Abstract

Air pollution is a significant problem in Lahore, Pakistan. The reasons for this are a combination of geographic, climatic and anthropogenic factors. A smog episode in November 2024, which was blamed for a large part on rice crop residue burning (CRB) in Punjab provided an incentive to analyse the contribution of emissions from CRB to the overall air quality in the urban area of Lahore. The air pollution simulation model HYSPLIT was used with a focus on real emission PM_2.5 sources, and pollutant concentrations rather than trajectories. The meteorological and atmospheric conditions were obtained from two sources: GDAS and METAR of the international airport. The pollution sources were CRB fires obtained from satellite information and the mobile sources in Lahore. Satellite detected fires were found to be underestimated compared to actual observations using Google Earth. GDAS weather information did not reflect the local conditions well. Using both GDAS and METAR weather information showed that PM_2.5 from CRB contributed very little to the air pollution in Lahore, however only METAR information resulted in PM_2.5 emissions similar to those measured in the city. This study indicates that the focus of the government of Punjab should be on reducing emissions from urban rather than rural areas.

1. Introduction

The city of Lahore Pakistan is habitually exposed to high levels of air pollution ^{1, 2, 3} but especially during the colder months of the year when the pollution becomes more visible, and has a more direct impact on the health and wellbeing of the population ^{4, 5, 6}. While this is a recurring phenomenon, to the extent that people describe it as the fifth season ⁷, the causes for this phenomenon to occur are diverse. It coincides with the major rice crop residue burning (CRB) in Punjab, both in Pakistan and India ^{1, 8, 9} and certain meteorological conditions that see the development of inversion layers, conditions of very little wind, and decreasing temperatures ¹⁰.

The city of Lahore has a sizable population of about 14 million people ¹¹ hence the urban pollution generated by the population itself is also significant. Reference ¹¹, and others ^{1, 7, 12, 13} identified a range of sources of the air pollution encountered in Lahore. Among the major sources identified were transport, industry, agriculture (crop residue burning CRB) and urban waste burning. The various groups have taken different approaches to the identification of the sources of pollution, varying from emission estimates coupled with remote sensing data, and ground observations ¹¹, to air sample analysis coupled with principal component analysis ¹², to back-trajectory modelling using HYSPLIT ^{1, 7, 14}.

In the beginning of November 2024, a particularly bad smog period occurred across most of Punjab, in Pakistan as well as India. The sun was barely visible in Lahore for about 3 weeks, starting about the end of October until about the 17^th of November 2024. Given the coincidence of this episode with a lot of crop residue burning by farmers in Punjab, the blame for this episode quickly shifted towards the farmers who were seen to be irresponsible to burn their crop residues under these smog conditions. The burning of residue was also against the directives of the government and fines were issued to those caught setting fire to their rice stubbles ¹⁵. In the meantime, the urban population of Lahore continued to enjoy the usage of motor vehicles and no ban on vehicular movement was imposed as is sometimes done in other countries ^{16, 17}. This episode and the social pressures on farmers provided an incentive to do an in-depth analysis of the possible contributions of CRB to the air pollution in Lahore. It was decided to use the HYSPLIT model for this analysis and focussing on the pollutant concentrations rather than trajectories. The pollution sources were the fires burning around Lahore at the time, as well as the area emissions of the metropolis itself, and the method was to be a forward trajectory approach. Given the level of detail required for this analysis would result in significant computational time and power hence we focussed on a limited time-period around the smog episode mentioned earlier. The meteorological conditions during that period were very significant and were obtained from different sources. While air pollution constitutes of many pollutants, in this study we focussed on PM_2.5 being a major contributor to the burden of disease ⁴.

The outcome of this analysis shed some light on the magnitude of CRB pollution in contrast to the contribution of the pollution generated by the urban area itself. The outcome of this work could help policy makers in prioritising measures combating air pollution if it wants to address the incidences of smog that is having an ever-increasing impact on the population of Lahore.

