Evaluating mangrove conservation and sustainability through spatiotemporal (1990–2020) mangrove cover change analysis in Pakistan
Introduction
Mangroves are one of the most productive ecosystems, which offer worthy services for climate change adaptation and mitigation. They protect coasts from natural disasters (e.g. tsunami, cyclone), help clean the creeks and channels, and act as a buffer against sea intrusion and soil erosion (Khan et al., 2010; Marois and Mitsch, 2015). They recycle nutrients from livestock and human settlements (Adhikari et al., 2010). Mangroves have the potential to sequestrate and store substantial amounts of carbon, approximately 18 times greater than other forest types (Alongi, 2012; Donato et al., 2011; Sanderman et al., 2018). Like other wetlands, mangroves are important components of the water cycle, as they absorb excess water flow during the time of floods. The primary productivity of mangroves in deltaic areas is seven times higher than coastal areas without mangroves (UNEP-WCMC, 2014). The Sustainable Development Goal (SDG) 14, “Life Below Water”, emphasizes restoring and protecting mangroves for sustainable use of oceans, seas, and marine resources.
Despite the ecological value of mangroves, they are significantly threatened and vulnerable to climate changes (e.g. sea level rise, variations in ocean current, etc.), natural disasters (e.g. tsunami, cyclone, earthquake, etc.) and anthropogenic exploitation (e.g. pollution, siltation, urban development, over-harvesting, construction of shrimp ponds, and grazing, etc.). According to the most reliable estimates, at the global scale, mangroves are being lost at an alarming rate of 2% per year, with 0.15–1.02 Pg (billion tons) of carbon dioxide being released annually, resulting in $US 6–42 billion annual economic damages (Pendleton et al., 2012). Alarmingly, the Intergovernmental Panel on Climate Change (IPCC) has predicted that in the following 100 years, ~30–40% of the coastal environment and 100% mangrove forested areas could vanish if the prevailing rate of loss continues (Mackay, 2008).
Pakistan's coastline is 1050 km long and 40–50 km wide zone, distributed between the Sindh (350 km) and Balochistan (700 km) provinces, and geographically placed between 24° and 25° N Latitude and 61°–68° E longitude (Abbas et al., 2013; Amjad and Jusoff, 2007). In 1958, mangroves of Pakistan were declared “protected forests” under the Pakistan Forest Act 1972, and the entire forest areas with the water channels present therein were declared “wildlife sanctuaries” in 1977 under the Sindh Wildlife Safety Ordinance of 1972 Act (Rafique, 2018; Saeed et al., 2019). In Pakistan, mangroves are distributed in five separate, unconnected geographical pockets: Indus Delta and Sandspit in Sindh province; and Sonmiani Khor (Miani Hor), Kalmat Khor (Kalmat Hor), and Jiwani (Gwadar Bay) in Balochistan province (Fig. 1). According to the United Nation's Food and Agriculture Organization (FAO), on average natural tree cover in Pakistan diminishes by about 316 km2 annually (FAO, 2010). In an effort to counter deforestation and forest degradation, and to ensure a climate-resilient future for Pakistan, the Government and private sector have taken several small to large-scale plantation initiatives across the country to improve the tree cover (Kamal et al., 2019).
Earth observation satellite datasets and geospatial tools offer a unique possibility for quantifying changes occurring on the earth's surface, whether through human impact or climate change (Wulder et al., 2019). Temporal satellite data is a viable solution for the quantification of land features in a comprehensive and systematic manner. With the advantage of freely available earth observation satellite images [e.g. MODIS (Moderate Resolution Imaging Spectroradiometer), Landsat, Sentinel, etc.] and Analysis Ready Data (ARD) products [e.g. Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index (NDWI), Land Surface Temperature (LST), etc.], it becomes much easier and transparent to quantify and report near real-time changes occurring on the earth's surface. In forestry and particularly in mangroves, satellite remote sensing is frequently being used to evaluate changes in cover, density dynamics, and biomass budgeting (Elmahdy et al., 2020; Giri et al, 2011, 2015; Kathiresan et al., 2013; Kauffman and Bhomia, 2017; Kuenzer et al., 2011; Mondal et al., 2019; Simard et al., 2006; Vo et al., 2013). Open access of big geospatial datasets, cloud computing platforms, and machine learning algorithms offer new possibilities for global developments and societal benefits.
The Google Earth Engine (GEE) is one of the cloud-based platforms which allows algorithms development and provides a convenient mechanism for scientists to share, visualize and analyze remote sensing data and by-products (Gorelick et al., 2017; Kumar and Mutanga, 2018). A number of researchers have utilized GEE for the assessment and reporting of mangrove dynamics.
