The Overview Of Blue Carbon Ecosystems Mapping Mangrove Ecosystems

0:04 Speaker 1: Welcome everyone to our RSET training series, Earth Observations of Blue Carbon Ecosystems. Today is part one of our training series, Overview of Blue Carbon Ecosystems and Mapping Mangrove Ecosystems with Earth Observations. My name is Brock Blevins. I'm a training coordinator for RSET and I'll be your host for this training. Before we begin today, I'll say a few words about the RSET program. RSET is an acronym which stands for the Applied Remote Sensing Training Program. RSAT is part of NASA's Earth Action Capacity Building Program.

0:38 Speaker 1: RSAT provides accessible, relevant, and cost-free trainings on remote sensing satellites, sensors, methods, tools, and applications. The goal of our training is to increase the use of Earth science remote sensing and model data in decision-making, and our trainings are tailored to audiences with a variety of experience levels, from introductory to advanced. On the right side of the panel, you can see the six different thematic areas in which we conduct trainings. Trainings are offered online and in person, and they're targeted for beginners and advanced practitioners alike.

1:11 Speaker 1: Trainings are freely available and conducted either live, instructor-led, or self-guided, such as our Fundamentals of Remote Sensing. Since 2009, the program has reached over 140,000 participants from over 185 countries. All our trainings are cost-free with materials offered in bilingual and multilingual languages and are freely available to use and adapt for your curriculum. If you use the methods and data presented in RSET trainings, please acknowledge the RSET Applied Remote Sensing Training Program.

1:43 Speaker 1: If you want to learn more, please visit our website. And now we'll start by giving an overview of this training series, Earth Observations of Blue Carbon Ecosystems. By the end of this training series, Participants will be able to identify blue carbon ecosystems and the role they play in the global carbon budget, map the extent of blue carbon ecosystems using satellite observations, measure the carbon stock of mapped blue carbon ecosystems, and identify contexts in which Earth observation data of carbon stocks and blue carbon ecosystems can inform reporting, monitoring, accounting, and advocacy.

2:25 Speaker 1: In part one today, there will be an introduction to the importance of these blue carbon ecosystems, key considerations for mapping mangrove ecosystems, and an overview of the existing mangrove data sets, and a demonstration in Google Earth Engine. In part two, we'll focus on salt marshes and seagrasses with demonstrations using Google Earth Engine.

2:48 Speaker 2: To map them.

2:50 Speaker 1: There will be one homework assignment after this training. It will be available after part two is complete. and due two weeks later on December 19th. A certificate of completion will be awarded to those who attend all live sessions and complete the homework assignment before the given due date.

3:07 Speaker 1: Now we'll begin part one of this training, overview of blue carbon ecosystems and mapping mangrove ecosystems with Earth observation.

3:19 Speaker 1: We are very fortunate today to be joined by Dr. Siti Mariam Yacoub, Senior Director for the International Blue Carbon Institute, Dr. Adia Bey, Assistant Research Scientist and Geospatial Analyst at NASA's Goddard Space Flight Center. We'll also be hearing from Dr. Lola Fatoyenbo, a Research Scientist also at NASA's Goddard Space Flight Center. If you have any questions during today's training, you can put them in the questions box within the WebEx at any time. At the end of the training, we'll try to address as many questions as we can, and we'll post these questions and written answers to the training page within a week after the training.

4:00 Speaker 1: And now I'm going to hand this off to Siti to start us off.

4:05 Speaker 3: Thank you very much.

4:12 Speaker 4: Hello, and welcome to the segment of the course where we will be introducing blue carbon ecosystems. The topics we'll be covering here today will provide context for understanding how the learnings from this training can be used to support policy and management of blue carbon. So let's begin with a definition of what blue carbon is. The technical definition of blue carbon is carbon that is captured, removed, and stored by ocean systems. both in the biotic components, or living things like flora and fauna, and the abiotic components.

4:46 Speaker 2: Such as sediments. The sediments in these coastal ecosystems are.

4:50 Speaker 4: Frequently inundated, which results in an oxygen-poor environment that causes slow breakdown of organic materials, which then results in significant carbon storage.