2. Methods

The area of study centred around Lahore, Pakistan but the fires considered to be possibly relevant for Lahore were located within the area marked by the red polygon in Figure 1. The area where actual fires were observed using Google Earth (GE) is marked by the blue polygon, while the urban area of Lahore considered under the possible exposure to smoke from the CRB is marked by the white polygon, see Figure 1.

The area where the fires were located included a large swath of Indian Punjab as well, the south-east of the white and the blue polygons.

There is a dearth of easily available meteorological information for Lahore, Pakistan however accurate, reliable and consistent meteorological information can be obtained from the Meteorological Aerodrome Report (METAR) from Allama Iqbal International Airport, Lahore, Pakistan. This can be obtained through sites like Windy.com ¹⁸ or a dedicated METAR website ¹⁹. The METAR provides air and dew point temperature, air pressure, visibility, cloud cover, wind speed and direction and some other aeronautical specific parameters and is updated every 30 minutes.

Another source of meteorological information was the National Oceanic and Atmospheric Administration (NOAA) web archive in the form of Global Data Assimilation System (GDAS) data, a standard gridded 3 hourly output, based on a 1º x 1º grid. There are 35 meteorological parameters in the GDAS format.

One of the significant parameters GDAS also provides is the planet boundary height layer (PBHL). In air pollution studies the PBHL is very significant because it indicates the height of the boundary layer where mixing of the pollutants occurs. The height varies strongly depending on the prevailing meteorological conditions but also the ground surface temperature.

The presence of CRB is inferred from temperature anomalies measured by satellites such as the NASA/NOASS Suomi national Polar- orbiting and NOAA-20 satellites that provide a product in the form of the VIIRS at 375 m spatial resolution. The temperature anomalies are detected on a daily basis and recorded in terms of longitude, latitude, brightness temperature, and acquisition date and time and some other parameters. This information is freely available via the NASA FIRMS website ²⁰ and updated daily. The FIRMS product for South Asia covers a very large area hence a subset of the fires in Punjab was selected. This area was set by 28.5/70 (Lat/Long) in the South-West corner and 32.5/76 (Lat/Long) in the North-East corner, see the red polygon in Figure 1. The total area for fire detection was 256,847 km², enclosing much of the Pakistan and Indian Punjab provinces.

The fires detection using the satellites does not capture all the active fires in the area ²¹ but it is difficult to verify this on a large scale. However, some verification of the information provided by the satellites regarding the number and location of stubble fires could be obtained via Google Earth (GE). It was discovered that one image of GE of an area close to Lahore was obtained on the 22^nd of October 2024, being a very recent image. The total area covered by that image was 714 km² (13.3 km wide and 53.67 km long) located southeast of Lahore, see the blue polygon in Figure 1. In this image of GE the recently burned fields were clearly visible by way of black linear marks in the fields showing the burned wind rows, and at times even actual fires with associated smoke columns could be seen. These recent fires were located and marked in GE and the marks exported as a text file and then counted. Where multiple fields side-by-side had been burned, the total number of those fields burned at that location were noted in the description of the mark in GE so that they could be counted later as individual fires. The vast majority of the fields in Punjab are 1-acre fields, so that the number of burns could be converted to a total area burned.

The emission from stubble fires was derived from published studies. There have been many studies examining the emission from rice stubble burning e.g. ^{22, 23, 24, 25}. Based on these studies an emission of 10 g PM_2.5 / kg stubble burned was determined. The total emissions are related to the total weight of stubble burned, which is related to the yield of the crop harvested. The average rice production (kg/ha) in Pakistan obtained from the USDA website, ²⁶ was 3.7 t/ha. According to the Punjab Bureau of Statistics ²⁷ the average rice yield in 2023 was 2.2 t/ha. For this study we assumed an average rice yield of 2.8 t/ha. The amount of straw was related to the rice-to-straw ratio which was assumed to be one (1) ²⁸. This gave us a total emission of 11200 g PM_2.5 per acre from the burning of the stubble. The duration of the burns was assumed to be 24 hours, resulting in an hourly PM_2.5 emission rate of 466 g/acre/hr.

