You are on page 1of 16

Estuarine, Coastal and Shelf Science 249 (2021) 107128

Contents lists available at ScienceDirect

Estuarine, Coastal and Shelf Science


journal homepage: http://www.elsevier.com/locate/ecss

Evaluating mangrove conservation and sustainability through


spatiotemporal (1990–2020) mangrove cover change analysis in Pakistan
Hammad Gilani a, *, Hafiza Iqra Naz a, Masood Arshad b, Kanwal Nazim c, Usman Akram b,
Aneeqa Abrar a, Muhammad Asif b
a
Institute of Space Technology (IST), Islamabad, Pakistan
b
World Wide Fund for Nature - Pakistan (WWF-Pakistan), Pakistan
c
Society for Environment and Mangrove Protection Welfare Association (SEMPWA), Karachi, Pakistan

A R T I C L E I N F O A B S T R A C T

Keywords: This study provides the first comprehensive mangrove cover change assessment from 1990 to 2020, at five-year
Mangrove in Pakistan intervals, across all five mangrove areas in Pakistan, i.e. Indus Delta, Sandspit, Sonmiani, Kalmat Khor, and
Google Earth Engine (GEE) Jiwani. Using the Google Earth Engine (GEE) geospatial cloud computing platform, Random Forest (RF) classifier
Random Forest (RF)
was applied on Landsat 30 m spatial resolution satellite images to classify three major land cover classes:
Mangrove fragmentation
‘mangrove’, ‘water’ and ‘other’. High temporal and spectral resolutions of Landsat images, with a low saturation
Mangrove conservation
level of spectral bands with the integration of indices, are the main factors that ensured >90% overall accuracy of
land cover maps. Over the last three decades (1990–2020), the annual rate of change calculation, cross-tabulated
method, and fragmentation analysis were carried out to identify the changes in mangrove cover. Overall, the
findings of this paper revealed that, at the national scale, an estimated 477.22 km2 was covered with mangrove in
1990, which increased to 1463.59 km2 in 2020, a 3.74% annual rate of change. Mangrove fragmentation
mapping results have also showed enhancement in mangrove tree canopy density. Due to planting and effective
conservation practices, the current study shows positive changes in mangrove cover across all five study sites.
The findings of this study will prove useful for design and implementation of mangrove ecosystem management
plans, initiatives for adaptation to extreme weather events, carbon budgeting, and others.

1. Introduction and marine resources.


Despite the ecological value of mangroves, they are significantly
Mangroves are one of the most productive ecosystems, which offer threatened and vulnerable to climate changes (e.g. sea level rise, vari­
worthy services for climate change adaptation and mitigation. They ations in ocean current, etc.), natural disasters (e.g. tsunami, cyclone,
protect coasts from natural disasters (e.g. tsunami, cyclone), help clean earthquake, etc.) and anthropogenic exploitation (e.g. pollution, silta­
the creeks and channels, and act as a buffer against sea intrusion and soil tion, urban development, over-harvesting, construction of shrimp
erosion (Khan et al., 2010; Marois and Mitsch, 2015). They recycle ponds, and grazing, etc.). According to the most reliable estimates, at the
nutrients from livestock and human settlements (Adhikari et al., 2010). global scale, mangroves are being lost at an alarming rate of 2% per
Mangroves have the potential to sequestrate and store substantial year, with 0.15–1.02 Pg (billion tons) of carbon dioxide being released
amounts of carbon, approximately 18 times greater than other forest annually, resulting in $US 6–42 billion annual economic damages
types (Alongi, 2012; Donato et al., 2011; Sanderman et al., 2018). Like (Pendleton et al., 2012). Alarmingly, the Intergovernmental Panel on
other wetlands, mangroves are important components of the water Climate Change (IPCC) has predicted that in the following 100 years,
cycle, as they absorb excess water flow during the time of floods. The ~30–40% of the coastal environment and 100% mangrove forested
primary productivity of mangroves in deltaic areas is seven times higher areas could vanish if the prevailing rate of loss continues (Mackay,
than coastal areas without mangroves (UNEP-WCMC, 2014). The Sus­ 2008).
tainable Development Goal (SDG) 14, “Life Below Water”, emphasizes Pakistan’s coastline is 1050 km long and 40–50 km wide zone,
restoring and protecting mangroves for sustainable use of oceans, seas, distributed between the Sindh (350 km) and Balochistan (700 km)

* Corresponding author.,
E-mail addresses: hammad.gilani@mail.ist.edu.pk, hammad.gilani@grel.ist.edu.pk (H. Gilani).

https://doi.org/10.1016/j.ecss.2020.107128
Received 5 June 2020; Received in revised form 24 November 2020; Accepted 25 November 2020
Available online 30 November 2020
0272-7714/© 2020 Elsevier Ltd. All rights reserved.
H. Gilani et al. Estuarine, Coastal and Shelf Science 249 (2021) 107128

Fig. 1. The five mangrove study sites of Pakistan: 1. Indus Delta, 2. Sandspit, 3. Sonmiani Khor (MianiHor), 4. Kalmat Khor (KalmatHor), and 5. Jiwani (Gwa­
dar Bay).

provinces, and geographically placed between 24◦ and 25◦ N Latitude offer new possibilities for global developments and societal benefits.
and 61◦ –68◦ E longitude (Abbas et al., 2013; Amjad and Jusoff, 2007). In
1958, mangroves of Pakistan were declared “protected forests” under 1.1. Literature review – mangrove mapping in Google Earth Engine (GEE)
the Pakistan Forest Act 1972, and the entire forest areas with the water
channels present therein were declared “wildlife sanctuaries” in 1977 The Google Earth Engine (GEE) is one of the cloud-based platforms
under the Sindh Wildlife Safety Ordinance of 1972 Act (Rafique, 2018; which allows algorithms development and provides a convenient
Saeed et al., 2019). In Pakistan, mangroves are distributed in five mechanism for scientists to share, visualize and analyze remote sensing
separate, unconnected geographical pockets: Indus Delta and Sandspit data and by-products (Gorelick et al., 2017; Kumar and Mutanga, 2018).
in Sindh province; and Sonmiani Khor (Miani Hor), Kalmat Khor (Kal­ A number of researchers have utilized GEE for the assessment and
mat Hor), and Jiwani (Gwadar Bay) in Balochistan province (Fig. 1). reporting of mangrove dynamics.
According to the United Nation’s Food and Agriculture Organization In GEE, Chen et al. (2017) developed a new classification algorithm
(FAO), on average natural tree cover in Pakistan diminishes by about based on the biophysical characteristics of mangroves in China using
316 km2 annually (FAO, 2010). In an effort to counter deforestation and 2015 Landsat-7/8 and Sentinel-1A imagery. Portengen (2017)
forest degradation, and to ensure a climate-resilient future for Pakistan, completed a master’s dissertation on classifying mangroves in Vietnam
the Government and private sector have taken several small to using Sentinel-1 and 2 satellite datasets in GEE platform, achieved with
large-scale plantation initiatives across the country to improve the tree up to an overall accuracy of 87% of Random Forest (RF) machine
cover (Kamal et al., 2019). learning classifier. Mondal et al. (2017) estimated mangroves extent on
Earth observation satellite datasets and geospatial tools offer a the coast of Sierra Leone using Landsat datasets (1990, 2000, 2010 and
unique possibility for quantifying changes occurring on the earth’s 2016) by adopting k-means unsupervised classification approach using
surface, whether through human impact or climate change (Wulder GEE. Pimple et al. (2018) carried out mangroves mapping study over the
et al., 2019). Temporal satellite data is a viable solution for the quan­ Trat province, Thailand and surrounding areas to analyze three decades
tification of land features in a comprehensive and systematic manner. Landsat imagery (1987 and 2017) in GEE. Based on the RF classifier,
With the advantage of freely available earth observation satellite images they obtained 87% and 96% overall accuracy for the years 1987 and
[e.g. MODIS (Moderate Resolution Imaging Spectroradiometer), Land­ 2017 respectively (Pimple et al., 2018). Mondal et al. (2019) evaluated
sat, Sentinel, etc.] and Analysis Ready Data (ARD) products [e.g. two machine learning algorithms, RF, and Classification And Regression
Normalized Difference Vegetation Index (NDVI), Normalized Difference Trees (CART) in GEE using Sentinel-2 along the coast of Senegal and
Water Index (NDWI), Land Surface Temperature (LST), etc.], it becomes Gambia, West Africa. They found for the mapping of mangrove cover, RF
much easier and transparent to quantify and report near real-time is better than CART, with an overall accuracy of 93.44% ± 1.37% and
changes occurring on the earth’s surface. In forestry and particularly 92.18% ± 1.29% respectively (Mondal et al., 2019). In GEE, for the
in mangroves, satellite remote sensing is frequently being used to mangrove cover mapping of entire Cambodia, Tieng et al. (2019)
evaluate changes in cover, density dynamics, and biomass budgeting applied RF classifier on Landsat-8, Sentinel-2 and Google Earth derived
(Elmahdy et al., 2020; Giri et al, 2011, 2015; Kathiresan et al., 2013; very high-resolution images. To study the status of the Brazilian man­
Kauffman and Bhomia, 2017; Kuenzer et al., 2011; Mondal et al., 2019; groves, Diniz et al. (2019) analyzed three decades (1985–2018) of
Simard et al., 2006; Vo et al., 2013). Open access of big geospatial Landsat imagery from GEE environment by computing the Modular
datasets, cloud computing platforms, and machine learning algorithms Mangrove Recognition Index (MMRI). Li et al. (2019) investigated the