In GEE, Chen et al. (2017) developed a new classification algorithm based on the biophysical characteristics of mangroves in China using 2015 Landsat-7/8 and Sentinel-1A imagery. Portengen (2017) completed a master's dissertation on classifying mangroves in Vietnam using Sentinel-1 and 2 satellite datasets in GEE platform, achieved with up to an overall accuracy of 87% of Random Forest (RF) machine learning classifier. Mondal et al. (2017) estimated mangroves extent on the coast of Sierra Leone using Landsat datasets (1990, 2000, 2010 and 2016) by adopting k-means unsupervised classification approach using GEE. Pimple et al. (2018) carried out mangroves mapping study over the Trat province, Thailand and surrounding areas to analyze three decades Landsat imagery (1987 and 2017) in GEE. Based on the RF classifier, they obtained 87% and 96% overall accuracy for the years 1987 and 2017 respectively (Pimple et al., 2018). Mondal et al. (2019) evaluated two machine learning algorithms, RF, and Classification And Regression Trees (CART) in GEE using Sentinel-2 along the coast of Senegal and Gambia, West Africa. They found for the mapping of mangrove cover, RF is better than CART, with an overall accuracy of 93.44% ± 1.37% and 92.18% ± 1.29% respectively (Mondal et al., 2019). In GEE, for the mangrove cover mapping of entire Cambodia, Tieng et al. (2019) applied RF classifier on Landsat-8, Sentinel-2 and Google Earth derived very high-resolution images. To study the status of the Brazilian mangroves, Diniz et al. (2019) analyzed three decades (1985–2018) of Landsat imagery from GEE environment by computing the Modular Mangrove Recognition Index (MMRI). Li et al. (2019) investigated the plant phenological trajectory for mangroves species mapping in GEE using Sentinel-2 derived time series NDVI over the 2.5 km2 Zhangjiang estuary in Fujian Zhangjiangkou National Mangrove Nature Reserve (FZNNR), China.
In Pakistan, only a handful of studies have been conducted to assess the changes in mangrove cover using satellite data or field-based inventory with statistical estimations.
Masood et al. (2015) assessed mangrove cover changes over the Indus Delta (~6200 km2 total area taken in this study) between 2009 and 2014, using Landsat-5 and 8 satellite images respectively (30 m spatial resolution respectively), by comparing pixel-based supervised maximum likelihood classification and onscreen digitization techniques. Giri et al. (2015) studied Indus Delta mangrove cover change in terms of gain and loss using 1973 Landsat MSS (60m) and 2010 Landsat TM (30 m) images by adopting supervised and object-based classification techniques respectively. Abbas et al. (2013) used Advanced Land Observation Satellite (ALOS) and Advanced Visible and Near Infrared Radiometer type 2 (AVNIR-2) 10 m spatial resolution satellite images for the year 2009, to assess nationwide mangrove cover through object-based classification technique. WWF-Pakistan (2005) conducted a study at Keti Bandar (a small portion of the Indus Delta), Sonmiani Khor, Kalmat Khor and Sandspit sites using Landsat TM (30 m), Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) Terra instrument (15 m) and QuickBird (0.5m) images through pixel-based supervised maximum likelihood classification and onscreen digitization land cover mapping techniques. Prior to this, in 1983 through onscreen digitization at the scale of 1:250,000, the Space and Upper Atmosphere Research Commission (SUPARCO) of Pakistan reported the mangrove cover of the Indus Delta using Landsat MSS 60 m spatial resolution datasets of 1978 (IUCN, 2005; WWF-Pakistan, 2005). For national level mangrove cover assessment, another study was conducted by SUPARCO using the 20 m spatial resolution SPOT multispectral imagery of the year 2003 through digitization at 1: 50,000 scale. In 1999, the Forestry Sector Master Plan (FSMP) land cover map produced from Landsat images of 1988–1991 through onscreen digitization technique, while the National Forest and Range Resources Assessment Study (NFRRAS) was done through pixel-based image classification of Landsat images of 1997–2001 (Abbas et al., 2013). Apart from these assessments, remote sensing, field-based, in-situ and statistical assumption based number of mangrove cover studies conducted between 1980 and 2005 were compiled and reported in a tabular form in the FAO (2007) mangroves country report.
Based on the literature review on mangrove cover mapping in Pakistan, it was observed that most of the studies incorporated in-situ information in satellite images classification and have used different spatial resolution satellite images, image classification methods, definitions and classification schemes, which provided contradictory results. The literature review also revealed that only a few studies have analyzed one-time satellite imagery to assess and map the status of mangrove cover. However, a number of studies used temporal satellite datasets to assess mangrove cover changes. As yet no comprehensive, systematic, well-recognized nationwide temporal assessment has been carried out. Further, there has been very little focus on spatial analysis on mangrove cover towards understanding the mangrove ecology. Indeed, globally produced one-time mangrove databases are openly and freely assessable, but at the national scale, spatiotemporally produced harmonized mangrove cover databases are needed for short to long terms strategic policy formations and implementation.
By taking advantage of open access 30 m spatial resolution Landsat satellite images and RF machine learning algorithm in GEE platform, the current study is designed to achieve the following objectives to assess the mangrove conservation and sustainability over the five mangrove areas in Pakistan.
- •
Detecting changes in mangrove cover using Landsat 30 m images at five-year intervals (1990, 1995, 2000, 2005, 2010, 2015 and 2020), across all five mangrove areas in Pakistan.
- •
Mangrove fragmentation analysis, to estimate changes over thirty years within the mangroves ecosystem.
Section snippets
Study sites
In Pakistan, mangroves exist at the following five sites: Indus Delta, Sandspit, Sonmiani, Kalmat Khor and Jiwani, which have been selected for this study (Fig. 1). The details of the sites are:
1. Indus Delta (central coordinates: 23.957 N latitude, 67.5015 E longitude, geographical area: 8000 km2), situated in Thatta district, Sindh province. It comprises 17 major creeks and numerous minor creeks (Amjad and Jusoff, 2007) and constitutes ~95% of total estuarine mangroves situated in Pakistan (
Results
The results of this study consist of accuracy assessment, mangrove cover change assessment, and mangrove fragmentation.
Discussion
This study used multi-date Landsat images and standard methodology to quantify mangrove cover changes from 1990 to 2020 at five-year intervals. High temporal and spectral resolutions of Landsat images with a low saturation level of spectral bands and integration of indices are the major factors that ensure the accuracy of land cover maps for temporal mangrove cover change estimation over the diverse study sites.
Spatially explicit and periodic mangrove cover information provided in this study
Conclusions
The study, being the first, provides mangrove cover and change assessment at five-year intervals over the five mangrove areas in Pakistan, for a period of thirty years (1990–2020), using consistent set of RF classification method and Landsat 30 m spatial resolution satellite images from open access GEE cloud computing platform. The current methodology was cost-effective and can be easily replicable over five-year intervals using the operation Landsat-8 or planned Landsat-9 satellite datasets.
CRediT authorship contribution statement
Hammad Gilani: Conceptualization, Methodology, Supervision. Hafiza Iqra Naz: Software, Formal analysis, Validation. Masood Arshad: Funding acquisition, Project administration. Kanwal Nazim: Writing - review & editing. Usman Akram: Validation. Aneeqa Abrar: Writing - review & editing. Muhammad Asif: Validation.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
References (82)
- et al.
A mangrove forest map of China in 2015: analysis of time series Landsat 7/8 and Sentinel-1A imagery in Google Earth Engine cloud computing platform
ISPRS J. Photogrammetry Remote Sens.
(2017) A review of assessing the accuracy of classifications of remotely sensed data
Remote Sens. Environ.
(1991)Assessing the accuracy of land cover change with imperfect ground reference data
Remote Sens. Environ.
(2010)NDWI—a normalized difference water index for remote sensing of vegetation liquid water from space
Remote Sens. Environ.
(1996)- et al.
Decadal land cover change dynamics in Bhutan
J. Environ. Manag.
(2015) - et al.
Distribution and dynamics of mangrove forests of South Asia
J. Environ. Manag.
(2015) - et al.
Google earth engine: planetary-scale geospatial analysis for everyone
Remote Sens. Environ.
(2017) - et al.
Uncovering the spatio-temporal dynamics of land cover change and fragmentation of mangroves in the Ca Mau peninsula, Vietnam using multi-temporal SPOT satellite imagery (2004–2013)
Appl. Geogr.
(2017) - et al.
Geospatial analysis of creeks evolution in the Indus Delta, Pakistan using multi sensor satellite data
Estuar. Coast Shelf Sci.
(2018) - et al.
Change and fragmentation trends of Zhanjiang mangrove forests in southern China using multi-temporal Landsat imagery (1977–2010)
Estuar. Coast Shelf Sci.
(2013)
Standardizing the calculation of the annual rate of deforestation
For. Ecol. Manage.
