Jamaican Mangrove Forests and Remote Sensing
Jamaican Mangrove Forests and Remote Sensing
Trevor Brinkman
GSCI 515-01
Unity Environmental University
October 2023
Abstract
Mangrove swamps are notoriously difficult to study and quantify, yet these tropical ecosystems serve as critical habitat for juvenile aquatic species and birds, manage sedimentation and storm damage while also sequestering large amounts of carbon into living biomass. Jamaica's eastern Portland Parish has several areas populated with mangrove, as these coastal wetlands are found along many tropical coastlines. The prospect of using remote sensing and GIS to identify and quantify the mangroves will help build existing knowledge by focusing on a specific area of Jamaica's coast. Understanding the current distribution of mangroves and calculating the changes and causes of changes from historic levels could be used to educate and influence policy. This data could also be used for predicting future changes as a result of human actions, climate change and the looming prospect of sea level rise. Much work has been done in refining the techniques of identification of mangroves using Landsat data, with the calculations of the NDVI (Normalized Difference Vegetative Index) and the NDWI (Normalized Difference Wetlands Index) being utilized in the determination of the CMRI (Combined Mangrove Recognition Index) as a reliable and accurate identifier of mangrove forests. This data can than be classified and comparisons with past data can help identify change causes of change and for making future predictions and models. While ground referenced GPS data is best for accuracy assessment, random sampling and classification of aerial imagery can be also be used. This project failed to generate quality results, but an effort was made to relate the maps produced to ground knowledge and mangrove resources available from other sources like NASA and the United Nations. There remain questions about image pre-processing and other issues in methodology which require additional research and practice.
<When I teach high school students to write lab reports, I tell them the Abstract should answer three questions: What did I do? Why did I do it? What did I learn? I also tell them there shouldn't be any commentary or editorializing in an Abstract.
I tried to map the mangroves of Portland Parish, Jamaica. I figured that remote sensing and GIS would be a great way to quantify large areas of difficult-to-map wetland forests, and there seemed to be a lot of work already done on this concept. I learned that there are some critical preprocessing steps that, if not done, result in non-existent data.>
Maps 1, 2 and 3 - Identifying the Coastline Zone of Portland Parish, Jamaica
Introduction
Portland Parish is located on the northeastern corner of the West Indian Caribbean island of Jamaica. One of 14 parishes found in Jamaica, Portland is considered one of the more “tropical”, as they directly receive the trade winds from the northeast. These winds collect moisture as they cross the ocean, but they drop the water as they move inland into the John Crow and Blue Mountains. The mountains drain down to the coast, which consists of beaches, rocky cliffs and mangrove swamps. Mangrove swamps are a tropical wetland dominated by trees with roots which are highly adapted to the dynamic and complex environment of the coastal ecosystem. They create natural barriers to severe weather, trap river sediments, have high diversity and are the perfect habitat for juvenile marine species. Mangroves have also been found to be incredibly good at sequestering carbon dioxide. Mangrove forests are found around the island, but larger populations are localized along the southern coast.
Jamaica is a developing nation and former British colony. Parts of its government and infrastructure are in line with American and European standards, but there are also remain other large gaps in the data and services. A cursory examination finds voids in certain specific data, but the larger scale data collected by LANDSAT can be used to quantify existing mangrove forests and determine factors impacting past and future changes in these critical habitats.
The primary research question this project seeks to answer:
How have the mangrove forests changed in the last 30-40 years along the coast of Portland Parish, Jamaica, WI?
Additional questions that may contribute to a deeper understanding of the role of mangrove forests and their role in Jamaican society:
What factors have contributed to the changes that have occurred in the mangrove forests of Portland Parish, Jamaica, WI?
How will sea level rise and climate change impact the mangrove forests of Portland Parish, Jamaica, WI?