2
H. Gilani et al. Estuarine, Coastal and Shelf Science 249 (2021) 107128

plant phenological trajectory for mangroves species mapping in GEE current study is designed to achieve the following objectives to assess
using Sentinel-2 derived time series NDVI over the 2.5 km2 Zhangjiang the mangrove conservation and sustainability over the five mangrove
estuary in Fujian Zhangjiangkou National Mangrove Nature Reserve areas in Pakistan.
(FZNNR), China.
• Detecting changes in mangrove cover using Landsat 30 m images at
1.2. Literature review – mangrove mapping in Pakistan five-year intervals (1990, 1995, 2000, 2005, 2010, 2015 and 2020),
across all five mangrove areas in Pakistan.
In Pakistan, only a handful of studies have been conducted to assess • Mangrove fragmentation analysis, to estimate changes over thirty
the changes in mangrove cover using satellite data or field-based in­ years within the mangroves ecosystem.
ventory with statistical estimations.
Masood et al. (2015) assessed mangrove cover changes over the 2. Materials and methods
Indus Delta (~6200 km2 total area taken in this study) between 2009
and 2014, using Landsat-5 and 8 satellite images respectively (30 m 2.1. Study sites
spatial resolution respectively), by comparing pixel-based supervised
maximum likelihood classification and onscreen digitization techniques. In Pakistan, mangroves exist at the following five sites: Indus Delta,
Giri et al. (2015) studied Indus Delta mangrove cover change in terms of Sandspit, Sonmiani, Kalmat Khor and Jiwani, which have been selected
gain and loss using 1973 Landsat MSS (60m) and 2010 Landsat TM (30 for this study (Fig. 1). The details of the sites are:
m) images by adopting supervised and object-based classification tech­ 1. Indus Delta (central coordinates: 23.957 N latitude, 67.5015 E
niques respectively. Abbas et al. (2013) used Advanced Land Observa­ longitude, geographical area: 8000 km2), situated in Thatta district,
tion Satellite (ALOS) and Advanced Visible and Near Infrared Sindh province. It comprises 17 major creeks and numerous minor
Radiometer type 2 (AVNIR-2) 10 m spatial resolution satellite images for creeks (Amjad and Jusoff, 2007) and constitutes ~95% of total estuarine
the year 2009, to assess nationwide mangrove cover through mangroves situated in Pakistan (Giri et al., 2015). It is the 5th largest
object-based classification technique. WWF-Pakistan (2005) conducted delta and 7th largest mangrove forest in the world (Memon, 2005; Zafar
a study at Keti Bandar (a small portion of the Indus Delta), Sonmiani Khan and Akbar, 2012). Since 1972, Keti Bandar, Shah Bandar, and Cut
Khor, Kalmat Khor and Sandspit sites using Landsat TM (30 m), Manarki Chach in the Indus Delta have been designated Wildlife Sanc­
Advanced Spaceborne Thermal Emission and Reflection Radiometer tuaries. In 2002, the Indus Delta listed as a one of Ramsar sites in
(ASTER) Terra instrument (15 m) and QuickBird (0.5m) images through Pakistan. In the Indus Delta, ~95% area is covered by Avicennia marina
pixel-based supervised maximum likelihood classification and onscreen while Rhizophora mucronata, Ceriops tagal and Aegiceras corniculatum can
digitization land cover mapping techniques. Prior to this, in 1983 be observed. The mangroves of the Indus Delta have been declared the
through onscreen digitization at the scale of 1:250,000, the Space and largest arid climate mangroves site in the world (Saifullah, 2017). The
Upper Atmosphere Research Commission (SUPARCO) of Pakistan re­ freshwater of the Indus River feeds Indus Delta’s mangroves (Barkati
ported the mangrove cover of the Indus Delta using Landsat MSS 60 m and Rahman, 2005).
spatial resolution datasets of 1978 (IUCN, 2005; WWF-Pakistan, 2005). 2. Sandspit (central coordinates: 24.8399N latitude, 66.9478E
For national level mangrove cover assessment, another study was con­ longitude, geographical area: 68 km2) is situated in the southwest of
ducted by SUPARCO using the 20 m spatial resolution SPOT multi­ Karachi, Sindh province. It is a popular tourist spot and its beach is also a
spectral imagery of the year 2003 through digitization at 1: 50,000 key nesting ground for Olive Ridley and Green marine turtles (Durranee
scale. In 1999, the Forestry Sector Master Plan (FSMP) land cover map et al., 2008). There are monospecific stands of Avicennia marina in the
produced from Landsat images of 1988–1991 through onscreen digiti­ area. The movement of cargo ships and other maritime transport
zation technique, while the National Forest and Range Resources through the Karachi port disturb the biodiversity of this area (Saher
Assessment Study (NFRRAS) was done through pixel-based image clas­ et al., 2019). Except during the rainy season (July–September) when
sification of Landsat images of 1997–2001 (Abbas et al., 2013). Apart there is larger water volume, the Lyari river flowing through the Sand­
from these assessments, remote sensing, field-based, in-situ and statis­ spit study region is heavily polluted with industrial effluents and solid
tical assumption based number of mangrove cover studies conducted waste from settlements (Barkati and Rahman, 2005; Durranee et al.,
between 1980 and 2005 were compiled and reported in a tabular form in 2008; Khan et al., 2010; Sultana and Mustaquim, 2003).
the FAO (2007) mangroves country report. 3. Sonmiani Khor (Miani Hor) (central coordinates: 25.5343N lati­
tude, 66.3945E longitude, geographical area: 830 km2) lies in Lasbela
1.3. Study objectives district, Balochistan province. It is a ~60 km long and ~7 km wide
coastal lagoon, ~5 km-wide towards the Arabian Sea. In Balochi lan­
Based on the literature review on mangrove cover mapping in guage, the term ‘Khor’ or ‘Hor’ is used for water channels. In 2001, this
Pakistan, it was observed that most of the studies incorporated in-situ site was declared a Ramsar site (Amjad and Jusoff, 2007). Porali and
information in satellite images classification and have used different Windor, two seasonal rivers flow south into the study area from the
spatial resolution satellite images, image classification methods, defi­ mostly desert areas of Balochistan province (Mirza et al., 1988; Rasool
nitions and classification schemes, which provided contradictory results. et al., 2002). Avicennia marina, Rhizophora mucronata, and Ceriops tagal
The literature review also revealed that only a few studies have analyzed are found in this site. Fishing is the main occupation of the local com­
one-time satellite imagery to assess and map the status of mangrove munities throughout the year (Saher et al., 2017).
cover. However, a number of studies used temporal satellite datasets to 4. Kalmat Khor (Kalmat Hor) (central coordinates: 25.4105N lati­
assess mangrove cover changes. As yet no comprehensive, systematic, tude, 64.0827E longitude, geographical area: 560 km2) in Gwadar dis­
well-recognized nationwide temporal assessment has been carried out. trict, Balochistan province. The Basol river drains in the Kalmat Khor
Further, there has been very little focus on spatial analysis on mangrove coastal lagoon and ultimately into the Arabian Sea. Several tidal chan­
cover towards understanding the mangrove ecology. Indeed, globally nels exist in this area, which are filled with seawater during spring tides.
produced one-time mangrove databases are openly and freely assess­ This site is a ~7–19 km long and ~2–27 km wide lagoon, which re­
able, but at the national scale, spatiotemporally produced harmonized sembles the shape of a tree (Rasool and Saifullah, 1996). Avicennia
mangrove cover databases are needed for short to long terms strategic marina mangrove species is dominant in this area.
policy formations and implementation. 5. Jiwani (Gwadar Bay) (central coordinates: 25.1678N latitude,
By taking advantage of open access 30 m spatial resolution Landsat 61.7558E longitude, geographical area: 80 km2) in Gwadar district,
satellite images and RF machine learning algorithm in GEE platform, the Balochistan province. This area provides two major natural habitats, i.e.

3
H. Gilani et al. Estuarine, Coastal and Shelf Science 249 (2021) 107128

Table 1 m spatial resolution satellite images were utilized for mangrove cover
Landsat images used for mangrove cover mapping. mapping. All cloud-free images were selected for every five-year interval
Landsat Indus Delta Sandspit Sonmiani Kalmat Jiwani from 1990 to 2020. For this study, neap tide height, winter or dry season
sensor Khor Khor images were preferred in which the mangroves are visible and easily
Path/Row
differentiable from other land features. Four sites (Sandspit, Sonmiani
151/ 152/ 152/043 153/042 154/042 156/ Khor, Kalmat Khor, and Jiwani) are small enough to be covered within
043 043 043
one Landsat scene (185 × 185 km) while the Indus Delta is covered
Landsat-5 17- 10- 17-Feb- 12-Mar- 15-Feb- 13-Feb- through two scenes (Table 1).
TM Feb- Feb- 1990 1990 1990 1990
Since June 2003, the Landsat-7 ETM + sensor acquired and delivered
1990 1990
Landsat-5 08- 04- 04-Apr- 10-Mar- 12-Jan- 15- data, gaps (strips) caused by the Scan Line Corrector (SLC) failure.
TM Feb- Apr- 1995 1995 1995 Mar- Before classification, in GEE a morphological mean circle filter for the
1995 1995 1995 gap-filling technique was performed on Landsat-7 ETM + images (year
Landsat-5 13- 06- 13-Feb- 07-Mar- 26-Jan- 24-Jan- 2005), (code). For the classification, only optical 30 m spatial resolution
TM Feb- Feb- 2000 2000 2000 2000
2000 2000
spectral bands (visible and infrared) were selected by neglecting thermal
Landsat-7 18- 27- 18-Feb- 13-Mar- 16-Feb- 18- and Quality Assessment (QA) bitmask bands. In addition to optical
EM+ Feb- Feb- 2005 2005 2005 Mar- spectral bands, NDVI (Normalized Difference Vegetation Index)
2005 2005 2005 (Tucker, 1979; Tucker et al., 1979) and NDWI (Normalized Difference
Landsat-5 16- 23- 14-Jan- 15-Feb- 18-Nov- 19-Jan-
Water Index) (Gao, 1996) were computed and integrated into a classi­
TM Jan- Jan- 2010 2010 2009 2010
2010 2010 fication algorithm. Over the Indus Delta site, as mentioned above, two
Landsat-8 15- 14- 10-Mar- 13-Feb- 08-Mar- 18-Feb- adjacent Landsat images (151/043 and 152/043) had to be mosaicked.
OLI Mar- Jan- 2015 2015 2015 2015 Before land cover mapping, within GEE a spatial clipping operation was
2015 2015 performed on images to extract the exactly defined area of study sites.
Landsat-8 28- 19- 03-Jan- 11-Feb- 17-Jan- 31-Jan-
OLI Jan- Jan- 2020 2020 2020 2020
2020 2020 2.3. Methodology

The methodology of this study consists of land cover mapping, ac­


mangroves swamps and sandy beaches. In 2001, this site was listed as a curacy assessment, mangrove cover change assessment, and mangrove
Ramsar site, particularly important as a nesting ground of the endan­ fragmentation analysis (Fig. 2).
gered Olive Ridley and Green marine turtles. This swampy and marshy
site is covered with Avicennia marina. 2.3.1. Land cover mapping
In this study, over the stacked 30 m spatial layers (optical spectral
2.2. Landsat imagery and spectral indices bands, NDVI and NDWI), three major land cover classes, i.e. ‘mangrove’,
‘water’ and ‘other’, were identified and mapped through RF machine
In the GEE platform, Landsat-5 TM (Thematic Mapper), Landsat-7 learning classification algorithm in GEE. The ‘other’ land cover class
ETM+ (Enhanced Thematic Mapper Plus) and Landsat-8 OLI (Opera­ consists of dry and wet soil, algae, and grassland features. Land cover
tional Land Image) calibrated Top-Of-Atmosphere (TOA) reflectance 30 maps were made separately for five-year intervals for each mangrove

Fig. 2. Methodological flow chart diagram.