5:03 Speaker 2: Three ecosystems in particular.

5:05 Speaker 4: Mangroves, seagrasses, and salt marshes, have been identified to play an outsized role in carbon sequestration and storage. These three habitats are actionable as climate mitigation solutions and can therefore be integrated into current climate mitigation policy. The three ecosystems share these following traits. Number one, the presence of high carbon stocks. Number two, evidence of long-term carbon storage. And three, the capacity for people to manage and effectively measure GHG emissions and removals resulting from changes to these ecosystems.

5:45 Speaker 4: In contrast, while some ocean and coastal ecosystems and organisms are essential components of the ocean carbon cycle, their climate mitigation capacity is not directly responsive to management actions and they do not store carbon long term. So let's talk a little bit more about these three coastal blue carbon ecosystems. As mentioned in the first slide, there are two main components where carbon is stored. Carbon that is stored in the plant biomass and carbon that is stored in the sediments that these plants grow in.

6:21 Speaker 4: In the context of remote sensing, it is useful to think of these components as what's above ground or what's emerging and sticks out of the ground. and what's below ground, things like roots of the plants and sediments, which.

6:34 Speaker 2: Are hidden and therefore harder to detect from space.

6:39 Speaker 4: In mangroves, the parts that are above ground make up roughly over a third of total ecosystem carbon stocks, whereas the percentage of carbon in the above-ground biomass of seagrass.

6:50 Speaker 2: And salt marsh ecosystems are much lower at around 5% or less.

6:56 Speaker 4: This is because unlike mangroves, seagrass and salt marsh plants are smaller in size and they do not have.

7:01 Speaker 2: Woody components that contain a lot of carbon.

7:06 Speaker 4: In contrast, the below-ground carbon stock is higher in all three ecosystems, comprising around 60% of total carbon stock in mangroves and more than 95% of total carbon stocks for.

7:18 Speaker 2: Seagrass and salt marshes.

7:22 Speaker 4: So when we think about detecting blue carbon from space, what we are detecting.

7:26 Speaker 2: Are the ecosystem components that are emergent.

7:31 Speaker 4: Mangroves are probably the most readily detectable because mangrove trees are emergent and their leaves have a unique spectral signature that makes them amenable to earth observation approaches. Salt marshes are tidally inundated and they frequently occur in the upper intertidal zones. This means that they are rarely fully submerged and the above-ground plant biomass can also be detected via remote sensing approaches. Seagrasses, on the other hand, are marine plants that are usually underwater, but they can still be detected from space.

8:07 Speaker 4: However, factors such as the size of the seagrass species, water.

8:11 Speaker 2: Quality conditions such as turbidity, or the.

8:14 Speaker 4: Presence of algal blooms can impede their detection via remote sensing approaches.

8:20 Speaker 2: And last but.

8:21 Speaker 4: Not least, we have the sediments, and this is usually not detectable from space because it is covered by the plants that grow on top of them. Carbon that is found in the sediments of coastal blue carbon ecosystems can be split based on their origins. There is carbon that originates or forms within the ecosystems, otherwise known as autochthonous carbon, and there is carbon that is brought into the ecosystem, usually by way of hydrology and hydrodynamics. such as river runoff, currents, and waves, and this is known as the loctinus carbon.

8:57 Speaker 2: These forms of carbon.

8:58 Speaker 4: Are virtually indistinguishable in.

9:00 Speaker 2: The field and to the naked eye.

9:05 Speaker 4: I'll now turn our attention to each of the three actionable blue carbon ecosystems.

9:10 Speaker 2: First up, we have mangrove forests.

9:12 Speaker 4: These trees are typically found along sheltered tidal coastlines around river mouths and deltas, where there is an accumulation of fine, soft, silty to sandy sediments. They are regularly inundated by tidal exchange and can tolerate a wide range of salinities. Of the three actionable blue carbon ecosystems, mangroves sequester and store the largest amount of carbon due to the combination of their large biomass and the typically rich organic sediments in which they grow. They can sequester between 6 and 8 tons of carbon dioxide per hectare annually.