Literature Review: Remote Sensing of Mangrove Swamps
Mangrove swamps are found in brackish coastlines that encircle the earth within the tropic zone. The primary producers of these ecosystems are highly adapted trees, and mangrove swamps serve numerous economic and environmental functions. In addition to producing firewood, charcoal and alcohol, the formation of pole like-roots and a broad canopy provide shelter and nutrition for an area of high biodiversity, while also slowing sediments and buffering against the impact of coastal storms. From a global perspective, evidence suggests that mangroves are two to four times more effective at sequestering carbon than mature tropical forests. (Murdiyarso, et al., 2015) Mangrove swamps have decreased by approximately 50% in the last century. (Long and Giri, 2011) Ghana saw a 24% decrease between the years of 1980-2006 (Mensah, 2013). Activities such as coastal development, aquaculture and factors such as pollution, climate change, sea level rise and other forces all contribute to these losses. In one estimate, 75% of the habitat loss in southern China’s Greater Bay Area are the result of anthropogenic factors (He, et al., 2022).
Given the importance of these swamps, it is important to accurately identify and quantify these regions to improve our understanding and improve efforts of conservation and restoration. However, the density and structure of the trees and their geographic inaccessibility make them historically difficult to study. Remote sensing and GIS provide an opportunity to observe, study and analyze these ecosystems. Improved sensor data and geospatial techniques have led to the growth of our knowledge of mangrove ecology. Evolving technologies and increased processing power allow for ever improved interpretation.
Any survey of mangrove research is a virtual tour around the equatorial belt. The diversity and characteristics of individual swamps are impacted by variations in salinity, soil mineral content, and annual precipitation patterns. In general, mangrove swamps tend to have low tree biodiversity, with many only having only a half dozen or so species, such as the swamps of Ghana (Mensah, 2013) and the proposed study areas on the island of Jamaica, which is home to three species (Jamaican National Environment and Planning Agency -NEPA, 2020). NEPA also identifies four specific locations within Portland Parish and provides specific details about the mangrove and other environmental features at each spot. Other studies identified as many as 35 species, but these involved study areas of a much larger area, like the Queensland coast of Australia (Chamberlain, Phian and Possingham, 2021) and the entire Philippine archipelago (Long and Giri, 2011).
Remote sensing enhanced our abilities to study mangrove forests. The majority of studies reviewed here used imagery provided Landsat 7 TM and Landsat 8 OLI, with some even using data obtained by Landsat 5 (He, et al., 2022 and Erfanifard, Mohsen and Sterneczak, 2022). Other articles identified RapidEye, JAXA’s AW3D, QuickBird, WorldView-2, Sentinel-2 and NASA’s Shuttle Radar Tomography Mission (SRTM) as sources of imagery. Higher resolution aerial photographs and LiDAR data were also used, but these type of data are not always accessible or available.
Further data sources were also used based on the needs of individual studies. An analysis of the Sundarbans utilized ground collected salinity data (Mondal, et al., 2021). The effort to better understand the phenology of Australian mangroves utilized data collected from weather stations along the coast (Chamberlain, Phian and Possingham, 2021). Iranian studies of Persian Gulf mangrove swamps also utilized tide data, as their analysis compared imagery from both high and low tides in an effort to determine the impact of daily episodic sea level variations on the classification of mangroves (Erfanifard, Mohsen and Sterneczak, 2022).
The science of mangrove studies using remote sensing has utilized a variety of different processing techniques, but most started with a similar set of steps. Most studies identified their study areas and located appropriate cloud-free imagery. These were usually band stacked, georeferenced and projected in a manner that was appropriate to the study area. In some cases, like Elmahdy (2022) and Gupta, et al. (2018), additional corrections were made for atmospheric interference and angle of illumination. In an effort to limit the data used in processing, they were cropped to a small area, usually with some sort of limited elevation buffer, as mangroves are only found in coastal environments.
A wide range of methods were used to analyze the imagery from this point. Many used some type of specified or unspecified classification. The literature review showed that many different techniques were used, with varying results. Some studies limited themselves to classification of mangrove forests based on NDVI (He, et al., 2022). Erfanifard, Mohsen and Sterneczak (2022), as shown in Fig. 1, compared data from four vegetative indices (VI) and eight mangrove specific (MSVI) indices. Gupta (2018) used only the CMRI, or Combined Mangrove Recognition Index, in their identification of mangrove in three locations across Southeast Asia.