4
H. Gilani et al. Estuarine, Coastal and Shelf Science 249 (2021) 107128

area. Table 2
In GEE, to classify Landsat images, 100 trees with the certain spec­ Definition of mangrove forest fragmentation classes.
ifications, i.e. 1 variable/split, 1 min leaf population, 0.5 bag fraction, Category Definition
false out of bag mode, and 0 seed parameters, were chosen to set up and
Patch Patch pixels are small fragments of forest that are surrounded by non-
run the RF classifier. The RF classifier uses ‘bootstrap aggregating’ or forest areas. It should be within 300 feet from the non-forest feature.
‘bagging’ method to generate random vectors with replacement of the The areas are represented by using yellow color.
size of the original input training data to select training data for each Edge Edge pixel makes the exterior boundary of core traces where they meet
class (Gislason et al., 2006; Mondal et al., 2019). In a RF classifier, the with non-forest areas. It determines the boundary between the non-
forested and core forest areas. These pixels are within 100 m and are
number of trees and variables per split are user defined and classifier along the outside edge of forest tract.
performance is not sensitive to the number of variables per split with Perforated Perforated pixel is the edge of small forest areas within the core forests
reducing the computational complexity and decreasing correlation like building a house within the forest. These are also within 100 m and
among the trees to the square root of the input variables (Gislason et al., are found along the edge of small forest gaps.
Core Core pixel is the last fragmented category which is very far from the
2006; Mondal et al., 2019). In this study, we utilized simple yet RF
non-forest boundary. Core pixel is divided into three subcategories
classification robust models with the slightly customized parameters based on the size of the patches:
available on GEE. • Small Core (having area of <1 km2)
For mapping of the defined three land cover classes, well distributed • Medium Core (having area of 1–2 km2)
random samples were taken from the Landsat images. The visual image • Large Core (having area of >2 km2)

interpretation techniques (tone, texture, and pattern), spectral signa­


tures/responses and mosaicked Google’s high-resolution background In the cross-tabulated or change matrix, diagonal values show the
satellite image in GEE helped to identify and differentiate land cover stability of land cover class, while omission and commission values
classes. In the study, from the collected random samples, ~70% were indicate a shift in area or percentage between the classes (Gilani et al.,
used to train the RF classifier while ~30% to assess the accuracy of 2017). In this study, using the change matrix method over each site,
developed land cover maps (Table S1). mangrove ‘loss’ (omission) and ‘gain’ (commission) statistically and
spatially were reported at five-year intervals from 1990 to 2020.
2.3.2. Accuracy assessment
In this study, using the error matrix method (Foody, 2002), overall, 2.3.4. Mangrove fragmentation analysis
producer, and consumer/user accuracies along with Cohen’s Kappa Forest fragmentation is the splitting up of large contiguous forest
coefficient value were generated and reported for each produced land fields into smaller or less contiguous areas. Several events or activities
cover map (Gilani et al., 2015) (Table S1). Through cross-tabular can lead to forest fragmentation, including road formations, woodcut­
method, statistics of overall accuracy express the probability of ting, forest conversion to agriculture, forest fires, human conflict over
correctly classified pixels with reference pixels. Overall accuracy and forest patches, etc. (Hauser et al., 2017; Li et al., 2013). At the site level,
Cohen’s Kappa value exists between 0 and 1; a value above 0.8 (80%) is the land cover maps were used to assess mangrove fragmentation. The
considered good accuracy, while a value equal to or below 0.4 (40%) is outcome of mangrove fragmentation analysis was represented into six
considered poor (Foody, 2010). In this study once the land cover maps categories: Patch, Edge, Perforated, Small Core (<1 km2), Medium Core
were developed at five-year intervals, for each land cover map an overall (1–2 km2), and Large Core (>2 km2) (Table 2). These categories are
accuracy reported in percentage from 0 to 100% while Cohen’s Kappa signs of mangrove ecosystem quality, which can be used to estimate the
value from 0 to 1. For additional review and checks, apart from statically amount of fragmentation and the potential habitat impacts (Li et al.,
generated accuracy assessment reports, the developed land cover maps 2013).
were overlaid on from derived respective Landsat scenes to visually
assess the synchronization among them. This visual interpretation was
3. Results
considered to be good enough and increased confidence in derived land
cover products (Congalton, 1991).
The results of this study consist of accuracy assessment, mangrove
cover change assessment, and mangrove fragmentation.
2.3.3. Mangrove cover change assessment
The final land cover maps were exported from GEE to desktops for
further statistical assessments, spatial analysis and cartography. To 3.1. Accuracy assessment
remove the “salt-and-pepper” effect in the land cover maps, the classi­
fied objects with an area smaller than Minimum Mapping Unit (MMU) (i. For each study site, using the confusion error matrix analysis, the
e. 1 ha ~ 3 × 3 pixels) were fused with the neighboring land cover accuracy assessment of developed individual land cover map of 1990,
classes (Saura, 2002). 1995, 2000, 2005, 2010, 2015 and 2020 was carried out (Table S2)
In this study, for mangrove cover change assessment, land cover using ~30% of total randomly collected samples. Table 3 shows a
maps for 1990, 1995, 2000, 2005, 2010, 2015, and 2020, were evalu­ summary report of the accuracy assessment. In this study, ≥90% overall
ated and compared in terms of area coverage. The annual rate of change accuracy and ≥0.8 Cohen’s Kappa coefficient value attained across all
and cross-tabulated methods were used to identify mangrove changes study sites for the generated individual land cover maps.
over the last three decades at five-year intervals, i.e. 1990-1995,
1995–2000, 2000–2005, 2005–2010, 2010–2015 and 2015–2020. 3.2. Mangrove cover change assessment
According to Puyravaud (2003), the annual rate of change is based
on the compound interest law, considering a non-linear change across Based on Landsat (30 m), the spatiotemporal (1990–2020) land
the timeline to estimate the percentage change per year. From 1990 to cover analysis at five-year intervals indicates that the overall area of
2020 at five-year intervals, the annual rate of change was calculated mangroves has been increased across all five study sites in Pakistan. It is
using equation 1 proposed by Puyravaud (2003): evident that ~95% of the total mangrove cover in Pakistan lies in the
Indus Delta site, while the remaining ~5% is spread across the four other
r ​ = ((1 / t2 − t1) ​ x ​ ln(A2 / A1)) ​ x ​ 100
study sites, i.e. Sandspit, Sonmiani Khor, Kalmat Khor, and Jiwani. The
Where ‘r’ is the annual rate of change in percentage, and A1 and A2 the findings of this paper reveal, at the national scale, out of the total area of
estimated area at time t1 and time t2 respectively. 9538 km2 at all five sites, an estimated 477.22 km2 mangrove area in
1990 increased to 1463.59 km2 in 2020 a 3.74% annual rate of change

5
H. Gilani et al. Estuarine, Coastal and Shelf Science 249 (2021) 107128

Table 3
Summary of accuracy assessment of three land cover classes over the five mangrove study sites in Pakistan (detailed accuracy assessment reports given in Table S2).
Study site Accuracy and Kappa 1990 1995 2000 2005 2010 2015 2020

Indus Delta Overall accuracy (%) 90 97 95 98 93 96 97


Kappa coefficient value 0.92 0.97 0.92 0.97 0.92 0.93 0.95
Sandspit Overall accuracy (%) 92 90 92 90 93 94 96
Kappa coefficient value 0.88 0.82 0.88 0.84 0.9 0.92 0.93
Sonmiani Khor Overall accuracy (%) 90 91 93 95 92 91 95
Kappa coefficient value 0.85 0.86 0.9 0.92 0.88 0.88 0.92
Kalmat Khor Overall accuracy (%) 93 97 94 93 94 92 93
Kappa coefficient value 0.9 0.95 0.91 0.89 0.9 0.87 0.89
Jiwani Overall accuracy (%) 94 95 91 95 90 95 96
Kappa coefficient value 0.91 0.93 0.85 0.92 0.86 0.92 0.94

Table 4
Mangrove cover assessment based on Landsat 30 m spatial resolution at five-year intervals (1990–2020) over the five mangrove sites in Pakistan (detailed land cover
assessment given in Table S3).
Study site km2

1990 1995 2000 2005 2010 2015 2020

Indus Delta 445.75 540.17 776.76 804.65 1045.31 1198.97 1395.66


Sandspit 12.09 9.79 10.74 12.41 13.05 13.33 13.56
Sonmiani Khor 14.80 27.66 32.41 35.56 36.15 40.30 42.80
Kalmat Khor 3.78 2.77 3.61 5.75 6.62 6.68 9.33
Jiwani 0.81 0.83 0.96 1.94 1.93 1.99 2.23
National scale 477.22 581.22 824.47 860.32 1103.05 1261.28 1463.59

Table 5
Mangrove loss and gain assessment based on Landsat 30 m spatial resolution at five-year intervals (1990–2020) over the five mangrove sites in Pakistan.
Class km2

