Salman Atif ( Institute of Geographical Information Systems, National University of Sciences and Technology, Islamabad, Pakistan. )
Ejaz Hussain ( Institute of Geographical Information Systems, National University of Sciences and Technology, Islamabad, Pakistan. )
Junaid Aziz Khan ( Institute of Geographical Information Systems, National University of Sciences and Technology, Islamabad, Pakistan. )
April 2020, Volume 70, Issue 4
Narrative Review
Abstract
The 2015 heat wave resulted in an estimated over 1200 deaths during the month of June. However, there were no records on the spatial distribution of the effects of this heat wave. An analysis of Moderate Resolution Imaging Spectroradiometer (MODIS) land surface temperature (LST) daily data was conducted to identify regions that experienced above normal temperatures in 2015. An analysis of the monthly averages showed that in general April and May were the warmer months in Karachi, unlike the case in 2015. In addition, the general warm trends were common in the highly industrialised Sindh Industrial Trading Estate (SITE) and Liaquatabad towns, while Gadap, with its mostly barren land, and New Karachi also experience higher temperatures. Coastal towns were naturally cooler and more habitable in the given scenario. A count of the spatial presence of health facilities for the city was also extracted where Gadap and Korangi were poorly served while the more affluent towns of Defence Housing Authority (DHA) and Gulshan-e-Iqbal appeared to be better served.
Keywords: Heat wave, Karachi, Urban heat islands, MODIS-LST, Remote sensing.
https://doi.org/10.5455/JPMA.5478
Introduction
The effects of climate change on quality of life and health of individuals and communities is a well understood prospect.1 One of the major concerns with the current scenario, however, is heat stress and resulting deaths. Additionally, an increase in respiratory and heat-related illnesses is commonplace. In a normal situation, humans are acclimatised to the prevailing conditions where they plan for housing, power availability and cooling mechanisms as well as their working hours according to the prevalent heat index. It is, however, believed that people do not live comfortably outside the temperature range of 17-31ºC. If, however, the temperature range changes, so will the pattern of lives altogether, ranging from adaptability measures in terms of clothing to water and power consumption. One adaptability feature, the urban land-use and land-cover, nevertheless remains critical. In case of Karachi, this is a worst-case scenario where urban growth has shown to be tremendous. For instance, the recorded population of the city is 16,051,521 inhabitants.2
The effects of excess surface heating have been broad and wide, like, for instance, the number of deaths in the 2015 heat wave, which were unprecedented.3,4 The event coincided with the holy month of Ramadan which meant no food/water consumption for more than 90% of adult population during the day, along with power shortages and the usual difficulty to access health services. Also, as high temperatures are considered a common occurrence in the city, people tend to try and adjust to it. But the situation went somehow out of control and blame-game between various stakeholders continued well into the crisis on improper provisioning and management of resources.5
In this context, the current analysis was conducted related to temperature from March, 2000, to August, 2016, for the entire city, based on Moderate Resolution Imaging Spectroradiometer (MODIS) land surface temperature (LST) data. MODIS was chosen as it provides a good approximation6 of the day and night surface temperature (ST). The overpass time for MODIS Terra satellite is 10:30AM in descending mode while its ascending overpass time is 10:30PM.7 These maps were then used to demarcate the dominant high temperature zones for the city. This was then examined in terms of the presence and geographic accessibility of health facilities in each heat island.
Literature Review
One of the key methods for studying geographic phenomena of large spatial and temporal stretch is the use of Remote Sensing (RS) data and Geographic Information System GIS.8 It has been widely employed for urban land-use and land-cover characterisation, classification, surface run-off modelling, disaster mitigation, sustainability planning, urban LST mapping, quality of life analysis and public health management.1
Surface heat islands can be mapped using either data gathered from meteorological stations or from a field surface with the help of a mobile sensor. While the former limits us to available records from only a few stations across an entire city, particularly in the case of Third World countries, the latter is bound to be limited to roads or commutable segments of a neighbourhood. Surface urban heat islands (SUHI) mapped using satellite images have a different character in comparison. These tend to work well for inaccessible areas or in resource-deficient scenarios. The maps of ST derived from satellite remote sensing (SRS) can then be further classified as the urban canopy layer (UCL) heat islands, situated between the roughness elements like trees and buildings but lower than the neighbouring rooftops, and as the urban boundary layer (UBL), situated above the first and subjected to the effects of the urban surface.9 It must be noted that with the help of GIS there is further possibilities of segregating regions close to large water bodies and industrial units etc.