9:49 Speaker 4: Seagrass are marine flowering plants that typically grow in sedimentary environments along shallow.

9:56 Speaker 2: And sheltered coastal waters.

9:57 Speaker 4: They have a global distribution and can be found in both tropical and temperate regions spanning.

10:04 Speaker 2: Every continent except Antarctica. There are more than 60.

10:07 Speaker 4: Species of seagrass globally, with the highest diversity occurring in the Indo-Pacific. Seagrass species have a wide range of morphologies with leaf sizes ranging from as small as one centimeter to as large as 1.5 meters in length. Most of the carbon stored in seagrass occur below ground in their deep root systems and the sediments that they grow in. The carbon sequestration rate in seagrass range from between 1.5 to 4 tons of.

10:36 Speaker 2: Carbon dioxide per hectare per year.

10:39 Speaker 4: And seagrasses are particularly effective at trapping sediments, and are a good barometer for both water quality and ocean health.

10:49 Speaker 4: Salt marshes are coastal wetlands characterized by their saline conditions, which result from the tidal influx of seawater and.

10:57 Speaker 2: The gradual accumulation of sediments. They are typically found in.

11:01 Speaker 4: Intertidal zones and river systems in temperate regions, although there.

11:05 Speaker 2: Are some examples of subtropical salt marshes.

11:09 Speaker 4: As with seagrass, most of the carbon in salt marsh ecosystems are found.

11:13 Speaker 2: Below ground in the roots and sediments.

11:16 Speaker 4: Their sequestration rate can range between two to six tons of carbon dioxide per hectare per year, and they are particularly effective at long-term carbon burial and trapping carbon-rich sediments transported from upstream and through tidal exchange. Although the focus on blue carbon emphasizes the importance of these ecosystems for their climate mitigation value, Mangroves, seagrass, and small marshes have many other functions and co-benefits. For example, they support high biodiversity. These ecosystems are rich nurseries for juvenile fauna of endangered and.

11:55 Speaker 2: Commercially valuable species.

11:58 Speaker 4: They also have high climate adaptation value in their protection of the coastline from erosion, floods, and storm surges. They support diverse livelihoods and income streams for coastal populations especially for local communities and indigenous people.

12:16 Speaker 2: And lastly, they also support.

12:17 Speaker 4: Ocean health through nutrient cycling, bioremediation, and maintenance of water quality. Despite their important role, coastal blue carbon ecosystems are at risk of loss and degradation, and this could result in carbon stored in these ecosystems being returned to the atmosphere.

12:36 Speaker 2: Common drivers of loss include.

12:39 Speaker 4: Land conversion to aquaculture and agricultural uses, unsustainable logging and harvesting, pollution and deterioration.

12:47 Speaker 2: Of water quality, as well as coastal development.

12:53 Speaker 4: There is urgent action needed to conserve, protect, and restore blue carbon ecosystems, and we need to develop effective policies that can support this. Some examples include international policy frameworks, such as the UNFCCC, and recognizing the value of blue carbon in countries' nationally determined contributions.

13:15 Speaker 2: The NDCs are a good.

13:16 Speaker 4: Mechanism because they allow countries to commit to restoring and conserving their blue carbon ecosystems based on best available information, with the intention of increasing ambition with each reporting cycle. This means that countries can initially commit by just acknowledging the existence of blue carbon ecosystems in the first cycle, and then raise their ambition to document, measure, and verify ecosystem.

13:40 Speaker 2: Extent at carbon stocks in the next cycle, and so on.

13:45 Speaker 4: There are also related mechanisms, such as the National GHG Framework, which is suited for countries where there is currently sufficient data on their blue carbon.

13:53 Speaker 2: Ecosystems to include them in the national inventory.

13:59 Speaker 4: Achieving climate and biodiversity goals requires innovative financing strategies for nature-based solutions that consider social, environmental, political.

14:09 Speaker 2: And economic contexts.