1990–1995 1995–2000 2000–2005 2005–2010 2010–2015 2015–2020

Indus Delta
Mangroves loss 41.85 65.64 110.99 63.11 123.03 132.80
Mangroves gain 125.27 293.25 138.85 299.28 274.59 322.46
Sandspit
Mangroves loss 2.90 0.71 0.74 1.15 0.70 1.43
Mangroves gain 0.65 1.65 2.35 1.73 0.99 1.62
Sonmiani Khor
Mangroves loss 0.19 1.17 1.82 2.42 2.06 4.55
Mangroves gain 12.77 5.62 4.83 3.10 6.09 6.72
Kalmat Khor
Mangroves loss 2.70 0.46 1.74 1.95 1.96 1.75
Mangroves gain 0.20 2.80 3.84 2.74 2.08 4.37
Jiwani
Mangroves loss 0.38 0.35 0.04 0.44 0.44 0.43
Mangroves gain 0.39 0.48 1.01 0.42 0.49 0.72
National scale
Mangroves loss 48.02 68.33 115.33 69.08 128.19 140.96
Mangroves gain 139.28 303.80 150.88 307.26 284.24 335.89

Table 6
Mangrove cover annual rate of change over the five mangrove sites in Pakistan, calculated based on equation 1 proposed by Puyravaud (2003).
Study site %

1990–1995 1995–2000 2000–2005 2005–2010 2010–2015 2015–2020 1990–2020

Indus Delta 3.84 7.26 0.71 5.23 2.74 3.04 3.80


Sandspit − 4.22 1.86 2.90 1.01 0.42 0.34 0.38
Sonmiani Khor 12.51 3.17 1.86 0.33 2.18 1.20 3.54
Kalmat Khor − 6.22 5.26 9.34 2.80 0.21 6.67 3.01
Jiwani 0.66 2.91 14.03 − 0.18 0.66 2.29 3.39
National scale 3.94 6.99 0.85 4.97 2.68 2.98 3.74

(Tables 4–6, Table S3). (17.45%) is covered in 2020 by mangroves, the rest of the areas were
The findings on mangrove cover change at the five sites are: under ‘water’ and ‘other’ land cover classes (Table S3). In the last thirty
years (1990–2020), a total of 950 km2 mangrove area increased in the
3.2.1. Indus Delta Indus Delta. In 1990, the mangroves covered 445.75 km2, which
In the Indus Delta, out of a total of 8000 km2 study area, 1395.66 km2 increased to 540.17 km2 in 1995, with an annual rate of change 3.84%

6
H. Gilani et al. Estuarine, Coastal and Shelf Science 249 (2021) 107128

Fig. 3. Indus Delta - Mangrove cover and change assessment based on Landsat 30 m spatial resolution at five-year intervals (1990–2020).

and 41.85 km2 and 125.27 km2 mangrove loss and gain from 1990 to increased by only 1.48 km2 in the Sandspit study site. In 1990, a total of
1995 respectively. Between 1995 and 2000, mangrove loss of 65.64 km2 12.09 km2 area was under mangroves, which reduced to 9.79 km2 in
and gain of 293.25 km2 was detected, with 7.26% annual rate of change; 1995 with − 4.22% annual rate of change and 2.90 km2 loss and 0.65
mangrove cover further improved from 540.17 km2 in 1995 to 776.76 km2 gain in mangrove cover between 1990 and 1995. Only 0.95 km2
km2 in 2000. In the next five years (2000–2005), only 28 km2 of was added to mangrove cover between 1995 and 2000. In 2000 man­
mangrove area was added, with a slight annual rate of change of 0.71%, groves covered 10.74 km2, which further improved to 12.41 km2 in
with an overall loss of 110.99 km2 and gain of 138.85 km2. In the last 2005, then 13.05 km2 in 2010, to 13.33 km2 in 2015, and finally
fifteen years (2005–2020) of the study period, thus, there has been a increased to 13.56 km2 in 2020. Based on the mangrove cover assess­
substantial improvement in mangrove cover, i.e.804.65 km2 in 2005, ment, in Sandspit the annual rate of change was 1.86% between 1995
1045.31 km2 in 2010, 1198.97 km2 in 2015 and 1395.66 km2 in 2020, and 2000, 2.90% between 2000 and 2005, 1.01% between 2005 and
with an annual rate of change 5.23% between 2005 and 2010, 2.74% 2010, 0.42% between 2010 and 2015, and 0.34% between 2015 and
between 2010 and 2015 and 3.04% between 2015 and 2020. Although 2020. In the Sandspit site, mangrove loss findings showed 0.71 km2 lost
the overall mangroves extent increased in the last fifteen years, between 1995 and 2000, 0.74 km2 between 2000 and 2005, 1.15 km2
mangrove loss has also increased, i.e. 63.11 km2 were lost between 2005 from 2005 to 2010, 0.7 km2 between 2010 and 2015, and 1.43 km2 from
and 2010, 123.03 km2 between 2010 and 2015, and 132.80 km2 be­ 2015 to 2020. On the spatial map the mangrove losses are observed in
tween 2015 and 2020. While mangrove gain was 299.28 km2 between the northeast and southwest areas (Tables 4–6, Fig. 4).
2005 and 2010, 274.59 km2 between 2010 and 2015, and 322.46 km2
between 2015 and 2020. The spatial distribution of mangrove cover and 3.2.3. Sonmiani Khor
change was mostly on the southeastern site of Indus Delta (Tables 4–6 According to the findings of this study, in Sonmiani Khor site, out of a
Fig. 3). total area of 830 km2, 42.80 km2 (5.16%) is covered by mangroves in
2020. In thirty years (1990–2020), an impressive increment has been
3.2.2. Sandspit observed in mangrove cover, i.e. from 14.80 km2 in 1990 to 42.80 km2
In the Sandspit out of a total study area of 68 km2, 13.56 km2 (20%) in 2020. From 1990 to 1995, the mangrove spread increased by 12.86
consists of mangroves in 2020. From 1990 to 2020, the mangrove area km2 area with 12.51% annual rate of change, while for the rest of the

7
H. Gilani et al. Estuarine, Coastal and Shelf Science 249 (2021) 107128

Fig. 4. Sandspit - Mangrove cover and change assessment based on Landsat 30 m spatial resolution at five-year intervals (1990–2020).

study period, at five-year intervals from 1995 to 2020, <5 km2 area and respectively, with 14.03% annual rate of change. In Jiwani, mangroves
<5% annual rate of change was detected (Tables 4–6, Fig. 5). are concentrated in the middle of the study area; loss and gain in
mangrove cover was observed due to fluctuation in water extent
3.2.4. Kalmat Khor (Tables 4–6, Fig. 7).
In the Kalmat Khor site, out of a total area of 560 km2 only 9.33 km2
(1.67%) is covered by mangroves in 2020. In 1990, the mangrove area 3.3. Mangrove fragmentation
was 3.78 km2 that reduced to 2.77 km2 in 1995, and then increased to
3.61 km2 in 2000. From 2005 to 2020, without any decline, an increase Mangrove forest fragmentation analysis results are presented in
in mangrove cover was observed with an annual rate of change of 2.8% Table 7 and Fig. 8. Based on thirty years of mangrove cover assessments
from 2005 to 2010, 0.21% from 2010 to 2015, and 6.67% from 2015 to (1990–2020), the fragmentation analysis revealed improvements in
2020. On this site, mangrove loss findings showed 2.70 km2 from 1990 mangrove tree canopy coverage.
to 1995, 0.46 km2 from 1995 to 2000, 1.74 km2 from 2000 to 20005, In the Indus Delta site, the ‘Large Core (>2 km2)’ fragmentation class
1.95 km2 from 2005 to 2010, 1.96 km2 from 2010 to 2015, and 1.75 km2 showed a continuous increasing trend from 86.4 km2 in 1990 to 734.1
from 2015 to 2020. Maximum mangrove gain, i.e. 4.37 km2 was km2 in 2020, while in ‘Small Core (<1 km2)’ and ‘Medium Core (1–2
observed between 2015 and 2020 and minimum gain of 0.2 km2 be­ km2)’ fragmentation classes increase was detected with a slight decline
tween 1990 and 1995 (Tables 4–6, Fig. 6). in 2005. In Indus Delta, the ‘Edge’ fragmentation class intensified from
144.21 km2 in 1990 to 187.45 km2 in 2000, with a slight drop to 148.11
3.2.5. Jiwani km2 in 2005 and then again rise to 300.44 km2 in 2020. Overall, in the
At the Jiwani site, out of a total area of 80 km2 only 2.23 km2 northwest side of the Indus Delta, spatially, improvement can be seen in
(2.79%) is covered by the mangroves in 2020. Until 2000 the mangrove the mangrove canopy cover.
cover was <1 km2 and <2 km2 until 2015. In thirty years (1990–2020), In the Sandspit site, ‘Large Core (>2 km2)’ fragmentation class was
there has been an overall 1.43 km2 increase in mangrove cover with <3 km2 till 2000, which increased to 8.11 km2 in 2005, then fell to 5.77
3.39% annual rate of change. Maximum mangrove gain and least loss km2 in 2010 and then again rose to 7.76 km2 in 2020. In thirty years
was observed from 2000 to 2005, i.e. 1.01 km2 and 0.04 km2 (1990–2020), the ‘Patch’ class overall dropped from 0.35 km2 in 1990 to

8
H. Gilani et al. Estuarine, Coastal and Shelf Science 249 (2021) 107128

Fig. 5. Sonmiani Khor - Mangrove cover and change assessment based on Landsat 30 m spatial resolution at five-year intervals (1990–2020).