For the study of UHI, images from several SRS sources have been used at various times. For large-scale studies, National Oceanic and Atmospheric Administration (NOAA) satellite Advanced Very High-Resolution Radiometer (AVHRR) (~1000m), MODIS-LST (~250m-1000m) and Landsat temperature bands (~60-120m) have been widely used.10-12
MODIS, however, remains a preferable choice as it can be used to get daily coverage for day and night, and for large areas. One downside, however, is the excessive cloud cover that hinders recording good quality surface emissivity from space. But for that, spatial masking and subsequent gap-filling proves sufficient and has been employed where necessary. It is by far its temporal consistency and continued availability that MODIS LST for ST mapping remains an important choice.6
Studies13 have successfully been able to scale the spatial resolution of LST maps and to render it a better consistency using Regularized Splines with Tension (RST) interpolation, preparing temperature maps of good quality for epidemiological research.
Studies14 have mentioned climate change as one of the most significant urban health hazards. In an urban environment, its structure, surface types, population, road network density, traffic density and priority fuel types used as well as the coherence and sustainability of public transport networks contribute to this change at large. In the case of Karachi, urban expansion is at a large scale, with its urban built-up area increasing from ~500 sq.km in 1991 to ~1500 sq.km in 2013.15 Similarly, its population size and lack of coherent and large-scale public transport network together mean that most people will prefer using private vehicles, hence, contributing to the pollutant concentration. A study16 done in Bogota, Mexico City and Santiago suggested that temperature and air pollution are class-independent in terms of their effects on mortality.
Karachi in 2015 experienced its worst heat wave since 1979. While some other cities of Pakistan, like Larkana, Turbat and Sibi, recorded 49ºC, Karachi was not behind with 45ºC recorded on June 20. Weng et. al.17 reported high temperatures, dehydration and very high humidity as major causes of deaths in the city. Romero et.al16, however, called these causes as irrelevant and suggested that urban segregation had a bigger role to play. Climate Change and Development Knowledge Network,18 an independent climate activism organisation, estimated that more than 1200 people died during the episode. However, some independent estimates put this death toll at multiple times higher as there are no substantial records of such incidents published.19
There is a general tendency of urban bias and class-oriented distribution of health facilities in Pakistan. It is believed that the far flung regions of Karachi are devoid of basic facilities, like water, power and healthcare.19 Currently, however, it is hard to find research on the emergency caused by UHI distribution, spatial variability of temperature across the city and the presence of reliable healthcare in case of emergency. The current study, therefore, focussed on addressing these spatial inequalities.
Study Site
Karachi is located in the south of Pakistan and is the only commercial port city in the region since the British era. On its east, it is bound by the fertile plains of the Indus while on its west, it has the desert province of Baluchistan towards which its coastline continues. To its northwest, however, are the dry and moderately high (~300-600 m) Kirthar ranges.20
The total area covered by the city is ~3,527km. It boasts of ~163km coastline of which ~48km is dedicated to the port activities, while the rest is being used for recreational purposes.21 Also, major parts of the city have grown along this coastline.
Citimertic22 puts Karachi as the 7th largest city in the world in terms of population. According to the Globalisation and World Cities Research Network (GaWC),23 Karachi is classified as an instrumental social and economic hub for Pakistan well above the league of its other cities such as Islamabad and Lahore.
Administratively, Karachi is divided into six districts: Karachi South, Karachi East, Karachi Central, Karachi West, Malir and Korangi. It is further divided into 18 towns (Figure-1),
which are subsequently divided into Union Councils.24 Health facilities in the city range from large government-owned hospitals to small private clinics owned by individual practitioners working under licensing from the government.
