1 Introduction

Floods are the commonly occurring natural hazards. They are a serious threat and cause damage to human life, property, and economy [1]. Floods at large damage built land, causes displacement of people, and affects infrastructure. Which in turn causes financial losses to the affected city/country [2, 3]. It is believed that more than 75 million people have been badly affected by floods across the world [3]. Extreme floods also affect physical and mental health, during and after the floods causing morbidity [4] carried out an analysis of 38 Organization for Economic cooperation and development (OECD) member countries and highlighted that post flood psychological effects are adverse especially in old age. While floods are primarily triggered by extreme rainfall, approaches in urban development, local geomorphology, condition of stream network and human activities also contribute the runoff and thus damages [5,6,7].

In comparison to developed countries, the risk of exposure to urban flooding is more in developing countries because of various anthropogenic and socioeconomic factors [8, 9]. One of these factors is the higher rate of urban growth [3, 10]. This brings the problems of poverty, land management, conversion of agriculture land to built-up areas, transportation and water supply, degradation of natural drainage system, and environmental issues [11,12,13,14,15]. Highlighting the need for a sustainable urban development plan, especially in mega cities. [16] suggested that urban development needs to be in line with sustainable development approaches. [17, 18] believe that governance by local authorities and pan urban administration is a key in controlling and managing urban development. [19, 20] are of the opinion that public participation plays a vital role in modern urban planning and sustainable growth. [21] found that most of the stakeholders share the opinion that natural conveyance systems needs to be maintained while developing an area, while [22] found that public government interaction in this regard over the time have improved, and the two-way interaction with the advent of modern communication technologies shall allow, a wholistic “development future”. It is believed that streams, and frequent water flow in them even by floods are accelerators for vegetation growth [23]. Which have been found important by [24] for effective flood management.

In Pakistan, between 1950 and 2010, more than 8000 people lost their lives, more than 100,000 villages were affected, and Pakistan suffered from economic loss of around $19 Billion [25]. In 2022, Pakistan was devastatingly affected by floods due to torrential rainfall. These floods affected Swat valley in the north and Sindh and Baluchistan province in south of the country. They affected more than 6 million people and damaged the communication infrastructure and led to spread of various infectious diseases [26]. Being its mega coastal city, Karachi has also suffered from floods from time to time. In 1977, severe flood in Malir and Lyari River caused loss of more than 200 lives while causing damage of 5 billion Rupees. Exponential urban sprawl, environmental degradation, increased in impervious surfaces and subsequent disruption of natural drainage system are major causes of urban flooding in the city [12, 27, 28].

When a multitudes of growth factors coexists, they lead to rapid urbanization [29]. Population growth, between and within cities migration, business and investment opportunities, commercialization and access to lifestyle amenities are the main contributing factors. Increase in built area leads to an increase in impervious surface, blockage and the lack of sufficient drainage system increases the rate of surface runoff thus causing floods [30,31,32]. Due to exponential urban growth, especially in a mega city, various changes in geomorphology are unavoidable. Due to these geomorphological landscape changes and industrialization because of transition of economies from rural to urban in over 200 years, significant rise to urban flooding has been seen [10, 33,34,35]. Moreover, the urban growth drivers are believed to be spatially dependent [36, 37]. It is believed that local urban sprawl has a positive correlation with urban sprawl of neighboring localities. This later has been prominent in the case of Karachi, where suburban growth has been multifold. Leading to built area increase along its natural stream channels, which is a cheaper alternative for agriculture and crop production and livelihoods [12, 13, 38,39,40]. Over that, these waterways have been poorly managed, and there are limited checks on encroachments along the stream banks [13]. Several authors emphasize that only structural solutions are not sufficient without proper solid waste management. Inadequate drainage capacity followed by poor management of urban drainage leads to urban flooding [41, 42]. It is considered one of the many anthropogenic factors causing urban flooding in Karachi [39]. In developing economies, urban planning directly conflicts with politics of planning and politicians [43, 44]. This conflict creates hurdles in following sustainable urban development plans in true form. Also, justified distribution of resources between different sections of local government become difficult due to these political struggles [45]. Availability of funds is another critical aspect in urban development. Finance is a basic element of executing urban planning and to tackle the problems associated with rapid urban development [46]. Availability and proper use of finance is linked to poverty, empowerment, land, and governance [47]. Cities, where both approved and non-approved slums exist, allocation and distribution of funds for the later, is a challenge and leads to unsustainable growth. Sustainable urban planning plays an important role in mitigating floods. It plays a vital role in controlling several factors, that leads to devastating effects of floods [48]. Factors such as urban sprawl, spatial planning for waterways, green area, and storage facilities [49]. In case of Karachi this is a major issue, where vertical development has been limited to a certain small part of the city, and most of the city has overlooked the need for compact development [50]. Compaction of built structures can help reduce the stress of natural resources, water pathways and open spaces [51, 52]. As there is a necessity to create more space for natural flow channels of the rainwater or preserve them. For instance, in Karachi, many that exist are blocked by solid waste [12, 13, 15, 38, 40]. Urbanization management, when inadequate, can lead to loss of permeable surfaces that otherwise reduces the surface runoff and mitigates floods. Natural vegetation and its widespread distribution are another factor that plays a vital role and acts as an important component of sustainable urban planning [53], recommend the need for a paradigm shift in urban land management. This paradigm shift will need to be inclusive of the public participation in urban planning and in conserving a large part of the natural ecosystem. A key element in this regard being maintenance of the natural conveyance system for water [53].