0.20 km2 in 2020, with slight oscillation between the five-year periods. 4. Discussion
In the Sonmiani Khor site, the ‘Medium Core (1–2 km2)’ and ‘Large
Core (>2 km2)’ fragmentation classes were completely absent in 1990. This study used multi-date Landsat images and standard methodol­
The ‘Edge’ fragmentation class remained above 6 km2 with slight fluc­ ogy to quantify mangrove cover changes from 1990 to 2020 at five-year
tuations from 1990 to 2020. The ‘Patch’ fragmentation class was 1.24 intervals. High temporal and spectral resolutions of Landsat images with
km2 in 1990, which reduced to 0.57 km2 in 1995, remained <0.5 km2 a low saturation level of spectral bands and integration of indices are the
until 2010, then climbed to 1.64 km2 in 2015, and fell to 0.47 km2 in major factors that ensure the accuracy of land cover maps for temporal
2020. mangrove cover change estimation over the diverse study sites.
In the Kalmat Khor site, the ‘Large Core (>2 km2)’ mangrove frag­ Spatially explicit and periodic mangrove cover information provided
mentation class was nonexistent in thirty years while ‘Medium Core in this study are essential for various purposes, including the formation
(1–2 km2)’ was also not present till 2010. In 2015, ‘Medium Core (1–2 and implementation of mangrove ecosystem management plans for local
km2)’ observed 0.79 km2 and 1.59 km2 in 2020. The ‘Edge’ fragmen­ communities, environmental impact assessment studies, to quantify
tation class continuously increased from 0.78 km2 in 1990 to 3.89 km2 in fuelwood demand and supply, for carbon budgeting, and performance-
2020. based REDD + payment distribution, among others.
In the Jiwani site, ‘Medium Core (1–2 km2)’ and ‘Large Core (>2
km2)’ mangrove fragmentation classes did not occur in thirty years. In
‘Small Core (<1 km2)’ class an upturn was observed, from 0.15 km2 in 4.1. Major findings and ground realities
1990 to 1.09 km2 in 2020, which slightly declined between 2010 and
2015. The ‘Edge’ fragmentation class increased throughout the thirty The findings of this research reveal that in Pakistan, over the thirty
years of study. years spanning from 1990 to 2020, mangrove cover has increased by a
total of 986.36 km2 with an annual rate of change of 3.74% (Tables 3
and 4). Throughout all five studied sites, the spatiotemporal mangrove
cover analysis discovered, from the year 2005 onward, the mangrove
cover losses substantially reduced with growing and extending trends. In

9
H. Gilani et al. Estuarine, Coastal and Shelf Science 249 (2021) 107128

Fig. 6. Kalmat Khor - Mangrove cover and change assessment based on Landsat 30 m spatial resolution at five-year intervals (1990–2020).

terms of mangrove cover increase, the current study outcomes are mangrove cover of three sites (Sonmiani Khor, Kalmat Khor, and Jiwani)
endorsed by several previously published investigations (Amjad and of Balochistan province is also attributable to the fact that these areas
Jusoff, 2007; Giri et al., 2015; Saeed et al., 2019). are geographically remote coastal lagoons with limited population.
On an annual basis from 2006, in the coastal areas of Pakistan, Indeed, growth of mangroves will have positive benefits (e.g. protection
mangrove plantation campaigns are being conducted by national and from tsunami, carbon sequestration etc.) but could reduce the area of
international Non-Governmental Organizations (NGOs), Pakistan Navy, habitat available for migratory wading birds.
Government of Pakistan, and local Community-Based Organizations In the Indus Delta and Sandspit sites, the erosion of mud flats in the
(CBOs) (Mukhtar and Hannan, 2012). In 2009, the Guinness Book of creeks (sea intrusion) is a continuous phenomenon (Ijaz et al., 2018).
World Records certified Pakistan on planting 541,176 mangrove sap­ Waqas et al. (2019) reported, in the Indus Delta, 3750 km2 area was
lings in 24 h near Keti Bandar within the Indus Delta study site (Dawn, encroached from 1976 to 2017. Additionally, they also found that along
2009). Apart from these efforts for plantation drives, the conservation of the deltaic region 38% of the barrier islands are vulnerable to oceanic
existing mangrove cover is also remarkable which can be seen through factors, while 7% were found to be partially vulnerable, 17% remained
the mangrove fragmentation analysis. In most of sites, improvements in partially sustainable, and only 8% of these barrier islands sustained
three core (small, medium and large) classes of fragmentation have been against the ocean controlling factors. Kanwal et al. (2020) assessed the
observed, which indicate dense canopy cover of existing mangrove. rate of land erosion increasing from west to east in the Indus Delta, with
Fig. 9 shows the temporal terrestrial photographs, a vast area of mud­ the highest average rate of 27.46 m/yr from 1989 to 2018. Zia et al.
flats converted to dense mangrove. By taking a small portion of Indus (2017) conducted an assessment of seawater inundation along Daboo
Delta, Saeed et al. (2019) reported extraordinary growth in mangrove creek, Indus Delta, through spatiotemporal analysis of Landsat-5 TM
plantation areas, through repeat photographs from 2010 to 2015 and data for the years 1987 and 2010. According to their findings, in
openly accessible Google Earth high-resolution and Landsat medium twenty-one years, the Daboo creek area increased ~9.93%, i.e. 260.86
spatial resolution satellite images. Rasool et al. (2002), described m2/year. Nazeer et al. (2020) observed that the shoreline of Sandspit is
various conservation techniques and practices, which were developed experiencing accretion and erosion. They investigated and reported that
and implemented in Sonmiani Khor to restore the mangrove ecosystem over a period of 76-year (1942–2018) the length of shorelines in the
through community participation. The slight increase observed in eastern and western zones of Sandspit study site have increased by 2.6%

10
H. Gilani et al. Estuarine, Coastal and Shelf Science 249 (2021) 107128

Fig. 7. Jiwani - Mangrove cover and change assessment based on Landsat 30 m spatial resolution at five-year intervals (1990–2020).

and 16.6% respectively. Indeed, in Pakistan, mangrove cover and den­ (2012) and Khan et al. (2010) studied the level of pollution along the
sities are showing increasing trends but at the same time, sea-land Karachi coastline. According to findings, the domestic and shipping in­
intrusion or soil erosion is an increasing threat for mangroves. Aziz dustry discharges, water pollution, and possible leakages and spills,
and Khan (2001) and Giri et al. (2015) declared, soil salinity, transport create negative impacts over this area. This pollution directly affects the
and uniform dispersal of suspended sediments are primary and proxy shoreline, including the mangroves, invertebrates, birds and can have
drivers and causes of mangrove detrimental and degradation in the serious implications on human health.
Indus delta.
Due to climate change and anthropogenic activities, Pakistan’s 4.2. Comparison with previous studies
mangroves have become monoculture with the most dominant species
being Avicennia marina. Once there were eight mangrove species (Bru­ Among all five studied sites, the Indus Delta is the most investigated
guiera conjugata, Ceriops tagal, Ceriops roxburghiana, Rhizophora apicu­ site by a number of researchers. According to the study conducted by
lata, Rhizophora mucronata, Aegiceras corniculatum, Avicennia marina, Masood et al. (2015), based on the supervised classification technique,
and Sonneratia caseolaris) present along the coast of Pakistan (Saifullah, the mangrove cover of Indus Delta was 946.52 km2 in 2009 which has
1982), out of which four species (Bruguiera conjugata, Ceriops rox­ increased to 960.83 km2 in 2014; their onscreen digitization results
burghiana, Rhizophora apiculata and Sonneratia caseolaris) have showed that in five years (2009–2014) mangroves extended up to 72.6
completely disappeared. By neglecting diversity, overall mangrove km2 (1010.11 km2 in 2009 and 1082.71 km2 in 2014). Based on a
plantation campaigns are being carried out of Avicennia marina to only one-time assessment, Abbas et al. (2013) reported that in 2009 man­
focus on increasing forest cover of Pakistan which is currently <3%. groves covered 981.29 km2 area over all five mangrove areas of
Camel grazing, fuelwood extraction, and over-fishing are the most Pakistan. Giri et al. (2015) used two different spatial resolution Landsat
common anthropogenic threats to mangrove conservation practices satellite images (60 m and 30 m) and classification techniques (super­
(Ahmed and Shaukat, 2015; Amjad et al., 2007). vised and object-based), reported 442.3 km2 increase in mangrove cover
In thirty years (1990–2010), Sandspit mangroves show an extension with of a total 177 km2 loss and 766.16 km2 no-change in the Indus
(1.48 km2) in mangrove cover with few losses, Fig. 10 demonstrates the Delta’s mangroves between 1973 and 2010 (Giri et al., 2015). According
conversion of land from one land type to another type. Hameed et al. to the area estimations of WWF-Pakistan (2005), mangroves decreased

11
H. Gilani et al. Estuarine, Coastal and Shelf Science 249 (2021) 107128

Table 7
Mangrove fragmentation results 1990, 1995, 2000, 2005, 2010, 2015 and 2020 based on mangrove cover maps developed using based on Landsat 30 m spatial
resolution at five-year intervals (1990–2020).
Study site Fragmentation class km2

1990 1995 2000 2005 2010 2015 2020

Indus Delta Patch 32.68 89.42 34.07 22.03 62.87 71.96 60.15
Edge 144.21 161.24 187.45 148.11 250.01 288.93 300.44
Perforated 22.78 32.43 37.23 37.88 42.16 46.49 32.63
Core (<1 km2) 115.20 118.51 121.60 87.58 125.57 159.90 155.23
Core (1–2 km2) 44.48 54.11 78.02 42.29 96.74 116.89 113.12
Core (>2 km2) 86.40 84.46 318.38 466.77 467.95 514.81 734.10
Total 445.75 540.17 776.76 804.65 1045.31 1198.97 1395.66
Sandspit Patch 0.35 0.48 0.38 0.15 0.41 0.33 0.20
Edge 3.42 2.88 2.75 2.33 3.08 3.04 2.73
Perforated 0.34 0.53 0.72 0.32 0.50 0.32 0.22
Core (<1 km2) 4.17 2.59 2.13 1.50 2.07 2.25 2.65
Core (1–2 km2) 1.26 1.09 2.54 0.00 1.22 0.00 0.00
Core (>2 km2) 2.55 2.22 2.22 8.11 5.77 7.40 7.76
Total 12.09 9.79 10.74 12.41 13.05 13.33 13.56
Sonmiani Khor Patch 1.24 0.57 0.46 0.36 0.50 1.64 0.47
Edge 6.15 8.44 8.64 7.56 10.19 10.21 8.93
Perforated 0.68 1.11 1.94 1.45 1.49 1.04 1.16
Core (<1 km2) 6.74 8.30 5.95 2.90 7.19 6.11 7.44
Core (1–2 km2) 0.00 6.87 3.51 2.98 2.10 6.05 2.52
Core (>2 km2) 0.00 2.37 11.92 20.30 14.69 15.26 22.27
Total 14.80 27.66 32.41 35.56 36.15 40.30 42.80
Kalmat Khor Patch 0.97 0.18 0.92 0.55 1.30 1.14 1.05
Edge 0.78 0.79 1.03 1.66 2.56 2.90 3.89
Perforated 0.66 0.03 0.65 0.77 0.12 0.08 0.08
Core (<1 km2) 1.37 1.77 1.01 2.77 2.64 1.76 2.72
Core (1–2 km2) 0.00 0.00 0.00 0.00 0.00 0.79 1.59
Core (>2 km2) 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Total 3.78 2.77 3.61 5.75 6.62 6.68 9.33
Jiwani Patch 0.35 0.38 0.21 0.25 0.55 0.54 0.14
Edge 0.31 0.28 0.48 0.67 0.66 0.66 0.91
Perforated 0.00 0.05 0.01 0.08 0.14 0.15 0.09
Core (<1 km2) 0.15 0.12 0.27 0.94 0.58 0.65 1.09
Core (1–2 km2) 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Core (>2 km2) 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Total 0.81 0.83 0.96 1.94 1.93 1.99 2.23