Methodology
The study used MODIS LST records version 5 data which is freely available from the National Aeronautics and Space Administration (NASA) Level-1 and Atmosphere Archive and Distribution System (LAADS) portal25 for download. Records from March 5, 2000, to August 28, 2016, for Karachi were downloaded. MODIS data is distributed as tiled set of rasters in highly compressed hierarchical data format (HDF). For Karachi, h24v06 tiles were downloaded. Once downloaded, these files had to be made suitable for use with GIS and image analysis software and, hence, were processed using the MODIS re-projection tool (MRT) available online on the NASA Land Processes Distributed Active Archive Center (LP-DAAC) portal.26
The tiles were then processed to extract day and night ST layers for and the quality assurance (QA) files that came with the data. The selected projection parameters were universal transverse Mercator (UTM) projection zone 42N and the datum was WGS84. For better pixel consistency and artefacts removal, data was re-sampled to a higher resolution from its original 1000m coarse resolution.13 Hence, the rasters were re-sampled to 200m resolution using nearest neighbours re-sampling. This latter method did not alter the adjacent pixel values and, hence, was suited for re-sampling such datasets. Once the initial processing was completed, the rasters were imported into Geographic Resources Analysis Support System (GRASS) Geographic Information System (GIS) and processed for removal of pixels with bad quality (Figure-2a).27
According to the MODIS LST v5 user manual,28 the values in the original downloaded MODIS LST Spatial Data Sets (SDS) can alter up to a few degrees and, hence, it was necessary that pixels with >1ºC of erroneous values be screened out. After masking out the bad pixels from the MODIS datasets, upper and lower outlier detection was done. This was necessary because MODIS data was prone to contain negative values due to the presence of cloud pixels representing cloud top temperatures. A simple exercise would be removing all values below 0ºC as Karachi has never experienced such temperature. However, quartiles for all rasters were derived and used to mask out the unusual pixel values (Figure-3a).
After masking the MODIS data for highly error prone pixels, it is common to be left with null spaces in the output rasters (Figure-2b). Mostly in coastal areas where cloud cover is common in Karachi and after quality masking evidently, an open patch was left. Regularised Splines with Tension-based raster interpolation was used to fill the null areas. A total of 1,752 rasters did not have sufficient points available for reliable gap-filling and were left out. MODIS LST has a scaling factor applied and the data therefore needed rescaling and conversion of values to ºC.
These were then interpolated for using harmonic analysis of time series. Once the rasters for the day and night LST temperatures were ready, they were averaged out to map high day and night temperature regions as SUHI. Subsequently, spatial distribution density maps for Karachi were produced highlighting the urban health inequalities.
The map of the health facilities included urban health centres/units, rural health centres (RHCs), conventional eastern medicine facilities, pharmacies, available records on general physicians, specialist doctors (not just those treating health stroke patients), children hospitals and general hospitals. These were then tabulated as count against the town-based population densities.
All health facility points were converted to heat maps. For these heat maps, the feature density for each town (polygon) was mapped.
Results
An ST analysis revealed that the night and daytime temperature differences were stark. Daytime rescaled time series range was calculated to be 7.17ºC, while those for night-time was at 5.89ºC.
The mean ST was 28.35ºC, with the lower average values observed in Defence Housing Authority (DHA) (27.2ºC), Kemari (27.8ºC) and Korangi (27.7ºC), which are coastal and benefit from cold sea breeze, particularly at night. Their night-time temperature mean was (27.6ºC).
Higher ranges of temperature were observed in Sindh Industrial and Trading Estate (SITE) (29.17ºC) and Liaquatabad (29.2ºC), where the former is highly industrialised and bears significant on the latter both in terms of temperature and economic effects. This also owes to their distance from the coast and central nature to the urban built-up area.
An analysis of the seasonal variability of temperature indicated that in general high temperatures were observed during the second quarter of the year from April to June. The mean for day was 38.32ºC and 24.78ºC for ST at night. The coastal localities of Kemari, DHA and Korangi, benefiting from the cold sea breeze, had a mean temperature of 30.7ºC.
In case of monthly averages, the months of April and May were the warmest. Of the summer, the daytime mean for April was comparatively higher at 39.64ºC compared to 38.85ºC and 36.47ºC in May and June. However, as a whole, May tended to remain warmer with higher temperatures recorded in the Gadap town and areas northwest of the urban built-up regions.
The rest of the years were almost at par with the exception of 2012 which was colder (Figure-4 a,b).
It is pertinent to mention that these are the daily average, and not the daytime STs for the city.
A simple overview of the health facility count for each town of the city suggested that there was a serious imbalance in terms of the presence of these facilities (Table-1).
It would have been interesting to map the computability in case of a heatstroke of the patients to these destinations, but no such records were available. However, the highest density of these facilities is in Saddar and Jamshed towns followed by DHA and Gulshan-e-Iqbal. It must be noted that these were the districts observed to bear comparatively lower temperatures in the wake of the 2015 heat wave and even in general (Figure-4b).