14:11 Speaker 4: Broadly, there are market and non-market mechanisms for financing blue carbon restoration, protection, and conservation. Market mechanisms include approaches in carbon markets, where the carbon sequestered in blue carbon ecosystems can generate carbon credits that can.

14:28 Speaker 2: Be sold and exchanged.

14:30 Speaker 4: Non-market mechanisms are more diverse and include examples like payments for ecosystem services and debt for nature swaps. Data continues to be a barrier in scaling up blue carbon projects, policy and finance, and remote sensing offers an important tool in building our understanding of blue carbon ecosystems. Applications of remote sensing in policy include understanding the aerial extent or an existence of blue carbon ecosystems, monitoring for national GHG inventories. In carbon projects, they can be used to identify project areas, setting baseline rates of ecosystem degradation, estimations of biomass, and monitoring.

15:15 Speaker 4: And in management, they can be used for monitoring and surveillance of illegal destruction or logging across a large scale. as well as remote detection of health and condition of these plants from space. And with that, thank you for your attention, and I'll now turn it over to the next speaker.

15:40 Speaker 5: So my name is Adia Bey, and I work for NASA Goddard Space Flight Center in Dr. Lola Fatayimbo's mangrove science team. It's a pleasure to join you today and to tell you more about our work using Earth observations to measure mangrove blue carbon ecosystems. In this first part of the RSET module on blue carbon ecosystems, we'll focus on mangroves. I'll explain the basic steps and key considerations for mapping the extent of mangrove ecosystems. Afterwards, Dr. Lola Fatayimbo, the head of the mangrove science team, will explain how to use Earth observation data to measure mangrove height, structure, and biomass.

16:21 Speaker 5: So, there's several learning objectives that we'd like to cover before the end of this presentation. We hope that you'll be familiar with existing global datasets that show mangrove extent, canopy height, and biomass. We hope that you'll understand how these datasets were produced, as well as some of their limitations. Our goal is that you'll have a basic criteria for assessing the suitability of existing data sets for your purposes, and that you'll know how to use Google Earth Engine to generate your own mangrove extent data.

16:58 Speaker 5: We also hope that you'll be able to estimate mangrove canopy height, biomass, and carbon stocks in your area of interest, and know what data is necessary for a more precise mangrove ecosystem carbon stock estimate. So before we dive into the module, I should mention that there are tons of resources available that can help you learn more. As this is an intermediate level module, we'll focus on remote sensing data and basic processing and analysis steps that are useful for assessing mangrove ecosystems for one point in time.

17:37 Speaker 5: If general introductory information about remote sensing would be useful or if you need more guidance on how to get started with Google Earth Engine, please see these three prerequisites. Conversely, if you're a more advanced user and you'd like more advanced guidance on using mangrove data to map loss, gain, and other ecosystem changes over time, you may want to see these additional resources that dive into depth on mangrove change mapping and using Google Earth Engine for other applications.

18:18 Speaker 5: Many of the resources that I'll show you in this first part have been developed by the mangrove science team. including several resources that are available on RSET's website that have been developed by Abigail Barenblit, Celio Resende de Souza, Mark Simard, and Lola Fatimbo. So today, I'll explain why mangrove-specific data is needed. We'll review key things to consider when mapping mangrove ecosystems. I'll provide an overview of existing mangrove datasets that can help you quickly generate metrics of mangrove extent and change, and we'll map mangrove extent at a sample site in Senegal's Seloum Delta.

19:07 Speaker 5: So mangrove ecosystems are extremely valuable locally and globally. Although mangrove forests are often compared to terrestrial forests, they differ in Several significant ways, including ecosystem services that they provide, the threats that they face, and their biophysical conditions. Mangrove forest metrics, models, and remote sensing techniques are not interchangeable with those of terrestrial forests. So if you're interested in assessing mangrove ecosystems, it's really important to use mangrove-specific data.

19:44 Speaker 5: Mangrove cycle nutrients and improve water quality by filtering nitrates, phosphates, heavy metals from water. They serve as important spawning grounds for nurseries. Over 4.1 million people depend upon mangrove forests for their food and livelihoods. And as much as 80% of global fishing catch can depend directly or indirectly on mangrove habitats. Mangroves also serve a critical role in combating climate change and also for supporting adaptation. Mangroves sequester vast amounts of carbon in soil and vegetation above and below ground.