Emissions from the urban area of Lahore come from mobile sources such as motor vehicles (small/large vehicles, motorbikes, rickshaws, buses and trucks) and stationary sources such as power stations and factories (steel works, brick kilns etc) and domestic burns. The latter three sources were ignored in this study. It would obviously add to the level of emissions if these were to be included as well. The number of motor vehicles (2 million) was taken from published data ²⁹, while the vehicular emissions have been obtained from other studies, assuming they apply to a Pakistani context as well. Motor vehicle emissions of fine particles (PM_2.5) were obtained from a study by ³⁰ in Wuhan. The PM_2.5 emissions were 1.3 mg/km travelled. The daily distance travelled of 30 km in cars in Lahore was taken from a publication by the Urban Unit in Lahore ³¹ and a small survey conducted by Numbeo ³². Motorbike/rikshaw emissions were taken from a study by ³³ as 53 mg PM_2.5/km travelled for an average of 30 km per day travelled. Similarly, as with motor vehicles this distance was also taken from the Urban Unit publication and the small Numbeo survey. The total number of motorcycles in Lahore was estimated to be 4.5 million ²⁹. Bus and truck emissions came from data published by ³⁴ based on work conducted in China and assumed to be 252 mg PM_2.5/km. It was also assumed that buses and trucks would travel 10 times the distance in the city compared to cars. The total number of buses and trucks in Lahore was 600,000 ²⁹ The total maximum urban hourly PM_2.5 emissions was 2200000 g/hr. This maximum was assumed to be emitted during the day, starting around 7 am until about 7 pm, with the emissions tapering off to about 15 % of the maximum during the night. The area representing the urban area of Lahore is marked by the white polygon in Figure 1.

The ambient air quality in Lahore was derived from the data collected at Forman Christian College (FCC), Lahore (Lat: 31.5206 and Long: 74.3368) where air quality has been monitored now for several years. The system employed is a Haz-Scanner system from Environmental Device Corporation (EDC), USA. It is located in the middle of Lahore, in a residential area about 500 m from major arterial roads. The air pollution parameters monitored were particulate matter with a particle size of < 2.5 um (PM_2.5), and particulate matter with a particle size of < 10 um (PM₁₀) using 90° infra-red (IR) light scattering systems. For carbon dioxide (CO₂), CO, NO₂, SO₂, O₃ Alpha Sense® sensors were used, while for the temperature and the relative humidity (RH) an HTM2500LF sensor was used. Wind speed and direction and rainfall were obtained with standard anemometer and a wind vane and a tipping bucket rain gauge from EDC. The data was recorded every ten minutes. The system has been serviced and calibrated every year by the manufacturer according to USEPA standards.

The simulation of the movement and quantity of air pollution in November 2024 was carried out using HYSPLIT v 5.4.2, an air pollution simulation program developed by NOAA Air Resources Laboratory, Maryland USA ³⁵. HYSPLIT, the acronym for HYbrid Single-Particle Lagrangian Integrated Trajectory, illustrates the approach used in air pollution modelling whereby the advection and diffusion in the atmosphere is simulated using fictitious particles representing gases or aerosols. The particles are considered small enough to follow the motion of smallest eddies and, at the same time, big enough to contain a large number of molecules. Each particle is moved at each time step by transport due to mean wind and diffusion, related to the turbulent wind velocity fluctuations ³⁶. Following the input from meteorological grids, calculations are done on the grid nodes. Air concentration calculations associate the mass of the pollutant species with the release of the particles. The dispersion rate is calculated from the vertical diffusivity profile, wind shear, and horizontal deformation of the wind field. Air concentrations are calculated as cell-average concentrations for the particles ³⁵. The number of particles released during emission cycles can vary and determines the accuracy but also the speed of the calculations. HYSPLIT can calculate the trajectory of single particles, showing the general movement of a pollutant released from a source, as well as grid concentrations both in the horizontal and vertical directions. In the vertical direction it is set by the number of layers of interest while in the horizontal direction it is determined by the grid resolution, set by the user.