in Keti Bandar from 95 km2 in 1992 (Landsat, 30 m data) to 75 km2 in data produced by Giri et al. (2011), Abbas et al. (2013) and Bunting et al.
2001 (ASTER, 15 m data), while in the Sonmiani Khor site the man­ (2018) (Table 8). At the national as well as site scales, differences in
groves increased from 25 km2 in 1989 (Landsat) to 37 km2 in 2001 mangrove coverage area were observed between current and previous
(ASTER), in Kalmat Khor mangroves doubled from 2.5 km2 in 1990 studies, most particularly in the Indus Delta region. Over the Indus Delta
(Landsat) to 5.2 km2 in 2004 (ASTER), while a one-time assessment a comparison done between 2000 mangrove cover reported in Giri et al.
based on 2003 QuickBird high-resolution satellite imagery showed ~4 (2011) and 2010 reported in Giri et al. (2015) with the results of current
km2 mangrove cover in Sandspit. In this study, mangrove cover was study (Fig. 11). Based on 2000’s mangrove cover between Giri et al.
extracted from Landsat imagery, which is spatially coarser than ASTER (2011) and current study, ~294 km2 difference was observed while in
and QuickBird, and compared with different classification techniques 2010 only 65 km2 between Giri et al. (2015) and current study results.
and mangrove classification legends (WWF-Pakistan, 2005). In 1983 Differences in area coverage and estimations are might be because of
SUPARCO reported ~2600 km2 mangrove vegetation over Indus Delta, input data sources, spatial resolutions, classification methods, and study
and in 2003 a total 867 km2 area was under mangroves in Pakistan scales. The present study used a robust machine learning classifier,
(IUCN, 2005; WWF-Pakistan, 2005). Over 1553 km2 mangrove area was well-distributed random samples, and field experienced analysts and
reported in the FSMP land cover map, and NFRRAS reported a total of experts that could ensure higher accuracy of the study as compared to
1580 km2 the mangrove areas (Abbas et al., 2013). According to a study other published studies.
conducted by the Sindh Forest Department, Memon (2012) observed
that in 2010 Pakistan’s mangrove cover was ~1080 km2. 4.3. Limitations of this study
At the global scale, Giri et al. (2011) mapped mangrove extent using
Landsat imagery of 2000, according to this study Pakistan had a total of This study was carried out in quite detail and included multiple steps
535.22 km2 mangrove area. Under the Global Mangrove Watch (GMW) from start to end, including: literature review, satellite data collection,
initiative, a 2010 global baseline of mangrove extent was produced by collection of training samples, classification methodology development,
Bunting et al. (2018) by applying extremely randomized trees classifier validation of land cover maps, and spatial analysis conducted to the best
on ALOS PALSAR and Landsat sensor datasets with the integration of of the authors’ knowledge and experience. There may be some flaws,
field observations and previously conducted studies on globally devel­ errors or uncertainty in data that may affect results and methodology, so
oped datasets (mangrove cover, DEM, surface water, shorelines, ba­ there are some limitations of this study.
thymetry) (Giri et al., 2011; Pekel et al., 2016; Soluri and Woodson, In this study, medium spatial resolution Landsat images were taken
1990; Spalding, 2010; Weatherall et al., 2015; Wessel and Smith, 1996). for the month of January to April (Table 1) but in 2010, over the Kalmat
As per this study’s findings, in 2010, Pakistan’s mangrove extent was Khor site, due to unavailability of January to April images, we selected
687.60 km2 area (Bunting et al., 2018). At the site level for the statistical November 2009 to produce the land cover map. Landsat-7 ETM + SLC
and spatial comparison of mangrove covers, we could only get access to off images were corrected and utilized, which could effect spectral

12
H. Gilani et al. Estuarine, Coastal and Shelf Science 249 (2021) 107128

Fig. 8. A, B & C: Mangrove fragmentation based on mangrove cover maps over the five sites at five years intervals (1990–2020).

Fig. 9. Extensive and impressive improvement in mangrove plantation in southern site of Indus Delta. Photograph credits: WWF-Pakistan.

13
H. Gilani et al. Estuarine, Coastal and Shelf Science 249 (2021) 107128

Fig. 10. Land transformation in Sandspit; (a) Water to Soil, (b) Wet-soil to Mangroves and (c) Mangroves to Settlement.

values and ultimately land cover maps. To be cautious, the final land
Table 8 cover maps were visually assessed to ensure synchronization among
A comparison of mangrove coverage area between current and global scale original satellite data and produced land covers. At times of spring tides,
conducted studies. mangroves roots and lower stems might be inundated, while at neap
Sites km2 tides the mangroves may be exposed for several hours to days. The tidal
Giri et al. Current Abbas Bunting Current cycle along the Pakistan coast is mainly a semi-diurnal mixed tidal
(2011) study Year et al. et al. study Year regime with daily inequality (Syed and Siddiqa, 2019). Although, in this
Year 2000 2000 (2013) (2018) 2010 study, the selected satellite images were at the neap tide height time,
Year 2009 Year 2010
however, the mangrove extent and changes may vary according to the
Indus 482.85 776.76 924.11 638.09 1045.31 tidal cycle and inundation. For the mangrove classification and change
Delta assessment of the thirty years, we were lacking ground samples. This
Sandspit 11.72 10.74 10.56 12.03 13.05
Sonmiani 37.69 32.41 40.20 33.72 36.15
might introduce uncertainties in the mangroves extent and estimated
Khor losses and gains for different years. Therefore, we utilized satellite data
Kalmat 1.64 3.61 4.07 1.84 6.62 derived vegetation indices and ground experts’ knowledge, to try and
Khor reduce the uncertainties of the generated mangrove cover and change
Jiwani 1.32 0.96 2.35 1.93 1.93
maps. The current study results are from only optical satellite images,
National 535.22 824.48 981.29 687.60 1103.06
scale which may be improved upon in future studies by employing microwave
and hyperspectral imagery as well as the integration of active and pas­
sive remotely sensed datasets.

Fig. 11. A comparison between Giri et al. (2011) and Giri et al. (2015) with current study over the Indus Delta.