Fig. 1 Image source: Erfanifard, Y., Mohsen, L., Sterenczak, K. (2022) Assessment of Iran’s Mangrove Forest Dynamics (1990-2020) Using Landsat Time Series. Remote Sensing; 14(19), 4912. https://doi.org/10.3390/rs14194912
A variety of classification systems were employed. Some studies processed the data through supervised classifications of visual analysis (Wang, 2003). Wang utilized the “mangrove” and “not mangrove” system of classification, while others developed systems with four categories (Mensah, 2013). Long and Giri (2011) used a process of unsupervised classification, which was appropriate for their large study area of the entire Philippine coastline.
An earlier in-depth ground study of a mangrove on the Port Royal peninsula in Kingston Harbor showed it to be remarkably stable in its location and size for the last 300 years (Allang, 1998). While this cannot be used as a comparison or stand-in for data regarding changes in Portland Parish’s mangroves in modern times, it suggests an interesting topic of worthy of additional investigation.
Another important factor was the methods and data used in determining the accuracy of the products resulting from processing. The ground truthing of mangroves provide their own problems, so there are a variety of techniques of assessing accuracy and determining the impact of inaccuracies on errors. In ideal situations, accurate GPS data provides opportunities to compare the GIS mangrove maps with the reality of geography. Mensah’s (2013) evaluation of the mangroves of Ghana were compared against data from 112 points on the ground. Wang, et al., (2003) study of the Tanzanian coastline was examined against 324 ground control points, with many including geotagged photos with species identification. Other papers within this review utilized analysis of pixels by either visual comparison with higher resolution photography (He et al., 2022), or with the machine learning powered analysis of up to 24,000 pixels (Erfanifard, Mohsen and Sterneczak, 2022).
Questions about errors, correlation and independent errors of both classification and truthing data should always be kept under consideration. This is especially true in examining the importance of ground collected data in support of analysis of data collected through remote sensing. Foody (2010) identified issues that may arise from data inaccuracies. They concluded small errors, in either the classification of data by GIS or the collection of ground data, may have large impacts. They also stressed that some errors may have easily identified correlation, while many others occur independently of one another.
A final topic in this review explored the role of UAV in the collection of data about mangrove forests. Zhu, et al., (2022) compared data collected from satellites with similar data collected by drones and found that drones can provide accurate, detailed information about mangrove, and their low cost and high resolution may provide higher quality, more accurate, timely and informative information. This provides another avenue of investigation that demands further investigation. The author of this review is interested in further exploring the role of UAV, especially when using multispectral imagery in wetland and coastal mapping.
Methods
Imagery Selection
Imagery was identified and downloaded from Glovis from Landsat 5, 7 and 8, spanning the years from 1987-2023. Additional 2023 data from "Collection 2, Level 2" was found using Earth Explorer. See Fig 2. The data was selected to provide a wide range of years, but also provide opportunities to observe changes in data from the same satellite, if possible. The data was also selected with preference for "cloud free" imagery collected between May and early August of any given year, to avoid any potential seasonal variations.
Additional data included a shapefile for Jamaican parishes to create a locator map and create buffers for clipping data to the coastal area of Portland Parish. Shapefiles of mangrove forest distribution were obtained from NASA Earth Data (2021) and the UN Environment World Conservation Monitoring Centre (2020). These provided references and results regarding the current status of mangrove forests from other sources.