The hospital-to-population ratio for 2010 was really poor. For Gulshan-e-Iqbal, where the number of hospitals was the highest, for every ~27,000 individuals, there was one available hospital. Most of these hospitals were private and not readily reliable or accessible. However, for Jamshed town that followed in terms of hospital count, the ratio stood poor at ~33,728 people per hospital. The town boundaries were not clearly demarcated and connectivity within the city was appropriately good, which means the actual load during a health emergency might be difficult to handle.
In Gadap town the situation was grave, with most of the population being rural and located at places that did not have access to good healthcare facilities.
The total number of people that one hospital roughly had to cater to was 82,420. By virtue of both the land area covered and difficulties in commuting, the ratio was dangerously low. Other poorly-served towns included Lyari and Korangi with the population-to-hospital ratio being one to ~50,000.
Discussion
The problem with the 2015 heat wave that were cited in general were unbearably high temperatures, but an alternative set of perspectives have been found as interesting contributors to the state of affairs. While temperatures remained below historically recorded highs (Figure-4b), as found in the MODIS LST analysis, and at par in some cases, it was mostly the provision of utility services, and health service amenities that must have resulted in a large number of deaths as most of the districts had never experienced such temperature highs. It was also interesting to see that densely populated areas with a high proximity to the sea had better amenities, and probably better provisioning of power and water supplies, but regions like SITE and Gadap were neglected at both levels and were apparently suffering from lack of appropriate provisioning of these facilities (Figure-5).
While it would be interesting to link the entire episode with power distribution patterns and availability, such information is not available. Another interesting perspective is that barren lands towards the north were highly affected, and one widely known reason for that is the lack of vegetation and exposed soil bed. The reason needs to be investigated by further studies.
Conclusion
The overall trend of temperature across the city was not exceedingly high on a long-term basis. However, some mismanagement on the part of the local authorities and power supply inconsistency during the season need to be analysed for other details. However, it is true that with an ever-growing population, the people-to-health facilities ratio is a cause for concern. Those in need of urgent attention were areas like Gadap, Kemari and Orangi. Also, Saddar, Jamshed town, SITE and Liaquatabad are areas that are under a constant threat from high temperature.
Disclaimer: None.
Conflict of Interest: None.
Source of Funding: None.
References
1. Masud MM, Akhtar R, Afroz R, Al-Amin AQ, Kari FB. Pro-environmental behavior and public understanding of climate change. [Online] [Cited 2013 September 13]. Available from: URL: https://link.springer.com/article/10.1007/s11027-013-9509-4.
2. Pakistan Bureau of Statistics, Province wise provisional results of census, 2017. Statistics Division, Government of Pakistan. [Online] [Cited 2017 June 12]. Available from: URL: http://www.pbs.gov.pk/sites/default/files/PAKISTAN%20TEHSIL%20WISE%20FOR%20WEB%20CENSUS_2017.pdf
3. Ebrahim ZT. Is Karachi ready for the next big heatwave. [Online] [Cited 2019 Jul 26]. Available from: URL: https://www.dawn.com/news/1395438/is-karachi-ready-to-fight-the-next-big-heatwave.
4. Network CDK. Living with heatwave, the people of Karachi need a heat plan. [Online] [Cited 2016 February 03]. Available from: URL: https://cdkn.org/2016/02/opinion-living-with-heatwaves-the-people-of-karachi-need-a heat-plan/?loclang=en_gb.
5. Chaudhry Q, Rasul G, Kamal A, Ahmad Mangrio M, Mahmood S. Technical report on Karachi heat wave June 2015. Government of Pakistan Ministry of Climate Change, Ministry of Climate Change, Pakistan. [Online] [Cited 2015 July 15]. Available from: URL: http://www.ndma.gov.pk/files/heatwave.pdf
6. Neteler MG, Rocchini D, Delucchi L, Metz M. Massive data processing in GRASS GIS 7: A new gap-filled MODIS Land Surface Temperature time series data set. In FOSS4G-Europe. [Online] [Cited 2014 July 12]. Available from: URL: http://www.foss4g-e.org/content/massive-data-processing-grass-gis-7-new-gap-filled-modis-land surface-temperature-time handle
7. Crosson WL, Al-Hamdan MZ, Hemmings SN, Wade GM. A daily merged MODIS Aqua-Terra land surface temperature data set for the conterminous United States. Remote Sens. Environ.2012; 119: 315-24.