14
H. Gilani et al. Estuarine, Coastal and Shelf Science 249 (2021) 107128

5. Conclusions Durranee, A., Hasnain, S.A., Ahmad, E., 2008. Observations on the birds of Sandspit/
Hawkesbay coastal wetland complex, Karachi coast. Pakistan J. Zool. 40, 229–237.
Elmahdy, S.I., Ali, T.A., Mohamed, M.M., Howari, F.M., Abouleish, M., Simonet, D.,
The study, being the first, provides mangrove cover and change 2020. Spatiotemporal mapping and monitoring of mangrove forests changes from
assessment at five-year intervals over the five mangrove areas in 1990 to 2019 in the northern emirates, UAE using random forest, Kernel logistic
Pakistan, for a period of thirty years (1990–2020), using consistent set of regression and naive Bayes tree models. Front. Environ. Sci. https://doi.org/
10.3389/fenvs.2020.00102.
RF classification method and Landsat 30 m spatial resolution satellite FAO, 2010. Asia-pacific forests and forestry to 2020 report of the second Asia-pacific
images from open access GEE cloud computing platform. The current forestry sector outlook study, Second Asia-Pacific Forestry sector Outlook Study.
methodology was cost-effective and can be easily replicable over five- Food and Agriculture Organization (FAO), Rome, Italy.
FAO, 2007. Mangroves of Asia 1980-2005: Country Reports. Italy, Rome.
year intervals using the operation Landsat-8 or planned Landsat-9 sat­ Foody, G.M., 2010. Assessing the accuracy of land cover change with imperfect ground
ellite datasets. The results of this study showed a significant increase in reference data. Remote Sens. Environ. 114, 2271–2285. https://doi.org/10.1016/j.
mangrove cover of Pakistan. The data and findings of this study could rse.2010.05.003.
Foody, M.G., 2002. Status of land cover classification accuracy assessment. Remote Sens.
help forest department and relevant agencies to make strategic decisions Environ. 80, 185–201.
to prevent mangrove loss and may be suitable for sustainable planning Gao, B., 1996. NDWI—a normalized difference water index for remote sensing of
for management and conservation of mangroves in Pakistan. vegetation liquid water from space. Remote Sens. Environ. 58, 257–266. https://doi.
org/10.1016/S0034-4257(96)00067-3.
Gilani, H., Shrestha, H.L., Murthy, M.S.R., Phuntso, P., Pradhan, S., Bajracharya, B.,
CRediT authorship contribution statement Shrestha, B., 2015. Decadal land cover change dynamics in Bhutan. J. Environ.
Manag. 148, 91–100. https://doi.org/10.1016/j.jenvman.2014.02.014.
Gilani, H., Sohail, M., Koju, U.A., 2017. From REDD+ MRV perspective: comparison of
Hammad Gilani: Conceptualization, Methodology, Supervision. two different forest management regimes using geospatial techniques in Ludi Khola
Hafiza Iqra Naz: Software, Formal analysis, Validation. Masood watershed, Gorkha district, Nepal. PFG – J. Photogramm. Remote Sens. Geoinf. Sci.
Arshad: Funding acquisition, Project administration. Kanwal Nazim: 85, 265–278. https://doi.org/10.1007/s41064-017-0028-x.
Giri, C., Long, J., Abbas, S., Murali, R.M., Qamer, F.M., Pengra, B., Thau, D., 2015.
Writing - review & editing. Usman Akram: Validation. Aneeqa Abrar:
Distribution and dynamics of mangrove forests of South Asia. J. Environ. Manag.
Writing - review & editing. Muhammad Asif: Validation. 148, 101–111. https://doi.org/10.1016/j.jenvman.2014.01.020.
Giri, C., Ochieng, E., Tieszen, L.L., Zhu, Z., Singh, A., Loveland, T., Masek, J., Duke, N.,
2011. Status and distribution of mangrove forests of the world using earth
Declaration of competing interest observation satellite data. Global Ecol. Biogeogr. 20, 154–159. https://doi.org/
10.1111/j.1466-8238.2010.00584.x.
The authors declare that they have no known competing financial Gislason, P.O., Benediktsson, J.A., Sveinsson, J.R., 2006. Random forests for land cover
classification. Pattern Recognition Letters. https://doi.org/10.1016/j.
interests or personal relationships that could have appeared to influence patrec.2005.08.011.
the work reported in this paper. Gorelick, N., Hancher, M., Dixon, M., Ilyushchenko, S., Thau, D., Moore, R., 2017.
Google earth engine: planetary-scale geospatial analysis for everyone. Remote Sens.
Environ. 202, 18–27. https://doi.org/10.1016/j.rse.2017.06.031.
Appendix A. Supplementary data Hameed, I., Soomro, Y.A., Butt, A.S., Shakoor, R., 2012. Coastline pollution problems of
Karachi. Int. J. Emerg. Trends Eng. Dev. 2, 528–537.
Supplementary data to this article can be found online at https://doi. Hauser, L.T., Nguyen Vu, G., Nguyen, B.A., Dade, E., Nguyen, H.M., Nguyen, T.T.Q.,
Le, T.Q., Vu, L.H., Tong, A.T.H., Pham, H.V., 2017. Uncovering the spatio-temporal
org/10.1016/j.ecss.2020.107128. dynamics of land cover change and fragmentation of mangroves in the Ca Mau
peninsula, Vietnam using multi-temporal SPOT satellite imagery (2004–2013). Appl.
References Geogr. 86, 197–207. https://doi.org/10.1016/j.apgeog.2017.06.019.
Ijaz, M.W., Mahar, R.B., Siyal, A.A., Anjum, M.N., 2018. Geospatial analysis of creeks
evolution in the Indus Delta, Pakistan using multi sensor satellite data. Estuar. Coast
Abbas, S., Qamer, F.M., Ali, G., Tripathi, N.K., Shehzad, K., Saleem, R., Gilani, H., 2013.
Shelf Sci. 200, 324–334. https://doi.org/10.1016/j.ecss.2017.11.025.
An assessment of status and distribution of mangrove forest cover in Pakistan.
IUCN, 2005. Mangroves of Pakistan, Status and Management. Karachi, Pakistan.
J. Biodivers. Environ. Sci. (JBES) 3, 64–78.
Kamal, A., Yingjie, M., Ali, A., 2019. Significance of billion tree tsunami afforestation
Adhikari, B., Baig, S.P., Iftikhar, U.A., 2010. The use and management of mangrove
project and legal developments in forest sector of Pakistan. Int. J. Law Soc. 1,
ecosystems in Pakistan. J. Environ. Dev. 19, 446–467. https://doi.org/10.1177/
157–165. https://doi.org/10.11648/j.ijls.20180104.13.
1070496510384392.
Kanwal, S., Ding, X., Sajjad, M., Abbas, S., 2020. Three decades of coastal changes in
Ahmed, W., Shaukat, S.S., 2015. Status of mangroves of North-Western part of Indus
Sindh, Pakistan (1989–2018): a geospatial assessment. Rem. Sens. 12, 8. https://doi.
Delta : environmental characteristics and populated structure. Pakistan J. Mar. Sci.
org/10.3390/rs12010008.
24, 61–85.
Kathiresan, K., Anburaj, R., Gomathi, V., Saravanakumar, K., 2013. Carbon sequestration
Alongi, D.M., 2012. Carbon sequestration in mangrove forests. Carbon Manag. 3,
potential of Rhizophora mucronata and Avicennia marina as influenced by age,
313–322. https://doi.org/10.4155/cmt.12.20.
season, growth and sediment characteristics in southeast coast of India. J. Coast
Amjad, A.S., Jusoff, K., 2007. Mangrove conservation through community participation
Conserv. 17, 397–408. https://doi.org/10.1007/s11852-013-0236-5.
in Pakistan: the case of Sonmiani Bay. Environ. Sci. Ecosyst. Dev. 1, 438–444.
Kauffman, J.B., Bhomia, R.K., 2017. Ecosystem carbon stocks of mangroves across broad
Amjad, A.S., Kasawani, I., Kamaruzaman, J., 2007. Degradation of Indus delta
environmental gradients in West-Central Africa: global and regional comparisons.
mangroves in Pakistan. Int. J. Geol. 3, 27–34.
PLoS One 12, e0187749. https://doi.org/10.1371/journal.pone.0187749.
Aziz, I., Khan, M.A., 2001. Experimental assessment of salinity tolerance of Ceriops tagal
Khan, Z.M., Babar, H., Ghalib, S.A., Zehra, A., Nazia, M., 2010. Distribution, population
seedlings and saplings from the Indus delta, Pakistan. Aquat. Bot. https://doi.org/
status and environmental impacts on reptiles in Manora, Sandspit, Hawkesbay and
10.1016/S0304-3770(01)00160-7.
Cape Monze areas of Karachi coast. Can. J. Pure Appl. Sci. 4, 1053–1071.
Barkati, S., Rahman, S., 2005. Species composition and faunal diversity at three sites of
Kuenzer, C., Bluemel, A., Gebhardt, S., Quoc, T.V., Dech, S., 2011. Remote sensing of
Sindh mangroves. Pakistan J. Zool. 37, 17–31.
mangrove ecosystems: a review. Rem. Sens. 3, 878–928. https://doi.org/10.3390/
Bunting, P., Rosenqvist, A., Lucas, R., Rebelo, L.-M., Hilarides, L., Thomas, N., Hardy, A.,
rs3050878.
Itoh, T., Shimada, M., Finlayson, C., 2018. The Global Mangrove Watch—a new
Kumar, L., Mutanga, O., 2018. Google Earth Engine applications since inception: usage,
2010 global baseline of mangrove extent. Rem. Sens. 10, 1669. https://doi.org/
trends, and potential. Rem. Sens. 10, 1–15. https://doi.org/10.3390/rs10101509.
10.3390/rs10101669.
Li, H., Jia, M., Zhang, R., Ren, Y., Wen, X., 2019. Incorporating the plant phenological
Chen, B., Xiao, X., Li, X., Pan, L., Doughty, R., Ma, J., Dong, J., Qin, Y., Zhao, B., Wu, Z.,
trajectory into mangrove species mapping with dense time series Sentinel-2 imagery
Sun, R., Lan, G., Xie, G., Clinton, N., Giri, C., 2017. A mangrove forest map of China
and the Google Earth Engine platform. Rem. Sens. 11, 2479. https://doi.org/
in 2015: analysis of time series Landsat 7/8 and Sentinel-1A imagery in Google Earth
10.3390/rs11212479.
Engine cloud computing platform. ISPRS J. Photogrammetry Remote Sens. 131,
Li, M.S., Mao, L.J., Shen, W.J., Liu, S.Q., Wei, A.S., 2013. Change and fragmentation
104–120. https://doi.org/10.1016/j.isprsjprs.2017.07.011.
trends of Zhanjiang mangrove forests in southern China using multi-temporal
Congalton, R.G., 1991. A review of assessing the accuracy of classifications of remotely
Landsat imagery (1977–2010). Estuar. Coast Shelf Sci. 130, 111–120. https://doi.
sensed data. Remote Sens. Environ. 37, 35–46. https://doi.org/10.1016/0034-4257
org/10.1016/j.ecss.2013.03.023.
(91)90048-B.
Mackay, A., 2008. Climate change 2007: impacts, adaptation and vulnerability.
Dawn, 2009. Guinness certifies mangrove sapling world record. Daw. Newsp.
Contribution of working group II to the fourth assessment report of the
Diniz, C., Cortinhas, L., Nerino, G., Rodrigues, J., Sadeck, L., Adami, M., Souza-Filho, P.,
intergovernmental Panel on climate change. Journal of Environmental Quality,
2019. Brazilian mangrove status: three decades of satellite data analysis. Rem. Sens.
Cambridge, UK. https://doi.org/10.2134/jeq2008.0015br.
11, 808. https://doi.org/10.3390/rs11070808.
Marois, D.E., Mitsch, W.J., 2015. Coastal protection from tsunamis and cyclones
Donato, D.C., Kauffman, J.B., Murdiyarso, D., Kurnianto, S., Stidham, M., Kanninen, M.,
provided by mangrove wetlands - a review. Int. J. Biodivers. Sci. Ecosyst. Serv.
2011. Mangroves among the most carbon-rich forests in the tropics. Nat. Geosci. 4,
Manag. https://doi.org/10.1080/21513732.2014.997292.
293–297. https://doi.org/10.1038/ngeo1123.