Several studies [54,55,56] have been done and methods developed to examine flood damages, risks, and their assessment. These methods include use of remote sensing [54, 57, 58], detection of flood water using sentinel 1 Synthetic Aperture Radar (SAR) imagery [55, 56, 59, 60], fuzzy logic [61, 62], flood modelling [63,64,65], spatio-temporal population modelling [66,67,68], flood hazard index model [69], analytical hierarchy process (AHP) [70, 71], Artificial Neural Network (ANN) [67, 72, 73], and Analytic Network Process [74, 75]. Most of these studies focused on detection of either flood inundation or flood damage and hazard, based on an analysis of urban sprawl. However, these studies have not considered the loss of historical natural stream network that carry flood water and the consequences of its removal by built areas. Since, an improvement of mapping technique was not the approach, this study uses a simplified technique to extract flood water using Sentinel 1 SAR imagery linking the flood inundation to rapid urban sprawl and blockage or damage of natural stream network over the years.

Although numerous studies have been conducted to find the relationship between large scale urban development and urban flooding. But a critical research gap exists, for developing an understanding of urban growth in Karachi, leading to its natural flood discharge infrastructure.. The little research that has already been conducted in the study area of Karachi relies upon a single source i.e., optical satellite imagery. This study illustrates how improperly led planning, where natural drainage pathways were covered by housing schemes, has led to increased flooding. For this it uses, satellite remote sensing data, topographic map and historical non spatial records for ascertain the loss of natural stream pathways to rapid urban development and subsequent risk factors created.

2 Study area

Karachi is located at the coast of Arabia Sea; its geographical bounds are 24o42’ N to 25o37’ N and 66o42’E to 67o34’ E with total area of 3,527 km2 (Fig. 1). Being the only coastal metropolitan city of Pakistan, it serves as an international trade and transport centre. Two seasonal rivers namely Lyari River (from east to south) and Malir River (from north to southwest) run through it.

Fig. 1
figure 1

Map of the study area, Karachi, the most populated city of Pakistan. Karachi is administratively divided into districts as well as union councils. It can be noticed that most of the built area of the city is enclosed in ridges and piedmonts of the Kirthar ranges on the west. While on the east of Malir River are anticlines formations

Full size image

Karachi experiences dry, hot, and humid weather conditions throughout the year [40, 76]. Karachi has hot and humid Summer from April–September and mild winter from December–February. Short duration heavy rainfall occurs from July to September. Minimum temperature ranged from 16 to 20 °C. Maximum temperature varied between 23 and 38 °C, whereas average temperature has remained between 20 and 34 °C [77].

Although mostly dry, Karachi receives heavy rainfall under the effect of tropical storms from July – September. Karachi has experienced multiple episodes of urban flooding due to heavy rainfall. Humid conditions in Karachi are caused by winds from the Arabian Sea. Wind speed is highest in summer with south-west to west direction [78].

Since 1729, Karachi has become a trade hub for the region. In 1941, Karachi’s population was 436,000 with 3.8% growth rate. After independence of Pakistan in 1947, with more than 900,000 migrants from India, Karachi’s population growth rate increased to 9.4%. According to 1981 census survey, population of Karachi increased from 3.6 million in 1972 to 5.43 million in 1981 with average growth rate of 4.96% (1972–1981), with 95.7% population living in urban areas. With an average growth rate of 3.56% between 1981 and 1998, Karachi’s population rose to 9.85 million in 1998. Karachi’s population has grown to 14.9 million (at 2.6%) as of 2017) [79, 80]. Population trend over the years (from 1941 to 2017) is shown in Fig. 2.