8. Clifford N, Cope M, Gillespie T, French S. Key methods in geography: Sage; 2016.
9. Oke TR. The energetic basis of the urban heat island. Quart J R Met Soc. 1982; 108:1-24.
10. Clinton N, Gong P. MODIS detected surface urban heat islands and sinks: Global locations and controls. Remote Sens. Environ. 2013; 134: 294-304.
11. Zhou D, Zhao S, Liu S, Zhang L, Zhu C. Surface urban heat island in China's 32 major cities: Spatial patterns and drivers. Remote Sens. Environ. 2014;152: 51-61.
12. Pareeth S, Delucchi L, Metz M, Buzzi F, Leoni B, Ludovisi A, et al. Inter-sensor comparison of lake surface temperatures derived from MODIS, AVHRR and AATSR thermal bands. In: The 35th EARSeL Symposium European Remote Sensing: progress, challenges and opp; 2011, pp 110-11.
13. Metz M, Rocchini D, Neteler M. Surface temperatures at the continental scale: tracking changes with remote sensing at unprecedented detail. Remote Sens. Environ. 2014; 6:3822-40.
14. Hajat S, Vardoulakis S, Heaviside C, Eggen B. Climate change effects on human health: projections of temperature-related mortality for the UK during the 2020s, 2050s and 2080s. J Epidemiol Community Health. 2014; 68:641-8.
15. Mahboob MA, Atif I, Iqbal J. Remote sensing and GIS applications for assessment of urban sprawl in Karachi, Pakistan. Sci Technol Dev. 2015; 34:179-88.
16. Lankao RP, Qin H, Cordova MB. Exploration of health risks related to air pollution and temperature in three Latin American cities. Soc Sci Med. 2013;83:110-8.
17. Weng Q. A remote sensing? GIS evaluation of urban expansion and its impact on surface temperature in the Zhujiang Delta, China. Inter J Remote Sensing. 2001; 22:1999-2014.
18 Commissioner Office, Karachi, personal communications; CDKN (2015). Karachi Heat Wave 2015: A Visual Guide.
19. Ghumman U, Horney J. Characterizing the impact of extreme heat on mortality, Karachi, Pakistan, June 2015. Prehosp Disaster Med. 2016; 31:263-6.
20. Kazmi AH, Jan MQ. Geology and tectonics of Pakistan. Pakistan: Graphic publishers, 2017.
21. Global Administrative Areas. GADM database of Global Administrative Areas, version 2.0. [Online] [Cited 2012 May 25]. Available from: URL: www.gadm.org.
22. Elledge J. Where are the largest cities in the word? [Online] [Cited 2015 March 13]. Available from: URL: https://www.citymetric.com/skylines/where-are-largest-cities-world-1051.
23. GAWC City Link Classification. [Online] [Cited 2018 Febuary 20]. Available from: URL: https://www.lboro.ac.uk/gawc/world2018link.html
24. Town Maps Karachi. [Online] [Cited 2017 January 17]. Avaiable from: URL: http://www.kda.gos.pk/Contents.aspx?id=94.
25. Wan Z, Hook S, Hulley G. MOD11A1 MODIS/Terra Land Surface Temperature/Emissivity Daily L3 Global 1km SIN Grid V006 [Data set]. NASA EOSDIS Land Processes DAAC.[Online] [Cited 2019 August 21]. Available from: URL: https://doi.org/10.5067/MODIS/MOD11A1.006
26. Dwyer J, Schmidt G. The MODIS reprojection tool. In Earth science satellite remote sensing. Berlin: Springer, 2006; pp-162-77.
27. Netler M. & Bowman H. (2015) GRASS GIS: r.fillnulls module NOT FOUND
28. Wan Z. MODIS land surface temperature products users’ guide. University of California, USA: Institute for Computational Earth System Science, 2006.
29. Bright EA, Coleman PR, Rose AN, Urban ML. Land Scan 2011. Oak Ridge, TN: Oak Ridge National Laboratory SE, 2010.
30. Health facilities maps. [Online] [Cited 2017 January 03]. Available from: URL: http://www.alhasan.com/sites/default/files/Karachi%20Health%20Facility.pdf
31. Open street maps Planet dump. [online]2017[cited 2018 dec 12]. Available from:URL: https://planet.openstreetmap.org/
Journal of the Pakistan Medical Association has agreed to receive and publish manuscripts in accordance with the principles of the following committees:
