019) Estimating the health impact of air pollution in Scotland, and the resulting benefits of reducing concentrations in city centres. Spat. Spatio-temporal Epidemiol 29: 85-96. http://doi.org/10.1016/j.sste.2019.02.003.
McGregor GR, Bamzelis D (1995) Synoptic typing and its application to the investigation of weather air-pollution relationships, Birmingham, United Kingdom. Theor Appl Climatol 51: 223-236. http://doi.org/ 10.1007/BF00867281.
McGregor GR, Walters S, Wordley J (1999) Daily hospital respiratory admissions and winter air mass types, Birmingham, UK. Int J Biometeorol 43: 21-30. http://doi.org/10.1007/s004840050112.
Malcolm AL, Derwent RG, Maryon RH. (2000) Modelling the long-range transport of secondary PM₁₀ to the UK. Atmos Environ 34: 881-894. http://doi.org/ 10.1016/S1352-2310(99)00352-0.
Met Office (2024) Air Pollution: information on how to use the daily air quality index. https://www.metoffice.gov.uk/weather/guides/air-quality#:~:text=The%20daily%20air%20quality%20index%20(DAQI)%20has%20been%20developed%20to,%2C%20Very%20High)%20is%20provided. Accessed 24 Jun 2024.
Oke TR (1987) Boundary Layer Climates, 2nd edition. London: Methuen, 435pp.
Pope RJ, Butt EW, Chipperfield MP, Doherty RM, Fenech S, Schmidt A, Arnold, SR. and Savage NH (2016) The impact of synoptic weather on UK surface ozone and implications for premature mortality. Environ. Res. Lett 11: Article Number 124004. http://doi.org/10.1088/1748-9326/11/12/124004.
Shahgedanova M, Burt TP, Davies TD (1998) Synoptic Climatology of Air Pollution in Moscow. Theor Appl Climatol 61: 85-102. http://doi.org/ 10.1007/s007040050054.
Stradling D, Thorsheim P (1999) The smoke of great cities: British and American efforts to control air pollution, 1860-1914. Environ Hist 4: 6-31. http://doi.org/10.2307/3985326.
Ward JH (1963) Hierarchical grouping to optimize an objective function. Journal of the American Statistical Association 58: 236-244. http://doi.org/10.2307/2282967.
WHO (2021) WHO global air quality guidelines. Particulate matter (PM_2.5 and PM₁₀, ozone, nitrogen dioxide, sulphur dioxide and carbon monoxide.

No competing interests reported.

The Synoptic Climatology of Winter and Summer Air Pollution in Glasgow, Scotland

Status:

Version 1

Abstract

Figures

1. Introduction

2. Data and Methods

3. Results

4. Discussion

5. Conclusion

Declarations

References

Additional Declarations

Status:

Version 1