Fig. 2
figure 2

Population Trend of Karachi over the years. Figure above the bar shows population in Millions for that respective year. The chart clearly shows a continuous increasing trend from 1941 to 2017. Population increase is almost double from 1981 to 1998

Full size image

Being an economic hub since 1729, Karachi has faced rapid urbanization in modern times. Urban sprawl of 8.35 km2 in 1946 rose to 785.45 km2 in 2006 and 820 km2 in 2010 [81]. In a Land Use Land Cover (LULC) mapping analysis by the European Space Agency (ESA) in 2018, it shows that only in a span of 12 years (2005–2017), there was 8% growth in urban extent in Karachi, with a change of 34% in urban context only [82].

With this development came a an increase migration to the city. The migrants mostly settled in squatter settlements, commonly known as Katchi Abadis. According to a report by Karachi Water and Sewerage Board, more than 50% population of Karachi lives in those settlements [83]. Karachi’s Strategic Development Plan 2020 has hoped to take the city towards a path of becoming a world class city with economically feasible growth with quality of life for its citizens [84].

3 Material and methods

Datasets selected and methodology (Fig. 3) followed in this study were formulated around the erasure of natural streams as a consequence of urban growth. Rainfall trends were mapped and used to select years of urban flooding. Rainfall intensity and duration is an important factor in urban flooding as excessive rainwater gets accumulated in low lying areas causing flooding situation [85]. Flood years identified based on literature and rainfall trends were further analyzed for areas flooded. Urban growth was mapped using historical topographic maps and satellite imagery. Topographic maps were a useful dataset when historical urban extent is required [86] and other dataset like satellite imagery or aerial photographs are not available. SAR imagery is an efficient source for flood water detection [87] Flood water and permanent water bodies were also mapped from SAR imagery. Permanent water bodies were masked where flood only pixels were required. For masking out permanent water bodies, Global Surface Water Explorer datasets were used [88, 89]. Change in urban extent was used to identify inundated areas, and how increase in urban extent contributes towards removal of natural drainage network system. Informal settlements were digitized using Google Earth Pro. It provided vital insight on how urban settlements alongside rivers and drainage system become vulnerable to damages during floods.

Fig. 3
figure 3

An illustration of the data used, and products drawn during our study. A variety data types were used since to ensure a holistic overview of the on-ground situation and damages due to floods

Full size image

3.1 Rainfall trend

The city of Karachi experiences both cyclonic and torrential rainfall. Therefore, a literature and online disaster database-based survey was done, and cyclonic rainfall as well torrential rainfall and their durations were tabulated (Table 1). The tabulated data shows that since 1998, major cyclones have hit Karachi with high repetitive frequency. Extreme rainfall events are also a recurring phenomenon. Most of these cyclones and rainfall events occurred during monsoon season. For climate historical monthly rainfall averages from the Pakistan meteorological division were acquired from the year 1960 to 2020. Since, the data is sparse and does not contain spatial distribution in the region, this was compensated for by ERA-5 (European Centre for Medium Range Weather Forecast (ECMWF)) re-analysis and Tropical Rainfall Measurement Mission (TRMM) and CHIRPS (Climate Hazard Group Infrared Precipitation) data. This information was then used for accessing relevant satellite remote sensing data and classification for flooded regions. Summary of datasets used in this study and map produced using these datasets are given in Table 2.

Table 1 Cyclones, and extreme rainfall events to hit Karachi since 1998, the information has been gathered from a variety of sources, that include published literature and reports
Full size table
Table 2 Published, satellite and other data used in this study
Full size table

3.2 Historical topographic records

Urban sprawl caused topographic changes that further contributed towards urban flooding. Some of these changes include the growth of slums and houses along the stream banks, removal, and blockage of natural streams and replacement of protective embankments for urban expansion. To understand the extent of effect urban growth has on natural stream removal, historical maps of the urban extent, stream network and embankments was needed. For this purpose, topographic maps from 1945 and 1970s topographic maps were used [90, 91]. These 1940s topographic sheets were developed by U.S Army Map Service at a scale of 1:250,000. While the topographic sheets of 1970s prepared by Soviet Military are at 1:200,000. Although cadastral maps and surveys are most valuable sources of information, these are not always available due to high costs or data sensitivity [92] or simply being too historic. In this situation, topographic maps are a good source of information for mapping various land cover classes. Topographic maps of 1944 and 1980 were georeferenced and digitized. Contour lines, spot elevation, stream network, roads network, embankments, green areas, and built-up area were digitized from these sheets.

Urban extent, stream network and embankments acquired from these topographic map sheets were further used in the analysis to check their present condition. Urban extent was analysed as the main precursor to floods. Blockage of stream network due to rapid urban development was analysed to explain how they also contributed towards urban flooding. Mapped embankments that were present in the low-lying region of Karachi before an urban boom showed how there was no resistance to urban floods in present time. Urban settlements along the streams of Karachi, like Malir and Lyari Naddi were also digitized from recent google earth images [93].

3.3 Landsat imagery

Historic urban development was mapped using topographic map sheets as explained in Sect. 3.2. For the more recent urban development, Landsat satellite remote sensing data for year 2000, 2008, 2015 and 2020 was used. From topographic map sheets and Landsat data, LULC classifications were developed. Before supervised classification, Landsat imagery was corrected for top of atmospheric reflection (TOAR). After atmospheric correction, imagery was stacked, mosaiced and clipped to study area. Landsat 8 Images for the year 2015 and 2020 were pan sharpened using band 8. For this, high pass filter (HPF) resolution merge tool was applied in ERDAS IMAGINE using technique provided by [94]. Supervised classification technique was then used to extract urban, water and green areas for the selected years.

3.4 Sentinel 1 SAR imagery

Sentinel 1 SAR imagery is being commonly used in flood mapping, extraction of water surface, built-up area extraction etc. [95]. The data is easily available and suited to a variety of floods and surface water related studies [40, 96,97,98]. Sentinel 1 SAR products are weather and time independent (overall post even repeat time of 3 days) and can capture water bodies under the tree canopy [59]. Images from Sentinel—1 SAR were used for mapping flood extent of study area. Year 2015, 2017, 2019 and 2020 were selected as identified by rainfall trends and floods literature. Data was downloaded from National Aeronautics & Space Administration’s (NASA) Alaska Satellite Facility (ASF) portal maintained by Earth Observing System Data and Information System (EOSDIS). Sentinel—1 ground range detected (GRD) product acquired with interferometric wide (IW) swath mode was chosen for water extraction. Post flood dates for each year were selected to extract flood extent.

Sentinel 1 SAR data for flood years was used to extract flood extent of study area using Sentinel Application Platform (SNAP) software. Pre-processing of raw SAR data was performed before extraction of flood water. Pre-processing steps included (i) slice assembly to mosaic SAR imagery, (ii) orbit correction, (iii) radiometric calibration, (iv) single product speckle filtering, (v) geometric correction and (vi) land mask to clip imagery to study area. Purpose of pre-processing the imagery was to remove noise and geometric distortions and improve spatial precision [3, 60, 99]. For extraction of flood extent, general thresholding technique was used. Pre-processed SAR imagery was converted to decibels (dB) and threshold value for water extent was selected after various trails from imager histogram. This threshold value was used in Band Maths window using the Eq. 1, where image band corresponds to the name of clipped SAR band of study area and threshold value is the selected threshold from image histogram.

$$if \, image\_band \, < \, threshold\_value \, then \, 1 \, else \, NaN$$
(1)

3.5 Shuttle Radar Topography Mission (SRTM) DEM

SRTM DEM is one of the widely used elevation datasets used in various GIS & RS related studies. Elevation measurements in SRTM DEM are radar based, thus providing higher accuracy than other types of DEMs. In this study, SRTM DEM was used to create DEM of Difference (DOD) surface using DEM created from 1945 elevation data [100].

4 Results

This section has been organized as (i) an outlook of the general climate leading to heavy storms and rains in Karachi and its vicinity, followed by (ii) an analysis of the urban growth mapped from topographic maps as well as satellite remote sensing data (iii) an analysis of the removal of embankments and natural flood protections that were present before the modern development (iv) flood distribution and damage analysis in the last five years and (v) an overview of potential urban development by the regularly flooded and clogged streams of Karachi.

4.1 General climate leading to heavy storms and rains

A Mann–Kendall’s analysis (Fig. 4) of the rainfall trend in study area (using TRMM data from 1998) was carried out which suggests that the city for its western part has a loss of average rainfall (Fig. 4c) and its northern and north-eastern regions showing a slight increase. An analysis of Pakistan Meteorological Department (PMD) data for the city’s weather station records showed a similar decrease in rainfall (from 1960 to 2019) (Fig. 4d). An analysis of daily rainfall station data from 1960 to 2018 revealed that months of July and August are the wettest. The average rainfall from June to September was 11.09 mm, 73.18 mm, 52.55 mm, and 19.12 mm, respectively. Indicating a higher rainfall in the monsoon months only. Rainfall analysis was also carried out using CHIRPS dataset (Fig. 5). Which also indicated high rainfalls for year 2015, 2016, 2017, 2019 and 2020. After which mapping floods and analysis of their impacts for the years 2015, 2017, 2019 and 2020 was done. These years were selected keeping in view the availability of high-resolution optical satellite imagery, Sentinel 1 SAR imagery, rainfall trend analysis coupled with historical cyclone data.