20:20 Speaker 5: It's estimated that over six billion metric tons of carbon are stored in the first meter of soil in mangroves worldwide, and mangroves can sequester four times more carbon than most rainforests per hectare. Mangroves also provide a natural buffer against coastal erosion, storms, floods, and extreme weather events that may be caused by climate change.

20:48 Speaker 5: Despite their global importance, 35% of the world's mangrove forests may have been lost between 1980 and 2000. On average, mangroves are being lost at a rate of 2% per year. Although several global studies have found that the rate of mangrove loss has been declining since 2010, the recently released IUCN Red List of mangrove ecosystems have found that half of the world's mangrove areas are considered threatened. Most drivers of mangrove loss are directly anthropogenic, such as the expansion of aquaculture, commercial agriculture, and urbanization.

21:27 Speaker 5: However, mangrove loss due to sea level rise and increasing storm intensity is becoming more prevalent. So recognizing the importance of mangrove ecosystems for sustaining life on land and also recognizing that these ecosystems are under threat, several international agreements have been established that support the conservation and restoration of mangrove ecosystems. These include the United Nations Framework Convention on Climate Change, the Convention on Biological Diversity, the Sustainable Development Goals, and the Ramsar Convention on Wetlands.

22:07 Speaker 5: Nevertheless, only 14% of the world's mangroves benefit from the highest level of protection, and mangrove loss continues inside and outside of protected areas. Mapping and monitoring mangrove ecosystems is really essential to understanding where and how these valuable ecosystems are changing. Earth observation data can be used to map current and historic mangrove extent, the condition of mangroves, for example, whether or not they're intact or degraded, the height and structure of mangroves, above ground biomass and vegetation, and some biophysical conditions and drivers of mangrove change.

22:54 Speaker 5: When mapping mangroves and other ecosystems, it's important to consider exactly what you're mapping and why. Maybe you're interested in mapping the extent or current carbon stocks or habitat distribution, or maybe you want to use earth observation to identify hotspots of mangrove loss to inform conservation and restoration efforts. Remote sensing can inform a wide variety of mapping objectives. Identifying the objective that you're most interested in will have an impact on the type of methods that you use.

23:29 Speaker 5: For example, guidance from the Intergovernmental Panel on Climate Change should be followed if you're interested in mapping mangrove extent and change for national-level reporting. The Center for International Forestry Research provides protocols for gathering ground-based data through forest inventories that can be used to estimate mangrove carbon stocks. If you're mapping mangroves to develop and sell blue carbon credits, you'll want to select and follow an approved methodology that covers the specific activities you plan to implement.

24:00 Speaker 5: These methods may include recommendations for specific types of remote sensing analysis, as well as ground-based data that should be collected. In short, there are many ways to map mangrove ecosystems, but before you start, you want to be confident that the methods that you're using are likely to produce the quality of results that you need. Some key considerations when you're starting to map mangrove ecosystems. There's a wide variety of data that you can be using, and you'll want to keep in mind the spatial, resolution, spectral resolution, and temporal resolution of the data.

24:46 Speaker 5: Data from passive optical sensors like Landsat and Sentinel-2 can be useful for mapping mangroves, as well as active radar and lidar sensors, such as data from NASA's Shuttle Radar Topography Mission and the Global Ecosystem Dynamics Investigation. For national-level mapping covering multiple decades, 30-meter Landsat data may be a great fit for your project. If you're mapping mangroves at a subnational level or for a more recent time period, a finer resolution data set like Sentinel-2 might be preferable.

25:23 Speaker 5: If your study area is persistently cloudy, using radar or LIDAR data that penetrates clouds can help map mangroves in these challenging conditions. Another thing to consider when mapping mangrove ecosystems is the tier of level, the tier of analysis that's needed. The IPCC provides a framework for structuring different methods that are used. The most basic method, the most basic tier is tier one. This draws upon course resolution data and globally available estimates. On the far other end of the spectrum, you have Tier 3 methods, which use higher resolution data in conjunction with national or subnational level data on biomass and carbon stocks.