Figure 2 - Landsat Data Acquisition Dates
Landsat 5 - May 4, 1987
Landsat 5 - July 10, 2000
Landsat 7 - June 19, 2001
Landsat 8 - August 5, 2015
Landsat 8 - May 7, 2023
Landsat 9 - August 3, 2023
Calculating the CMRI
Gupta, et al. (2018) showed that the calculation of the Combined Mangrove Recognition Index (CMRI) was a highly reliable mechanism for identifying mangrove forests. As illustrated in Fig. 1, the CMRI is calculated by using the formula (NDVI-NDWI), where NDVI is the Normalized Difference Vegetative Index and NDWI is the Normalized Difference Wetland Index. These are calculated using the Green, Red and Near Infrared (NIR). The appropriate bands were selected from the imagery collection and they were clipped to the coastline zone. The band rasters were then run through a series of calculations to determine the NDVI and NDWI, which were then geoprocessed to determine the CMRI, as shown by Fig. 3. It is important to note that Landsat 5 and 7 used similar band combinations, whereas Landsat 8 and 9 used bands 3, 4, and 5.
Figure 3 - Model Builder workflow for generating the NDVI, NDWI and CMRI.
Raster Classification
The maps would then indicate areas of mangroves. By applying an unsupervised classification system, the data would be divided into categories of "mangrove," "non-mangrove tree," "not forest" and "water." These categories would then provide data for comparison between the Landsat images. Simple statistical analysis of the data would provide insight into patterns of change. The process of change detection indicates areas of undergoing loss or growth, which can then be further investigated using higher resolution imagery.
It would also be possible to run supervised classification with training data.
Ground Referencing
Following reclassification, the process of stratified random sampling will be employed to create a pool of points for ground referencing with appropriate higher resolution imagery. Considering the small concentration of mangroves in the area and the large proportion of water, the random sampling would be done by weight rather than proportional. A larger portion of sample points will be selected from mangrove areas and very few will be chosen from the water, even though it makes up half of the data processed.
The identified points would then be referenced against higher resolution imagery to determine their accuracy.
Survey Information
In the study of changes in Jamaican Mangroves there are benefits to be able to identify them accurately both in the past and today. Hoping to establish a baseline for better measuring future changes from the collection of accurate data about their current status using drones and ground referencing would be ideal. However, surveys provide additional information about the current understanding and attitudes towards these coastal wetlands and could provide additional insight into making better decisions in the future.
Runs through a simple tool, Google Forms the surveys have a simple interface and a versatility in question construction. Options for response include binary choice, multiple choice, multiple answer, grid response, linear scale and text entry. These short, direct questions are divided between determining respondents’ knowledge of the history and importance of mangroves, their understanding of potential future changes, and their attitudes about the environment. This information should provide a (somewhat spatially related) better understanding of the community’s connection to mangroves and an additional set of baseline data for future investigation.
To make the data geospatial, the plan would be to use posters with QR codes placed along the coastal zone. The survey would be uniform along the entire coast, but the area would be subdivided into districts with individualized QR codes. This would provide a general, regional data set without identifying the spatial qualities of the data to the individual respondents. The interface would be maximized for cell phone access and rapid response. Jamaicans primary access to the internet is through their cellular phones and there are many different areas of focus along Portland’s coastal zone to target for sharing these surveys.
This survey would be totally voluntary and would try to target a wide range of the population along the coast. The coastal road is the easiest and most travelled way to move around the island. The population and the things associated with greater density are often localized along the coast- schools, churches, businesses, public transportation hubs, libraries, parks. The posters would need to be attractive, interesting and stimulate curiosity. They should also be available in standard paper sizes in black & white and full color, to ensure ease in printing and distribution. The process could start small, with emails to secondary school principals/science teachers, libraries and business owners who advertise on Google Maps.
In performing this survey there needs to be recognition of cultural differences in regards to technology, surveys and the environment. Do Jamaicans feel comfortable answering questions like those Offered? Are they worried about personal data collecting? Could their answers be traced back to them as individuals? Will they answer truthfully?
Survey Questionnaire Rationale:
1. Multiple Choice – this is a fact-filled prompt to engage the participant with information and is intended to prime and embed the concept of mangrove forests in their head.
2. Multiple Choice – Proximity. This could allow more focused attention in certain areas in which people report being closer to mangrove forests based on spatial data collected by QR code region.
3. Text Entry – Location of closest mangrove – This is the most difficult and probably least likely to receive responses, but it would be very useful in building a comprehensive knowledge of the local geography.
4. Multiple Answer – How have they learned information?
5. Multiple Answer – Who are the responsible parties?
6. Likert Scale – Past Changes – opinion about change until now…
7. Likert Scale – Future Changes – opinion – optimism or pessimism?
8. Multiple Choice – Threats – forced classification
9. Multiple Answer – Future engagement attitude
10. Multiple Choice – Age Range- four choices only, purposefully broad, but could really allow for some interesting analysis…
Link to Portland Parish Mangrove Survey - There are some inherent limitations to working with Google Forms as they only work through Google platforms…
Results
Map 4 illustrates the results of producing a CMRI using Landsat 8 data from 2023 and layering the data provided by NASA on top of it. At this scale it is difficult to determine whether or not the process resulted in producing recognizable mangrove identification, but it does demonstrate that remote sensing has been used by others to identify the mangroves in Jamaica. As can be seen, there is not a high concentration of mangrove forest in Portland Parish.
Map 4 - CMRI of Portland, Jamaica and NASA Mangrove Data
The appropriate band data from the various satellites over the years were processed using the raster calculating formulas. None yielded any clear, definitive results. Some areas that contain known mangroves were slightly exaggerated, but without any clear distinctions. Several of these resulting CMRIs were run through unsupervised classification to narrow the range of symbology, but areas that "popped out" as mangrove ended up with similar symbology as densely populated urban environments. I also manipulated the variables and methods for classifying data and displaying imagery, but none showed any distinct mangrove areas.
Maps 5-8 examine an area in eastern Portland Parish using Landsat 9 OLI Collection 2 - Level 2 data acquired on August 3, 2023. This data was selected for analysis as it has had a higher degree of preprocessing and was most likely to generate positive results. As the imagery shows, there are no distinct indications of mangroves in the CMRI analysis. (Click the map number to view higher resolution imagery)
Had I been able to successfully identify mangroves using the CMRI technique, I would then perform random sample analysis to determine the degree of reliability. Areas of known mangrove and non-mangrove would be identified on higher resolution imagery and the random samples would be compared to known data. Ideally, ground referencing would include GPS collected data points - with the potential for utilizing drone imagery for greater accessibility and resolution. However, personal experience in this area over the last two decades provides a degree of local knowledge. See Map 6. This understanding of the geography of the region does allow a small degree of analysis and interpretation in comparing the CMRI and NASA's mangrove map to the reality on the ground, as illustrated in Map 8.
NASA's Global Mangrove Distribution data shows a patchwork of mangrove forests along the coast. Some, like those at Priestman's River and White Sand are properly identified, although an unusually large data set is indicated at White Sand. You will find mangroves there, but only a narrow band between the coastal road and the Caribbean Sea. Other areas that have relatively large mangrove forests, like Reach River in the east and Turtle Crawle Swamp in the northwest appear in neither the CMRI nor the NASA data. When I think about Portland Parish's mangroves, these two areas are the first that come to mind, but they do not appear in either my data or that from NASA. There are also numerous locations in NASA's data that I've flagged as false positive "Improperly Identified" mangrove forests. Several of the areas towards the east that are improperly marked as mangrove are either residential or areas of elevated cliffs over a rocky coast. I do not have personal experience with the area immediately north of Priestman's River, so I did not assess the reliability of that area. I have spent a significant amount of time in the area marked as Fair Prospect & Long Bay. The areas marked as mangrove in this area are either areas of dense salt-tolerant plants that sit on exposed rock above the high tide line or they are swampy areas slightly inland where streams flow down from the steep slopes to the west of the coast.
Discussion
Mangroves are easy to identify with Landsat Imagery. That's what the literature tells us. Over the years there have been advances in sensors and processing techniques which provide more accurate information about the state of these hard-to-study biomes. The CMRI is a reliable, accurate tool that identifies mangrove forests using low resolution (30m) Landsat data. (Gupta, 2018) I set the goal of identifying changes in mangroves over time, yet I was unable to even identify mangroves using remotely sensed data. I was also unable to assess the accuracy of my data or examine any secondary research questions.
Figure 4 - Mangrove Workflow Summary
Image source: Gupta, K., Mukhopadhyay, A., Giri, S., Chanda, A., Majumdar, S.D., Samanta, S., Mitra, D., Samal, R.N., Pattnaik, A.K., Hazra, S. (2018) An index for discrimination of mangroves from non-mangroves using LANDSAT 8 OLI imagery. MethodsX 2018, 5, 1129–1139.
With proper pre-processing it is possible to locate mangrove swamps with a high confidence in the results. Fig. 4. The important part of this figure is the part that says, "Conversion of DN values to Reflectance," and this has been my struggle through this process, as it has prevented me from reaching the classification and referencing portion of this project. I had hoped to identify mangroves using remote data, and then identify changes over time. Ideally, I would also be able to model future changes and make predictions. Instead, I was limited by processes that lead to dead ends. That said, I learned a lot along the way.
Late in the project, while still scrabbling for potential solutions, I stumbled across A Survival Guide to Landsat Preprocessing from the journal Ecology. It did not solve my problems, but, among other things, it provided me with this quotation:
“This difficulty is exacerbated by preprocessing approaches that are similar but distinct, each with numerous possible workflows that analysts must navigate. Furthermore, explanations of the specific preprocessing steps taken in ecological studies vary; some offer detailed technical accounts, while others provide just a few broad summary sentences. These inconsistencies result in confusion and ambiguity, particularly for researchers and managers who would like to use Landsat data in their work but do not have a clear roadmap for how to do so.” (Young, et al., 2017)
This quotation summarizes my frustrations as I've worked through this process, and it provides some solace. The learning curve can be steep, and sometimes lacks road signs or maps. I have a better understanding of mangroves and their ecological importance. I have found that people are successfully utilizing remote sensing and backing it up with accuracy assessments using both GPS and remotely sensed data (such as aerial photographs and supervised classification.) As the tools, data sources and analytical processes improve, we have increased understanding of the current patterns of mangroves. We are also able to identify changes and causes of change, as well as make models and predict how potential disruptions will impact these important ecosystems.
Jamaican politics is a complex, ever evolving entity. The divisions between parties is deep and they have a difficult history. Post-colonial bureaucracies and past grievances, coupled with current fiscal consideration and political tensions, create conflicts which can hinder policy decisions and implementations. Understanding the current status of Jamaican mangrove forests could provide a good foundation of baseline data. Determining past changes and causes of change could help build understanding of the current situation, while also providing information for improved regulation and future actions regarding these critical habitats. Having accurate current data will also improve any modelling or predictions about future changes. The use of surveys and educational outreach could improve both data collection and informed decision making. As Jamaica develops its economy and improves its educational, research, infrastructure and other capacities it has a chance to utilize already existing data and analytical tools to make more effective and economically sound decisions to benefit both the environment and the population of the island.
Deep learning provides another potential source of analytics to improve mangrove identification and predictive modeling for future changes as a result of direct human activity and climate change. Efforts were made in this project to install and employ deep-learning mechanisms for mangrove analysis but the installation of the Python programs was never successful. Efforts to uninstall and reinstall the software were not undertaken when others managed to delete their entire ArcGIS Python library in an effort to work through this problem.
One final drone side note...there are lots of benefits to using drones for mangrove mapping (see The Nature Conservancy Story Map and hit "UAVs" on the toolbar or scroll down to that section). As I learn more about GIS and remote sensing and consider my future topics of research, I am drawn more and more to near-shore and wetland applications of drone collected sensor data. This may be a path of future exploration, as well as temperate wetlands or other shallow water, near-shore features.*
Map 9 - CMRI with gamma value = 10
*At one point a classmate suggested I adjust the gamma values of my raster projection. When I pushed the value to the high end on my CMRI, most of the imagery got washed out. The one thing that jumped out...coral reefs. Another remote sensing identification rabbit hole?...