Fig. 4
figure 4

TRMM (1998–2018) derived rainfall trend for South-eastern Pakistan, where our focus study area is located. Figure a represents the autocorrelation among the daily raster data used, b the P-value signifying trend and c is the Mann–Kendall trend map d trend of rainfall across Pakistan as calculated from data acquired from the Pakistan Meteorological Department. This trend calculation is based on data from 1960 to 2020. In figure (d) only stations with data available continuously since 1960 have been included)

Full size image
Fig. 5
figure 5

High peaks have been common in the monsoon season for Karachi in the past too. Regional estimation of rainfall for Karachi derived from global IR based CHIRPS (Climate Hazard Group Infrared Precipitation) data. Peaks for which the outcomes have been studied in this paper are marked in red

Full size image

4.2 Urban growth and development

Historical topographic maps of 1940s and 1980s were resampled to same spatial resolution of satellite remote sensing data i.e. 30 m. It is pertinent to mention that these historical topographic maps were produced at small scale. A small-scale map shows more area but less details as compared to large scale map. Map of 1944 was prepared at the scale of 1:250,000 whereas map of 1980 is at the scale of 1:200,000. Being small scale, these topographic maps do have some limitations. For example, map of 1980 does not show urban extent of northeast corner whereas it is marked in the map of 1944. Despite this limitation, the two maps of 1944 and 1980 provided useful information to further use in the study.

Urban extent vectorized from 1944 topographic maps in study area shows that in 1940s only 62.27 km2 of land was developed as urban region. This urban region grew up to 89.24 km2 by 1980 which is nearly 43% increase from 1944 (Fig. 6). However, in year 2000, this urban extent increased to ~ 345 km2, which amounts to a 286% gain from the previously vectorized records of 1980s. This increase continued up to 552.53 km2 by 2020 (Fig. 7) and amounts to a total of 380% increase from 1940s [15], found this later to be ~ 584 km2. Temporal urban growth trends both in terms of area and percentage change from year 1944 to 2020 is given in Table 3. Urban area %age was calculated by dividing urban extent of that year divided by total area of study area using formula given in Eq. 2.

$${\text{Urban Area }}\left( \% \right) = \left( {{\text{A}}\_{\text{i}} / \left( {\text{A}} \right)*{ 1}00} \right)$$
(2)

where Ai is urban extent of specific year and A is total area of study area, both in Km2.

Fig. 6
figure 6

Study area maps of the 1944–1980 topographic sheet based urban and green areas. The increase in urban area is ~ 29% before the major population expansion in 2000 to 2020. The river embankments seem to have been rapidly remove for urban development, hence lending the city susceptible to floods

Full size image
Fig. 7
figure 7

Urban built area as of 1940s (a), 1980s (b), 2000 (c), 2008 (d), 2015 (e) and 2020 (f) derived using Landsat 5 and Landsat 8 images. An outline of the growth with intermittent steps from the last two decades

Full size image
Table 3 Urban extent and growth of covered area derived from topographic maps and satellite data
Full size table

4.3 Removal of embankments and natural flood protections

Historic topographic map of 1944 shows protective embankments built throughout the study area for floods. As urban population and spread increased, these embankments and streams were gradually removed to make new urban settlements keeping in view the growing needs of the city. Removal of these embankments was assessed using DEMs of Difference (DOD) technique. It is used for calculating and analysing volumetric changes of a surface [101].The DOD technique simply subtracts latest DEM from old DEM producing a raster surface with negative and positive elevation values. Negative values show erosion of surface whereas positive values show deposition.

In this case, 1945 DEM and SRTM 2000 were used. 1945 DEM was created using digitized contours and spot elevations of 1945 topographic map. DOD was created by subtracting 30 m SRTM DEM of year 2000 from 1945 DEM. DEM created using 1945 elevation data was resampled to same resolution as of SRTM i.e. 30 m. Elevation range of DOD surface range from −153 to + 218 m. This DOD DEM was classified to positive and negative values. Positive values show deposition and negative values show erosion. Results show that from 1945 to 2000, 58% of area has been eroded and 42% of area had sediments deposited. After creating DOD surface, embankments present in 1945 topographic maps were overlaid on DOD surface. Overlay analysis showed that majority of those embankments were in areas that have now been eroded to create urban settlements (Fig. 8). Removal of these embankments was resulted into disturbance of natural topography of study area that further contributed towards urban flooding.

Fig. 8
figure 8

DEM of Difference (DOD) map of study area prepared using DEM of 1945 elevation data and SRTM DEM. Embankments of 1945 were overlaid that clearly indicates that majority of 1945 embankment were present in areas that have been eroded between years 1945 and 2000

Full size image

4.4 Flood distribution and damage analysis in the last five years

Urban floods extents from SAR datasets were extracted for 2015, 2017, 2019 and 2020 using Sentinel SNAP software. Permanent water bodies were masked. Flood extent of each 24 towns of Karachi were calculated. Flooded area in percentage for each town in each flood year was calculated (Fig. 10). Results show that core urban areas of Karachi was affected by urban flooding in 2015, 2017, 2019 and 2020. A comparison of urban development and floods in Karachi was drawn to check the relationship between urban extent and flooded areas during flood events. Both urban extent and flooded area showed a continuous increasing trend over the years. (Figs. 9, 10).

Fig. 9
figure 9

Floods maps of study area developed for each year and each towns using Sentinel 1 SAR data. Tens of neighborhoods are inundated during the floods in Karachi

Full size image
Fig. 10
figure 10

Comparison between urban extent and flooded areas in study area. Statistics show an increasing trend both in urban extent and floods between 2015 and 2020

Full size image

Statistics show that urban extent in 2015 expanded from 741 to 553 Sq. Km in 2020. Flooded area also followed increasing trend with 14.81 Sq. Km in 2015, and 64.3 Sq. Km in 2020. Overall, core urban areas faced an increase in inundation. Other, major urban settlements affected by flood water were present in Gadap town, followed by Korangi town, Faisal Cantonment, Shah-Faisal town, and Gulshan-e-Iqbal (Table 4).

Table 4 Area flooded in 2015, 2017, 2019 and 2020 per town in Karachi extracted from Sentinel 1 (SAR) Imagery
Full size table

4.5 Stream blockages

According to [NO_PRINTED_FORM] [13] most of the streams of Karachi remain clogged and lead to a greater damage during rains. Recent (2021) pictures acquired from Orangi in Karachi highlights the plight caused by poor sanitation and waste disposal measures by the town administration (Fig. 11).

Fig. 11
figure 11

Pictures of a drainage line, meant for sewage and water discharge is clogged by industrial waste. Most of the lines remain clogged due to poor management of waste and disposal. These pictures were taken near the Orangi Pilot Project office in Orangi Town

Full size image

As the city of Karachi was built to harbour business activities in the city and at port, a stream network was built in 1940s keeping in view the existing topography of the area. The purpose of these streams was to collect water from catchment areas and drain into Arabian Sea and in the process ensuring ground water recharge and urban supply for storage. Temporal analysis of natural stream network was carried out to identify how much of these natural streams have been blocked or removed over time. Urban extend of 2000–2020 extracted from satellite imagery was overlaid on 1940s natural stream network. Streams covered by built area were identified. Figure 12 shows how stream network of 1945 was replaced by expanding urban infrastructure.

Fig. 12
figure 12

Stream network blockage over time by overlaying urban extent from year 2000–2020 on natural stream network digitized from 1980s topographic map of Karachi

Full size image

Streams and change percentage covered by built area in each town of study from year 2000 to 2020 is presented in Table 5. From 2000 to 2008, 28% of originally built stream network of 1945 was removed or blocked due to outward urban sprawl. From 2008 to 2015, additional 12% stream network was disturbed and from 2015 to 2020, this ratio further increased to 16% in addition to previously removed streams. Of the places that have lost most natural streams to are Lyari, North Nazimabad and Saddar (most affected by recent floods), followed by Liaquatabad, Jamshed Town, New Karachi, Gulberg and Clifton Cantt as well as other towns with Katchi Abadis like Orangi and Korangi towns.

Table 5 Stream blockages because of urban development, for each town in Karachi
Full size table

The Malir River is fed by 25 smaller drains, while the Lyari River is fed by 7 smaller drains. While there are 41/46 major drains mapped as important source of flood water discharge. There are many confusions that remain on the part of the city’s administration to clarify. According to estimates 550 smaller feeder channels in the city that are not under the municipal administration and are supposed to be managed by the local district municipal administration. However, the state of repair, mapping and cleanliness is not clear hence, during major floods the uncertainty should remain. After 2020 floods, an extensive effort by provincial and federal governments was done to clean the major drainage network, however management of storm water is yet to be ascertained. An analysis of each town of Karachi and the effects of imperviousness showed that about 89% of the land classifies as impermeable in the study area of Karachi.

5 Discussion

Unplanned and uncontrolled urban sprawl, disturbance to natural drainage system and urban settlements along banks of natural drainage are main reasons of urban flooding [102]. Pakistan was severely impacted by floods in 2022 because of heavy precipitation. The northern Swat Valley and the southern provinces of Sindh and Baluchistan were both damaged by these floods. Over six million people were affected by these terrible floods, which also caused significant harm to the communication infrastructure and triggered the spread of several infectious diseases.

An analysis of the changes to urban landscape shows that barren land has been mostly build over. In the northwestern parts of the city however, urban growth has dominated natural stream covered channels (Figs. 6, 13). Also, green areas as well as coastal parts of the city have been built over. The green areas in places were observed to be the green alluvial parts along the riverbanks. It has also been observed that development in the last two decades has been in places, where floods tend to accumulate more. A higher correlation exists between the newly built settlements in open areas with the cumulative floods observed (Fig. 14).

Fig. 13
figure 13

An overview of the spread of urban growth between 2000 and 2020 and the distribution of floods in Karachi. Development of Urban Area has taken over on open patches, green areas and water areas both along the streams and coasts

Full size image
Fig. 14
figure 14

The coefficient of determination for the land area that was converted from open areas to Urban and Green to Urban. A stronger relationship exists between areas that have been converted from Open to Urban. This is because of that these areas were previously attributed to natural streams

Full size image

Karachi is mostly warm and dry with precipitation ranging below 250 mm/year on average. However, the monsoon months are the wettest. It is worth mentioning that monsoon begins towards the last week of July and ends towards the third week of August, over southern Pakistan with some exceptional rainfall towards eastern Sindh province in 2011 and 2012 [103]. Rainfall analysis brought out the fact that Karachi is experiencing short but intense rainfall after every 2–3 years.

Urban governance in Karachi has been a matter of grave concern. Of the many problems that Karachi is facing, the marginalization of its poor population is a key factor. Housing and a planned allocation for these communities is missing [104]. Others include ethnic diversity, political violence, and the development of informal economies [105,106,107], that tend to strain the governance structure of a city [108]. These later can be attributed to the post-colonial development trajectory of the city, where the aim was to portray it as a world class city and keeping an exclusive control on development along its central regions [109]. This led to unchecked growth in its peripheries. These communities led development along the natural streamways north of the central old city and landfilling of the existing stream pathways. Even now many marginalised communities inhabit its stream banks in the city [13, 27]. For instance, there is a growing population along the Gujjar Nala, Malir River, and Lyari Rivers. These communities from time to time are under the direct effects of floods. In places, where large swaths of lands have been transformed into planned societies, such as Nazimabad, North Nazimabad, Baldia and North Karachi. The natural flow paths of the rivers have been landfilled and flood water in the last few years had accumulated in the streets.

Zeiderman et al. [110] suggests that in case of Karachi, there is a challenging urban planning environment that exists. While there is necessity to shift paradigms and adopt a modernist urban planning approach. There is, a need to look beyond the Karachi Development Authority’s (KDA) approach at managing and developing more land. Since the organization is already embattled with various land ownership issues, rapid community growth and sprawling urban societies, that tend to overlook best resource allocation practices and guidelines. An urban resource centre, research on pressing issues of social inclusion, information base creation, monitoring, and modelling growth, both current and future are some of the steps that can be taken to allow for informed decision making [111, 112]. One key aspect of this can also be reestablishment of urban waterways, cleaning of clogged drainage lines and subsequent control on development along these channels. While rehabilitating the communities, in planned areas with adequate amenities and livelihood opportunities [113].

Karachi like many cities of the developing world and region is no different in terms of the issues, faced by it. The city’s growth is dominated by its informal economy which provides vital livelihood for its immigrants. Its, rapid, urban growth being a direct threat, to its sustainability and green spaces [114,115,116]. Also, a myriad of health issues, growing. Of these those related to post flood scenarios being the most common. Incidents of Malaria and Dengue are on the rise, due to conducive, breeding environment and its widespread presence. Water contamination related gastrointestinal issues, diseases of the eyes and skins have been commonly reported in the post flood periods [117]. For a city as big as Karachi one the issues remain solid waste management. Which is usually deposited by its informal settlements in the river channels [118, 119]. And even the waste collected by garbage disposal corporations, has a similar fate, leading to a mixing with floodwaters and higher level of contaminant percolation.

As compared to other studies, this research incorporated historical topographic maps in the flood analysis. Making those historical records a baseline, comparison was made to observe the changes in topography, urban development, and protective infrastructure. Analysis was done based on this comparison and results were drawn. While the development before 1980s was not large scale, there has been a massive increase in urban growth since then. By year 2000, Karachi had experienced 286% increase in urban extent that was caused by various factors like lack of oversight on management of land, absence of urban planning etc. Relationship between urbanization and economic growth should be positive, ideally [120]. However, in Pakistan, this relationship shows that GDP growth has decreased between 2015 and 2021 while urban population percentage is constantly increasing. Thus, increase in urban extent from 1980s to 2000 is massive as in compared to the size of the economy of Pakistan and the state’s ability to invest in sustainable development at such a massive scale. Unregulated and unchecked allocation of parcels for development of housing societies also contributed towards rapid sprawl. Temporal LULC analysis shows that increase in urban extent was mostly at the expense of decrease in agricultural land. This type of urban extent and downward natural slope of core urban area (Fig. 1) rendered this urban development susceptible to the negative effects of urban floods.

Rapid urbanization also affected city’s natural drainage system. This natural drainage/stream network was either removed or damaged due to rapid urban development over the years. Both urban development (Table 3) and stream network covered or damaged by urban development (Table 4) shows an increasing trend. By year 2020, Stream network in 12 out of 24 towns is more than 70% as compared to more than 50% in year 2000 (Table 4). Most of the streams have been removed and the little natural streams that are left, are clogged with sewage and industrial and household waste. This continue to be a major problem as water does not get its natural flow and tends to inundate the roads or houses nearby these drains. Analysis from 2000 to 2020 showed that during urban floods, same areas are inundated repeatedly, where stream blockage/damage is maximum. Although the drainage network has been cleaned many times, especially after year 2020, but this temporary measure needs to be replaced with some permanent structural changes.

Urban settlements developed on banks of drainage/stream networks are a major concern during floods. Due to their proximity and structure, they are badly affected during each urban flood event. Analysis showed that these settlements are increasing at a high rate, and block flood water flow towards Arabian Sea. Major urban settlements affected by flood water were present in Gadap town, followed by Korangi town, Faisal Cantonment, Shah Faisal town and Gulshan-e-Iqbal. These settlements need to be relocated to alternate locations to provide easy passage to flood water.

Data availability and dissemination is a major issue in Pakistan. In the absence of modern data keeping, dissemination and analysis by the authorities. It was necessary to look for alternatives, such as published public map records and satellite remote sensing data. Moreover, statistical records and data on streams, stream flow rates, environmental data such as rainfall and temperature records are extremely limited, and some are impossible to find. Published records are mostly in the form of news archive, but they seldom have a spatial component associated. It would be therefore important to lead this work into databasing all forms of published records on urban growth and make it available to the research community. Another important analysis can be the effect of the loss of natural water conveyance on disease prevalence and urban heat accumulation.

6 Conclusion

Urban Flooding in Pakistan is a recurring phenomenon. Raid urban sprawl and poor maintenance of natural drainage system make flooding further worse necessitating identification of flood prone urban areas during floods. LULC information show that Karachi experienced rapid urban grown over the years. This rapid sprawl linked with poor management of water channels cause urban flooding in the city of Karachi during short but heavy rainfall. SAR imagery is an effective data source for detection of flood water in urban areas as it is not affected by weather conditions thus providing better identification of flooded areas.

Using thresholding, flood water was extracted from SAR imagery. LULC classification was carried out using Landsat imagery. This classification information shows that natural streams and flood protection structures were replaced with urbanization with the passage of time. This stream blockage is almost 100% in core urban areas of the city. When overlaid both LULC information and urban flood water, analysis showed that urban development has blocked stream network of the city. Due to this blockage, low lying urban areas face flood inundation during rainfall. It was also observed that town with the highest number of natural streams filled up are the ones that were worst affected recently. Examples are that of Lyari, North Nazimabad, Saddar, Liaquatabad, Jamshed Town, New Karachi, Gulberg, Clifton Cantt and Orangi. A visual analysis of topographic sheets showed several embankments along these streams. And we assume that the streams must have been filled, with soil from the very embankments. Hence leading to easy and cheap preparation of land for construction.

One of the limitations of the study was small scale historical topographic maps. Although information extracted from these historical records like elevation, urban extent, stream network and embankments etc. was effectively utilized in the analysis, however same analysis could have been performed with more accuracy if large scale historical records were available. Sentinel 1 SAR imagery is an excellent source to extract water extent of an area. In this study, open-source data of 10 m resolution was used. Better water extraction can be done if higher resolution SAR imagery was available, which is mostly not available as open source.