26:14 Speaker 5: This usually is drawn from ground-based forest inventories, and it's geared toward capturing the variability within ecosystems over space and over time. And this is really geared toward estimating mangroves, at a quality and confidence that's known with the highest level of precision possible.

26:41 Speaker 5: This module on mapping mangrove extent is really geared toward tier one level mapping. And if you review the IPCC guidelines that are provided in the wetland supplement, you can see some global default values for mangrove carbon stocks per hectare. However, Colleagues in this part of the module, as well as in the second part, will dive into more detail about how you can improve upon these global default values in mangroves, tidal marshes, and in seagrass environments. Nevertheless, if you're lacking ground-based data and you want to get started, using IPCC's Tier 1 default values for mangroves are a good place to start.

27:25 Speaker 5: But of course, you'll need to combine this data with information on mangrove extent. The Mangrove Science Lab has produced a large number of apps that analyze Earth observation data available through Google Earth Engine to help people quickly generate metrics on mangrove extent, height, biomass, and carbon stocks. Google Earth Engine's cloud computing resources facilitate remote sensing analysis of vast amounts of information And these apps can also be useful for disseminating results and sharing spatial data in a user-friendly format.

28:04 Speaker 5: This can be really advantageous if you're trying to make data accessible to non-technical audiences. However, there are some limitations to relying upon existing data. One key limitation is that the data and methods, the limitations of these, may not always be clear to end users. So this is why some people may choose to generate their own maps of mangrove extent. To help users better navigate the data that is available, the mangrove science team developed a dataset exploration app. This app shows national level estimates of mangrove extent from five different datasets, as well as estimates on above ground biomass and carbon stocks, which will be discussed a bit later.

28:49 Speaker 5: If we zoom into a specific country like Senegal, and if you click on the country using the app, shown in the slide before. You can see a wide range of values. For Senegal, these estimates of mangrove extent range from 16,500 hectares to over 140 hectares. It's a huge range, and part of this difference may be attributable to differences in methods, differences in types of earth observation data used and calibration data, as well as differences in the time period assessed. To help people understand where these differences may be impacting the spatial extent of mangroves identified, the mangrove science team developed a mangrove data selector app.

29:41 Speaker 5: With this app, users can select certain parameters like the date range, the temporal resolution, the priority metric, and the spatial resolution preferred to get a recommendation for a certain dataset that may be a good fit. For many users, Global mangrove watch data set and app will be very useful. Version 3 draws upon 25 meter SAR data to map mangrove extent and change for over two decades with an accuracy of above 86%. In 2024, version 4 will be released, which draws upon Sentinel-2 data, so finer spatial resolution data and it improves the accuracy of its mangrove extent mapping by nearly 10%.

30:34 Speaker 5: The app is really easy to use, and it's very easy to generate metrics for an area of interest. You can search for specific locations like countries or protected areas, or you can upload your own boundaries for an area you're interested in as a shapefile. Global Mangrove Watch app or platform online, it'll generate extent over time, it'll show you change over the past 20 years, and it will also show you a summary of alerts of mangrove loss. Global Mangrove Watch may be sufficient depending upon what your objective is, but for those of you who'd like to generate your own data, We provide a script.

31:22 Speaker 5: We prepared a script that runs in Google Earth Engine to guide you through this process. So many of the data sets that the team uses to map mangroves are already available within Google Earth Engine, including Landsat 8 and really the entire Landsat archive, Sentinel-2, Sentinel-1 radar data, PALSAR, SRTM, and to a certain extent, JEDI data. And all of this information can be accessed through the cloud and you can essentially use your algorithms to run your analyses on the pieces or the parts of this data that are within your area of interest.

32:06 Speaker 5: So understanding the basic components of Google Earth Engine will make it easier for you to use and adapt the script that we've prepared. But to provide a very brief sort of crash course, basically there are four panels. Google Earth Engine is comprised of four panels. When you click on the link provided in the next slide, it'll open a script in the middle. In this middle pane, you can edit the code, you can save a script, you can share a script, and you can run it. On the left-hand side, you can see all of the scripts that you've written and saved.