The meteorological input for HYSPLIT was provided in two different ways. Firstly in the form of GDAS data. HYSPLIT has been developed around the GDAS format and provides all the options of displaying and extracting meteorological data for the relevant places from the GDAS files. There are 35 meteorological parameters presented, with a minimum of 4 required to run HYSPLIT, being U and V (the horizontal wind components, T (temperature), and Z (height) or P (pressure). Secondly the METAR derived meteorological data were used. In HYSPLIT the option exists to input the user-defined weather information. For that 4 parameters are required: the wind speed (m/s), wind direction (deg), Mixing Height Layer, and Stability classes. The first two parameters were obtained from the METAR, while the third was obtained from the GDAS data as the Planetary Boundary Layer Height (PBLH). The METAR data had a different recording frequency (every 30 minutes) while the GDAS data were given for every 3 hours. The PBLH for the missing METAR times was therefore obtained by linear interpolation using the GDAS data. The stability classes were determined based on the time of day (sunrise/sunset), and the wind speed according to the Pasquill stability classes ³⁷. This user generated weather information for a period of 20 days (1/11/2024 – 20/11/2024) was submitted to HYSPLIT which then developed a GDAS format for that period over a 250 by 250 km domain at a horizontal resolution of 10 km, centered around Lahore.

Other input consisted of emission source locations including emission heights, emission rates, duration, grid output (resolution and vertical layers), and emission types (particles or gases). For this study only the distribution of PM2.5 pollution from two areas has been considered: the first area, being the multiple sources as fires present in Punjab during the rice stubble burning season and the second area representing the metropolitan area of Lahore. The locations of the multiple sources were based on the information obtained from the VIIRS satellite with the understanding that it was under-estimating the number of fires. The number of fires changed every day, varying from 100 to 1500 per day, and a total of 11631 fires for the whole 20 day period. The area of each fire was assumed to be 4000 m², while the emission output of each fire was 466 g/hour of PM_2.5. The fires were therefore not treated as point sources but as sources with a certain footprint. The total contributing metropolitan area was determined to be 788 km² with emissions as discussed earlier.

The output of HYSPLIT covered a large swath of Punjab and was converted to an ASCII output. This output was then filtered in order to extract the values geolocated within the boundaries of the metropolitan area of Lahore, bound by 31.35/74.0 (Lat/Long) and 31.65/74.5 (Lat/Long), see the white polygon in Figure 1. In this way the contribution of the metropolitan emissions could be compared with the emissions caused by CRB around Lahore within the boundary of Lahore.

Where applicable in the presentations the output of HYSPLIT which is in UTC, was changed to Pakistan time by adding 5 hours.

3. Results

The METAR observation of the air- and dew point temperature, the wind speed and the visibility at the airport are presented in Figure 2a and 2b respectively.

In all four parameters the METAR observation show a very strong diurnal pattern, driven by the daily temperature fluctuations. It is also clear that there was an episode that started in the first week of November (5/11) until about the 17^th of November where the air temp was similar to the dew point, indicating fog conditions, the wind speed reduced significantly, and the visibility became extremely poor. After that period visibility at the airport returned to acceptable levels.

Weather input parameters into HYSPLIT are essential for reliable simulations. A comparison between the GDAS data and the actual observed METAR data during the time that the smog episode occurred is presented in Figure 3 a, b and c.

There are some significant differences between the GDAS data and the observed METAR data. Both the GDAS maxima and minima are larger than the METAR observations by several degrees C. The GDAS windspeed also remained consistently higher, not at times when wind was recorded at the airport but the times when no wind was recorded at the airport. From the data presented in Figure 3b it shows that GDAS did not have any zero wind observations while the METAR data indicated that more than 45% of the observations recorded zero wind speed.

The concentration of PM_2.5 and the wind speed measured at FCCU campus in the middle of the city for October and November 2024 is presented in Figure 4.