References and Data Sources
References
Alleng, G. P. (1998). Historical Development of the Port Royal Mangrove Wetland, Jamaica. Journal of Coastal Research, 14(3), 951–959. http://www.jstor.org/stable/4298847
Bunting, P.; Rosenqvist, A.; Hilarides, L.; Lucas, R.M.; Thomas, T.; Tadono, T.; Worthington, T.A.; Spalding, M.; Murray, N.J.; Rebelo, L-M. Global Mangrove Extent Change 1996 – 2020: Global Mangrove Watch Version 3.0. Remote Sensing. 2022
Chamberlain, D.A.; Phinn, S.R.; Possingham, H.P. (2021) Mangrove Forest Cover and Phenology with Landsat Dense Time Series in Central Queensland, Australia. Remote Sens. 2021, 13, 3032. https://doi.org/ 10.3390/rs13153032
Classify Mangroves Using Deep Learning (2023) ESRI training tutorial. URL: https://learn.arcgis.com/en/projects/classify-mangroves-using-deep-learning/
Erfanifard, Y., Mohsen, L., Sterenczak, K. (2022) Assessment of Iran’s Mangrove Forest Dynamics (1990-2020) Using Landsat Time Series. Remote Sensing; 14(19), 4912. https://doi.org/10.3390/rs14194912
Elmahdy, S., Ali, T. (2022) Monitoring Changes and Soil Characterization in Mangrove Forests of the United Arab Emirates Using the Canonical Correlation Forest Model by Multitemporal of Landsat Data. Frontiers in Remote Sensing. https://doi.org/10.3389/frsen.2022.782869
Foody, G. (2010) Assessing the accuracy of land cover change with imperfect ground reference data. Remote Sensing of Environment, 114: 2271-2285. DOI: 10.1016/j.rse.2010.05.003
General Information on Jamaica’s Mangrove Wetland Resources 2020 (2023) Jamaican National Environment and Planning Agency. URL: https://websitearchive2020.nepa.gov.jm/policies/draft/Mangroves%20and%20Coastal%20Wetland%20Protection/MANGROVE%20AND%20COASTAL%20WETLANDS%20PROTECTION%20annex.html#:~:text=Types%20of%20Mangrove%20Wetlands,different%20characteristics%20and%20special%20values
Giri, C. (2016) Observation and Monitoring of Mangrove Forests Using Remote Sensing: Opportunities and Challenges. Remote Sens. 2016, 8, 783.
Giri, Chandra ed. (2021) Remote Sensing in Mangroves. MDPI. DOI 10.3390/books978-3-0365-0851-1
Gupta, K., Mukhopadhyay, A., Giri, S., Chanda, A., Majumdar, S.D., Samanta, S., Mitra, D., Samal, R.N., Pattnaik, A.K., Hazra, S. (2018) An index for discrimination of mangroves from non-mangroves using LANDSAT 8 OLI imagery. MethodsX 2018, 5, 1129–1139.
He, T.; Fu, Y.; Ding, H.; Zheng, W.; Huang, X.; Li, R.; Wu, S. (2022) Evaluation of Mangrove Wetlands Protection Patterns in the Guangdong–Hong Kong–Macao Greater Bay Area Using Time-Series Landsat Imageries. Remote Sens. 2022, 14, 6026. https://doi.org/10.3390/ rs14236026
Lymburner, L., Bunting, P., Lucas, R., Scarth, P., Alam, I., Phillips, C., Ticehurst, C., Held, A. (2020) Mapping the multi-decadal mangrove dynamics of the Australian coastline.Remote Sensing of Environment. 2020, v 238. https://doi.org/10.1016/j.rse.2019.05.004.