Red and photographic infrared linear combinations for monitoring vegetation
Remote Sens. Environ.
Monitoring corn and soybean crop development with hand-held radiometer spectral data
Remote Sens. Environ.
Current status of Landsat program, science, and applications
Remote Sens. Environ.
An assessment of status and distribution of mangrove forest cover in Pakistan
J. Biodivers. Environ. Sci. (JBES)
The use and management of mangrove ecosystems in Pakistan
J. Environ. Dev.
Status of mangroves of North-Western part of Indus Delta : environmental characteristics and populated structure
Pakistan J. Mar. Sci.
Carbon sequestration in mangrove forests
Carbon Manag.
Mangrove conservation through community participation in Pakistan: the case of Sonmiani Bay
Environ. Sci. Ecosyst. Dev.
Degradation of Indus delta mangroves in Pakistan
Int. J. Geol.
Experimental assessment of salinity tolerance of Ceriops tagal seedlings and saplings from the Indus delta, Pakistan
Species composition and faunal diversity at three sites of Sindh mangroves
Pakistan J. Zool.
The Global Mangrove Watch—a new 2010 global baseline of mangrove extent
Rem. Sens.
Guinness certifies mangrove sapling world record
Daw. Newsp.
Brazilian mangrove status: three decades of satellite data analysis
Rem. Sens.
Mangroves among the most carbon-rich forests in the tropics
Nat. Geosci.
Observations on the birds of Sandspit/Hawkesbay coastal wetland complex, Karachi coast
Pakistan J. Zool.
Spatiotemporal mapping and monitoring of mangrove forests changes from 1990 to 2019 in the northern emirates, UAE using random forest, Kernel logistic regression and naive Bayes tree models
Front. Environ. Sci.
Asia-pacific forests and forestry to 2020 report of the second Asia-pacific forestry sector outlook study, Second Asia-Pacific Forestry sector Outlook Study
Mangroves of Asia 1980-2005: Country Reports
Status of land cover classification accuracy assessment
Remote Sens. Environ.
From REDD+ MRV perspective: comparison of two different forest management regimes using geospatial techniques in Ludi Khola watershed, Gorkha district, Nepal
PFG – J. Photogramm. Remote Sens. Geoinf. Sci.
Status and distribution of mangrove forests of the world using earth observation satellite data
Global Ecol. Biogeogr.
Random forests for land cover classification
Pattern Recognition Letters
Coastline pollution problems of Karachi
Int. J. Emerg. Trends Eng. Dev.
Mangroves of Pakistan, Status and Management
Significance of billion tree tsunami afforestation project and legal developments in forest sector of Pakistan
Int. J. Law Soc.
Three decades of coastal changes in Sindh, Pakistan (1989–2018): a geospatial assessment
Rem. Sens.
Carbon sequestration potential of Rhizophora mucronata and Avicennia marina as influenced by age, season, growth and sediment characteristics in southeast coast of India
J. Coast Conserv.
Ecosystem carbon stocks of mangroves across broad environmental gradients in West-Central Africa: global and regional comparisons
PLoS One
Distribution, population status and environmental impacts on reptiles in Manora, Sandspit, Hawkesbay and Cape Monze areas of Karachi coast
Can. J. Pure Appl. Sci.
Cited by (43)
Sinking delta: Quantifying the impacts of saltwater intrusion in the Indus Delta of Pakistan
2023, Science of the Total EnvironmentWill short-term constraints affect long-term growth? Empirical analysis from the Beibu Gulf Mangrove National Natural Reserve of China
2023, Ocean and Coastal ManagementAutomatic forest change detection through a bi-annual time series of satellite imagery: Toward production of an integrated land cover map
2023, International Journal of Applied Earth Observation and GeoinformationRemote sensing for cost-effective blue carbon accounting
2023, Earth-Science ReviewsLeveraging machine learning and remote sensing to monitor long-term spatial-temporal wetland changes: Towards a national RAMSAR inventory in Pakistan
2023, Applied GeographyCitation Excerpt :Additionally, the inventory is also useful to quantify the spatial-temporal dynamics of several services provided by wetlands including flood protection, carbon storage, and economic incentives—left for future studies. Though there are no studies at hand to compare our large-scale and higher-resolution spatial-temporal assessment of Pakistan's wetlands, the increasing trends in the wetlands of Pakistan as identified in our study are in line with some local studies (Ahmad & Erum, 2012; Gilani et al., 2021). For example, Gilani (2021) performed a study on only mangroves in 5 main regions of Pakistan and found an overall increasing pattern in the mangrove area.