While the PM_2.5 in October was already high, in the first 3 weeks of November the PM_2.5 concentrations reached alarming levels. This coincided with a period of very little wind even during the day. By the time the pollution diminished, the wind during the day started to pick up as well. From the general observation there tends to be a strong correlation between the magnitude of the PM_2.5 and the wind speed. Even in October, elevated levels of PM_2.5 coincided with lower wind speeds. At all times however the diurnal changes in PM_2.5 were very strong and consistent even when levels of more than 600 ug/m3 were obtained.

The PBLH is a significant determinant in the expression of air pollution at any given location. The GDAS PBLH is presented in Figure 5.

The daily fluctuations of PBLH is clearly visible as is the period around the 10^th of November when the PBLH also during the day did not change.

The number of daily fires that occurred during the last four months of 2024 in a large area of Punjab are presented in Figure 6. In total 25067 fires were detected over that period.

The number fires increase towards the beginning of November with a maximum of 1511 observed fires on 17 November. There is great variability in the number of fires from day to day, but the crop stubble burning season was basically finished by the 24^th of November.

The satellite observations tend to underestimate the number of fires. There is anecdotal evidence that suggests that farmers know when the satellite passes over Punjab, and they wait before they light up their fields. From GE we obtained the actual number of fires that happened over a period in October 2024 and were able to compare that with the satellite information.

The visual detection of the burned areas from the GE image and the comparison with VIIRS detected crop fires is presented in Figure 7 while a summary of the data is presented in Table 1.

We assume that the fires had occurred over a space of 22 days (01 – 22 Oct 2024), hence we used the VIIRS data for that period. A summary of the data is presented in Table 1.

There is a discrepancy between the number of fires detected by the satellite and those occurring. More than thirteen times as many fires were detected using GE. That number equated to about 959 ha burned according to GE. Using the satellite information the number was much less (185) but the representative area was larger (2601 ha) because the spatial resolution of the satellite is 14.06 ha per pixel. In this study we focussed on the number of fires for a given area rather than the area burned. However, in the simulations the fires were allocated a burned footprint of 1 acre each, which compensated for the lack of number of fires.

The forward trajectory simulations track the general particle movement from the source over a given time. The simulation of the trajectories of pollution particles starting on 14 different days in the beginning of November all starting in Lahore for a duration of 24 hours is presented in Figure 8 using two different weather inputs, GDAS and METAR.

Initially the general movement of the particles was away from Lahore in a south-easterly direction, but then returning to Lahore, which occurred on the 1 and the 2 of November. On the other days the general movement was away from the city in a westerly direction, tracking at a speed of about 150 km per day. On the 6^th of November, the movement of air was fairly restricted, doing a turn on itself during the day which also happened on the 13^th of November. This latter movement would be detrimental for the air quality in Lahore, when considering pollution generated by the city itself. The east-west movement could carry smoke from fires east of the city (in Indian Punjab) to the city. Changing the weather input from GDAS to METAR show how the trajectories change. Comparing Figure 8A with 8B, which have the same Lat/Long range, the trajectories have become much more confined and localised using METAR inputs which means that the pollution does not travel very far.

The use of trajectories is helpful in as much it visualises where and what altitudes particles reach following the emission of certain pollutants. It does however not indicate what concentrations are encountered and where, unless specific concentration simulations are performed. Based on the number of particles in each grid cell, concentrations can be derived which can then be used to determine e.g. the effect of the movement of smoke from stubble fires on air quality in Lahore as will be discussed in the next section.

The geolocations of the fires from 01 Nov until the 20^th of Nov 2024 in relation to the geolocation of the metropolitan area of Lahore are presented in Figure 9.

In total 11631 fires were identified for that period. The number of daily fires for a given date was presented in Figure 7. We know that this is an underestimate in terms of number of fires based on the comparison with GE, while it would be an overestimate in terms of the area burned. The total area burned was 4652 km² based on an area of 4000 m² per fire.

The results of the simulation of the effect of the fires around Lahore on the air quality represented by the PM_2.5 concentration, in the Lahore metropolitan area is presented in Figure 10. The METAR weather input data and the GDAS data were used for this simulation.