15
H. Gilani et al. Estuarine, Coastal and Shelf Science 249 (2021) 107128

Masood, H., Afsar, S., Zamir, U. Bin, Kazmi, J.H., 2015. Application of comparative Sanderman, J., Hengl, T., Fiske, G., Solvik, K., Adame, M.F., Benson, L., Bukoski, J.J.,
remote sensing techniques for monitoring mangroves in Indus Delta, Sindh, Pakistan. Carnell, P., Cifuentes-Jara, M., Donato, D., Duncan, C., Eid, E.M., Ermgassen, P.Z.,
Biol. Forum Int. J. 7, 783–792. Lewis, C.J.E., Macreadie, P.I., Glass, L., Gress, S., Jardine, S.L., Jones, T.G.,
Memon, A.A., 2005. Devastation of the Indus river delta. Impacts of Global Climate Nsombo, E.N., Rahman, M.M., Sanders, C.J., Spalding, M., Landis, E., 2018. A global
Change, pp. 1–12. map of mangrove forest soil carbon at 30 m spatial resolution. Environ. Res. Lett. 13,
Memon, S.H., 2012. An overview of mangrove restoration efforts in Pakistan. In: 055002 https://doi.org/10.1088/1748-9326/aabe1c.
IUCN, Gland (Ed.), Sharing Lessons on Mangrove Restoration. Switzerland with Saura, S., 2002. Effects of minimum mapping unit on land cover data spatial
Mangroves for the Future,, Bangkok, Thailand, Mamallapuram, India, pp. 51–62. configuration and composition. Int. J. Rem. Sens. 23, 4853–4880. https://doi.org/
Mirza, M.I., Hasan, M.Z., Akhtar, S., Ali, J., Sanjrani, M.A., 1988. Remote sensing survey 10.1080/01431160110114493.
of mangrove forest along the coast of Balochistan. Mar. Sci. Arab. Sea 339–348. Simard, M., Zhang, K., Rivera-Monroy, V.H., Ross, M.S., Ruiz, P.L., Castañeda-Moya, E.,
Mondal, P., Liu, X., Fatoyinbo, T.E., Lagomasino, D., 2019. Evaluating combinations of Twilley, R.R., Rodriguez, E., 2006. Mapping height and biomass of mangrove forests
Sentinel-2 data and machine-learning algorithms for mangrove mapping in West in Everglades National Park with SRTM elevation data. Photogramm. Eng. Rem.
Africa. Rem. Sens. 11, 2928. https://doi.org/10.3390/rs11242928. Sens. 72, 299–311. https://doi.org/10.14358/PERS.72.3.299.
Mondal, P., Trzaska, S., de Sherbinin, A., 2017. Landsat-derived estimates of mangrove Soluri, E., Woodson, V., 1990. World vector shoreline. Int. Hydrogr. Rev. 67, 28–35.
extents in the Sierra Leone Coastal landscape complex during 1990–2016. Sensors Spalding, M., 2010. World Atlas of Mangroves. Routledge, London. https://doi.org/
18, 12. https://doi.org/10.3390/s18010012. 10.4324/9781849776608.
Mukhtar, I., Hannan, A., 2012. Constrains on mangrove forests and conservation projects Sultana, R., Mustaquim, J., 2003. Some physical parameters of the Sandspit backwaters,
in Pakistan. J. Coast Conserv. 16, 51–62. https://doi.org/10.1007/s11852-011- Karachi coast. Pak. J. Sci. Ind. Res. 46, 333–343.
0168-x. Syed, N.A., Siddiqa, T., 2019. The study of tidal current dynamics and impact of
Nazeer, M., Waqas, M., Shahzad, M.I., Zia, I., Wu, W., 2020. Coastline vulnerability bathymetry in training the currents along the coast of Karachi, Pakistan. Int. J. Mar.
assessment through landsat and Cubesats in a coastal mega city. Rem. Sens. 12, 749. Sci. Ocean Technol. https://doi.org/10.19070/2577-4395-1900015.
https://doi.org/10.3390/rs12050749. Tieng, T., Sharma, S., MacKenzie, R.A., Venkattappa, M., Sasaki, N.K., Collin, A., 2019.
Pekel, J.F., Cottam, A., Gorelick, N., Belward, A.S., 2016. High-resolution mapping of Mapping mangrove forest cover using landsat-8 imagery, Sentinel-2, very high
global surface water and its long-term changes. Nature 540, 418–422. https://doi. resolution images and Google earth engine algorithm for entire Cambodia. IOP Conf.
org/10.1038/nature20584. Ser. Earth Environ. Sci. 266, 012010 https://doi.org/10.1088/1755-1315/266/1/
Pendleton, L., Donato, D.C., Murray, B.C., Crooks, S., Jenkins, W.A., Sifleet, S., Craft, C., 012010.
Fourqurean, J.W., Kauffman, J.B., Marbà, N., Megonigal, P., Pidgeon, E., Herr, D., Tucker, C.J., 1979. Red and photographic infrared linear combinations for monitoring
Gordon, D., Baldera, A., 2012. Estimating Global “Blue Carbon” emissions from vegetation. Remote Sens. Environ. 8, 127–150. https://doi.org/10.1016/0034-4257
conversion and degradation of vegetated coastal ecosystems. PLoS One 7, e43542. (79)90013-0.
https://doi.org/10.1371/journal.pone.0043542. Tucker, C.J., Elgin, J.H., McMurtrey, J.E., Fan, C.J., 1979. Monitoring corn and soybean
Pimple, U., Simonetti, D., Sitthi, A., Pungkul, S., Leadprathom, K., Skupek, H., Som- crop development with hand-held radiometer spectral data. Remote Sens. Environ.
ard, J., Gond, V., Towprayoon, S., 2018. Google Earth Engine based three decadal 8, 237–248. https://doi.org/10.1016/0034-4257(79)90004-X.
Landsat imagery analysis for mapping of mangrove forests and its surroundings in UNEP-WCMC, 2014. The Importance of Mangroves to People: A Call to Action. United
the Trat province of Thailand. J. Comput. Commun. 247–264. https://doi.org/ Nations Environment Programme World Conservation Monitoring Centre (UNEP-
10.4236/jcc.2018.61025, 06. WCMC), Cambridge, UK.
Portengen, L., 2017. Classifying Mangroves in Vietnam Using RARAR and Optical Vo, Q.T., Oppelt, N., Leinenkugel, P., Kuenzer, C., 2013. Remote sensing in mapping
Satellite Remote Sensing: Processing Sentinel-1 and Sentinel-2 Imagery in Google mangrove ecosystems - an object-based approach. Rem. Sens. 5, 183–201. https://
Earth Engine. Delft University of Technology. doi.org/10.3390/rs5010183.
Puyravaud, J.P., 2003. Standardizing the calculation of the annual rate of deforestation. Waqas, M., Nazeer, M., Shahzad, M.I., Zia, I., 2019. Spatial and temporal variability of
For. Ecol. Manage. 177, 593–596. https://doi.org/10.1016/S0378-1127(02)00335- open-ocean barrier islands along the Indus Delta region. Rem. Sens. 11, 1–17.
3. https://doi.org/10.3390/rs11040437.
Rafique, M., 2018. A review on the status, ecological importance, vulnerabilities, and Weatherall, P., Marks, K.M., Jakobsson, M., Schmitt, T., Tani, S., Arndt, J.E., Rovere, M.,
conservation strategies for the Mangrove ecosystems of Pakistan. Pakistan J. Bot. 50, Chayes, D., Ferrini, V., Wigley, R., 2015. A new digital bathymetric model of the
1645–1659. world’s oceans. Earth Sp. Sci. 2, 331–345. https://doi.org/10.1002/2015EA000107.
Rasool, F., Saifullah, S.M., 1996. Mangroves of Kalmat Khor, Balochistan. Pakistan J. Bot. Wessel, P., Smith, W.H.F., 1996. A global, self-consistent, hierarchical, high-resolution
28, 143–150. shoreline database. J. Geophys. Res. Solid Earth 101, 8741–8743. https://doi.org/
Rasool, F., Tunio, S., Hasnain, S., Ahmad, E., 2002. Mangrove conservation along the 10.1029/96JB00104.
coast of Sonmiani, Balochistan, Pakistan. Trees Struct. Funct. 16, 213–217. https:// Wulder, M.A., Loveland, T.R., Roy, D.P., Crawford, C.J., Masek, J.G., Woodcock, C.E.,
doi.org/10.1007/s00468-001-0151-5. Allen, R.G., Anderson, M.C., Belward, A.S., Cohen, W.B., Dwyer, J., Erb, A., Gao, F.,
Saeed, U., Ahmad, S.R., Gilani, H., Nawaz, R.A.B., Shahzad, N., Ashraf, I., Qazi, W.A., Griffiths, P., Helder, D., Hermosilla, T., Hipple, J.D., Hostert, P., Hughes, M.J.,
2019. Monitoring mangroves plantation sites through integration of repeat Huntington, J., Johnson, D.M., Kennedy, R., Kilic, A., Li, Z., Lymburner, L.,
terrestrial photographs and spaceborne imagery. Biol. 65. McCorkel, J., Pahlevan, N., Scambos, T.A., Schaaf, C., Schott, J.R., Sheng, Y.,
Saher, N.U., Amanat, Z., Gondal, M.A., Qureshi, N.A., 2017. Distribution, abundance and Storey, J., Vermote, E., Vogelmann, J., White, J.C., Wynne, R.H., Zhu, Z., 2019.
population ecology of Ashtoret lunaris (forskel, 1775) and Matuta planipes fabricius, Current status of Landsat program, science, and applications. Remote Sens. Environ.
1798 from the Sonmiani Bay (lagoon), Pakistan. Pakistan J. Zool. 49, 455–465. 225, 127–147. https://doi.org/10.1016/j.rse.2019.02.015.
Saher, N.U., Siddiqui, A.S., Kanwal, N., Narejo, A.H., Gul, A., Gondal, M.A., Abbass, F.I., WWF-Pakistan, 2005. GIS Remote Sensing Based Assessment of Mangrove Resources of
2019. An overview of pollution dynamics along the Pakistan coast with special Selected Project Sites of Indus Delta and Makran Coast. Lahore, Pakistan.
reference of nutrient pollution. Marine Ecology: Current and Future Developments, Zafar Khan, M., Akbar, G., 2012. In: Boon, P.J., Raven, P.J. (Eds.), In the Indus Delta it Is
pp. 136–172. https://doi.org/10.2174/9789811412691119010012. No More the Mighty Indus, River Conservation and Management. John Wiley &
Saifullah, S.M., 2017. The effect of Global warming (Climate Change) on mangroves of Sons, Ltd, Chichester, UK, pp. 69–78. https://doi.org/10.1002/9781119961819.ch6.
Indus Delta with relevance to other prevailing anthropogenic stresses - a critical Zia, I., Zafar, H., Shahzad, M.I., Meraj, M., Kazmi, J.H., 2017. Assessment of sea water
review. Eur. Acad. Res. V 2110–2138. inundation along Daboo creek area in Indus Delta Region, Pakistan. J. Ocean Univ.
Saifullah, S.M., 1982. Mangrove ecosystem of Pakistan. The Third Research on China 16, 1055–1060. https://doi.org/10.1007/s11802-017-3350-4.
Mangroves in Middle East. Japanese Cooperation Research Centre, Tokyo, Japan,
pp. 69–80.

16

You might also like