Long, J., Giri, C. (2011) Mapping the Philippines’ Mangrove Forests Using Landsat Imagery. Sensors, 11: 2972-2981. DOI:10.3390/s110302972
Mapping Mangroves (2023) Then Nature Conservency. URL: https://storymaps.arcgis.com/stories/75615bb17fcd4704940b97fea318ad59
Mapping the Mighty Mangrove (2023) NASA Landsat Science. URL: https://landsat.gsfc.nasa.gov/article/mapping-the-mighty-mangrove/
Mensah, J. (2013) Remote Sensing Application for Mangrove Mapping in the Ellembelle District in Ghana. USAID Integrated Coastal and Fisheries Governance Program for the Western Region of Ghana. Narragansett, RI: Coastal Resources Center, Graduate School of Oceanography, University of Rhode Island. 24 pp. https://www.crc.uri.edu/download/GH2009DAZ004_508.pdf
Mondal, I., Ghosh, P., Thakur, S., Kumar De, T., & De, T. K. (2021). Assessing the Impacts of Global Sea Level Rise (SLR) on the Mangrove Forests of Indian Sundarbans Using Geospatial Technology. In S. Singh, S. Kanga, & G. Meraj (Eds.), Geographic Information Science for Land Resource Management. Wiley. Retrieved September 1, 2023, from https://search.credoreference.com/articles/Qm9va0FydGljbGU6NDkyNDk3OA==.
Murdiyarso, D., et al. (2015) The potential of Indonesian mangrove forests for global climate change mitigation. Nature Climate Change. 2015, 5, 1089–1092. https://www.nature.com/articles/nclimate2734
Wang, Y. et al. (2003) Remote Sensing of Mangrove Change Along the Tanzania Coast. Marine Geodesy, 26: 35-48. DOI: 10.1080/01490410390181243
Wikipedia contributors. (2023). Mangrove forest. Wikipedia. https://en.wikipedia.org/wiki/Mangrove_forest#/media/File:Mangroves_at_sunset.jpg
Wu, JiaYu. (2023). The Classification of Mangrove subtypes in Northwestern Madagascar based on Planet data. https://storymaps.arcgis.com/stories/1d29f12064d64fe7b386961b84eedae9
Young, N., Anderson, R., Chignell, S., Vorster, A., Lawrence, R., and Evangelista, P. (2017). A Survival Guide to Landsat Preprocessing . Ecology. 98. 920-932. DOI:10.1002/ecy.1730
Zhu, Z., Huang, M., Zhou, Z., Chen, G. and Zhu, X. (2022) Stronger conservation promotes mangrove biomass accumulation: Insights from spatially explicit assessments using UAV and Landsat data. Remote Sens Ecol Conserv, 8: 656-669. https://doi.org/10.1002/rse2.268
Data Sources
Global Mangrove Distribution, Aboveground Biomass, and Canopy Height, 2021 [download] Nasa Earth Data. https://daac.ornl.gov/cgi-bin/dsviewer.pl?ds_id=1665 [September 4, 2023].
Global Mangrove Watch 2020 data [download] UN Environment World Conservation Monitoring Centre. URL: https://data.unep-wcmc.org/datasets/45 [September 24, 2023].
Jamaica Parish Boundaries 2020 [download] Esri, Michael Bauer Research GmbH 2021, Statistical Institute of Jamaica, UN. URL https://www.arcgis.com/home/item.html?id=eb43bde24cca4bce9ac5e0f8a797dc42 [August 27, 2023].
Landsat 5 data for July 10, 2000 [download] GloVis. URL: https://glovis.usgs.gov/app [September 04, 2023].
Landsat 5 data for May 4, 1987 [download] GloVis. URL: https://glovis.usgs.gov/app [September 04, 2023].
Landsat 7 ETM+ data for June 19, 2001 [download] GloVis. URL: https://glovis.usgs.gov/app [September 04, 2023].
Landsat 8 OLI data for May 7, 2023 [download] GloVis. URL: https://glovis.usgs.gov/app [September 04, 2023].
Landsat 8 OLI data for August 5, 2015 [download] GloVis. URL: https://glovis.usgs.gov/app [September 23, 2023].
Landsat 9 OLI data for August 3, 2023 [download] GloVis. URL: https://glovis.usgs.gov/app [October 05, 2023].