The simulation illustrates that the fires surrounding Lahore have some impact on the air quality of the city when using METAR weather input. There is a diurnal effect caused by the daily temperature fluctuation, wind speed and PBLH changes. When the wind speed reduced significantly such as around the 14^th to the 17^th of November, see Figure 2B, the impact of the fires on the air quality in Lahore became negligible. During the worst of the smog episode with visibility reducing to 50 m (see Figure 2B), the contribution of the smoke of CRB to the levels of air pollution in Lahore was simulated to diminish completely due to the lack of wind speed. The use of GDAS for the simulation significantly reduced the contribution of the fires to the air quality in Lahore to a fraction what the METAR simulation estimated.

The results of the simulations of the mean PM_2.5 concentrations in Lahore based on the variable emissions from solely the urban area are presented in Figure 11. The weather data input was based on the GDAS data for the period 01 Nov – 20 Nov 2024.

The simulated levels of the PM_2.5 in the city of Lahore solely based on its own emissions displayed a diurnal pattern caused by the changes in wind speed, the PBLH which are in turn caused by temperature changes, and other atmospheric changes. The magnitude varied from close to zero to a maximum of about 400 ug/m3 at about the 13^th of November. There was little evidence of an impact of the wind-still period around the 14 to the 18^th of November because the weather input was based on the GDAS information, which did not provide accurate enough weather information. The average levels of pollution were significantly less than the measured PM_2.5 concentrations obtained at FC College in Lahore. Only towards the end of the period did the simulated values become more in agreement with the measured values when there was an increase in simulated values and a decrease in the measured ones.

The use of GDAS data as weather input allowed for too much wind compared to the actual conditions as was highlighted in Figure 3B. GDAS did not include any period with zero wind speed, hence simulations of the emissions from the urban area would see too much dispersion. The use of user-generated weather information using the METAR data provided an opportunity to simulate the actual weather conditions.

The concentration of PM_2.5 in the metropolitan area of Lahore at two different heights using METAR weather information is depicted in Figure 12.

The concentration showed again a strong diurnal pattern, particularly at the beginning of the simulation period and toward the end (18 – 20 Nov). In the middle of this period the very low readings were absent while the overall concentrations during that period remained between 600 – 1200 ug/m3, with only a few exceptions. The same order of magnitude during that period was also found at the monitoring station of FCCU. The simulation exceeded the measured concentration at the beginning of November and at the end of the 20-day period. Particularly at the end, very high levels of PM_2.5 were still being simulated while the actual measured PM_2.5 concentration had decreased to between 50 and 400 ug/m3.

The movement of the pollution is very sensitive to the presence or absence of wind speed. In Figure 13 the average simulated concentrations of PM_2.5 in Lahore as a function of the wind speed class are presented.

With an increase in the wind speed class, the pollution levels decreased.

4. Discussion

The weather information obtained from various sources illustrated the conditions that was experienced in Lahore in the first 3 weeks of November. The dew point temperature and the air temperature came very close, indicating periods of fog, while the wind speed measured at Lahore airport reduced drastically during the day. In Lahore it is normal to have very little wind at night, but this lack of wind also extended to the day. The air quality monitoring at FCCU indicated excessive levels of PM_2.5. The PM_2.5 readings obtained by the FCCU air quality monitor are based on the common 90º infra-red light scattering technique. The equipment is not a low-cost monitor and has had an annual calibration, but it would be prone to over-estimating the PM_2.5 in the presence of high relative humidity ³⁸, conditions that were encountered during the November 2024 period. Much more reliable beta-attenuation based monitors ³⁹ are now available as part of an expanding network of monitors initiated by the Government of Punjab, but they were not available when we initiated this study.

The source of this pollution, while perhaps not this smog episode in particular but the smog season in Punjab in general, has been discussed by several people ^{1, 8, 7, 10, 14, 40, 41}. Much of the work focused on the connection between crop burning and the elevated levels of air pollution in the urban centres of India and Pakistan. The work presented in this paper suggests that the main contributing factor to the pollution is caused by the emissions from the city itself with only a small contribution from CRB that are surrounding the city. The simulated concentrations of the mobile urban emissions yielded sometimes one hundred times higher concentrations of PM_2.5 compared to the contributions from CBR.

Reference ⁴² concluded from their work in the Delhi (India) region, based on the use of 30 low-cost air quality monitors, and trajectory modelling using HYSPLIT that there was only a low coupling between crop burning and air quality in Delhi. Others however, such as ^{1, 14} concluded from their work that there is a very clear connection or coupling between the air quality in Lahore and crop stubble burning around Lahore. Much of their work has been based on the use of trajectories, which only tell a part of the story, namely the trajectory of particles. That still needs to be converted to concentrations based on the number of particles reaching a particular area. Based on our work, using only trajectories we would have said also that there is a strong connection between the two, however using concentrations, the coupling becomes very weak. The city emits much more pollution by itself compared to additional pollution caused by the thousands of fires burning at some distance from Lahore. As ⁴² indicated, the effect of the fires are of course much more severe closer to the source of the fires, but that is not under consideration here.

In this approach the emissions from the urban area have been limited to just emissions from the transport sector based on estimated distances travelled per day, number of vehicles and average emission rates. While this might perhaps exaggerate the emissions from the transport sector for this case study, it is compensated by omission of emissions from other sectors such as industry and households. Their incorporation would add to the urban emissions that lead to the development of smog during the winter months.

The use of actual weather information is critical when using simulation models such as HYSPLIT for local problems. The scale of the GDAS information is simply not sufficient to reflect the actual conditions. This need was highlighted by ⁴³ for the use of real-time boundary conditions which improved the PM₁₀ forecasts in the region of Hungary. In our case we only used real time wind speed and direction. The MLH was still derived from the GDAS model. We showed that wind speed is very critical in these, hence having access to accurate wind information preferably for different heights is paramount.

Uncertainty of the PBLH is a significant issue for the accurate and reliable determination and modelling of the air pollution in Lahore. There is a complete lack of information regarding the PBLH in Lahore. There is no provision of information regarding the PBLH derived from actual observations such as the regular release and tracking of weather balloons or radiosondes. The nearest to Lahore where such regular releases do occur is in New Delhi, India ¹⁸. At the moment modelled PBLH information needs to be used, which, as has been demonstrated in this study, does not always provide accurate meteorological information.

Proper and accurate forecasting and source allocation of air pollution for the city of Lahore is essential if policies are to be developed addressing the real sources of air pollution in the city. Currently, large sums of money are being spent on addressing the burning of stubble ⁴⁴ but this might be a futile effort towards the air quality improvement of Lahore when the urban pollution sources are poorly addressed. The latter would benefit from additional financial resources if they perhaps were to be redirected from addressing stubble burning.

It should be said, off course, there are very sound agronomical reasons why crop stubble should not be burned ⁴⁵ but that is not relevant for this paper.

Conclusion

Smog episodes such as in November 2024 trigger a flurry of government initiatives funded both locally and internationally, addressing the sources of the pollution. In the process farmers burning their stubble received a lot of attention which, with the insight obtained from this study, might have been misplaced.

The atmospheric conditions for November 2024 obtained from various sources all pointed towards very high levels of pollution which in our estimation was a coincidence of unusually low consistent windspeeds, low temperatures (air and dewpoint) combined with high emission loads from the metropolitan area rather than emissions from crop residue burns. The additional burden from smoke caused by the CRB to the levels of pollution experienced in Lahore has been estimated as being very low, despite this coincidence of the timing of the episode and the crop stubble burning in Punjab.

The need for quality and timely and readily available detailed weather information is critical if a proper understanding of current air quality issues in Lahore is required. Relying on global weather updates is simply not good enough.

These findings might assist the government of Punjab in the development and implementation of directed policies that will ultimately see an improvement in the air quality of Lahore.

References