32:49 Speaker 5: You can also view the assets that you have uploaded, and Earth Engine assets are basically data sets that you have uploaded yourself, such as shape files, CSV files, and GeoTIFFs. You can also access documentation that can help you learn how to use Google Earth Engine. When you run a script, like the script that's shown on the next slide, layers that you're mapping will appear below. And you can also turn these layers on and off, change the transparency in the layer manager. You can also generate new polygons and new points.

33:32 Speaker 5: This is something that is useful if you're trying to generate some on-the-fly training data for running a classification algorithm or for validating it. And you can also print results. So some of your results will appear on the right-hand panel. So in that panel you have a console output and you have tasks. So if your script generates exports, they will appear there. Lastly, on the right-hand panel there's an inspector. So if you take your mouse and click anywhere within the map and you have some data loaded, it'll tell you the value of that layer at the point that you've selected.

34:15 Speaker 5: So for more information on Google Earth Engine to get started or to learn how to start writing your own scripts, feel free to visit some of these key resources listed on the left. So we're going to stay in Senegal and we're gonna visit the Seloum Delta. And if you click on the link in the middle, top in the middle, and you click run, you'll see five different layers will be added to your map down below, and 10 different outputs will be printed on the right in the console, and two tasks will be added to your task manager on the right.

34:58 Speaker 5: So I'm gonna walk you through these outputs, just showing you the highlights, and then we can dive into detail about what this script is doing and how. So basically, in the maps, in the map panel, you'll see the results of a mangrove extent, well, supervised classification and the mangrove extent pulled from that. This classification is drawing upon Landsat imagery for the year 2020, as well as the year before and after, so plus or minus one year. You'll see the relative importance of each of the input variables.

35:40 Speaker 5: So in addition to using bands from Landsat to train the algorithm, we've also generated some additional features, such as different vegetation indices, which we'll go through. We have tested, we've generated some training data and testing data quickly within Google Earth Engine to validate the classification algorithm. But to do a more robust accuracy assessment, we have also generated stratified sampling points that we can assess either on the ground or perhaps with very high resolution satellite imagery in Google Earth Pro.

36:18 Speaker 5: And lastly, we are also loading to the map the results of Global Mangrove Watch version three for our area of interest so that we can compare our supervised classification with the results of Global Mangrove.

36:36 Speaker 2: Watch.

36:37 Speaker 5: I'm going to revisit the script in a minute, but I just want to walk you through some of the basic parts of doing a supervised classification. Basically, you need some input data. This is one of the aspects of the script that you would have to change in order to make this work in your area of interest. We generate polygons and from those polygons we extract a random selection of points. We also are using in the supervised classification a composite. This is a multi-temporal composite drawing upon three years of Landsat imagery.

37:16 Speaker 5: And we use our training data at each point where we have samples, we extract the value from our composite and from our different features, and we use that information to train a random forest classifier.

37:35 Speaker 5: So this is essentially on the far right at part E. We run that classifier on our input image, which is the Landsat image from 2020, the composite, and we use a subset of the data. a set of testing data points to validate the classifier, just to make sure that the results that we're getting are, just to make sure that the classification algorithm is functioning properly. And as a result of running that classification algorithm on the imagery with our training data, we'll have a classified image. This script is very basic.

38:20 Speaker 5: It just has mangrove areas and non-mangrove areas. You might wanna make this more complex by adding additional land cover types, but to help you get started, we've just built a basic one for now. And to assess the accuracy of that classified image in a more robust way, we've generated stratified random samples. So meaning we have a random selection of samples in mangrove areas and a random selection in non-mangrove areas. And you can use that for your next step either in the ground or assessing those sites with high resolution imagery.

38:58 Speaker 5: So how does that translate in the script? We have a couple of steps. First, we define our area of interest, which is Seloum Delta in Senegal, and we set up the map. We're not going to, I can show you how the script looks in more detail, but just to walk you through the basic parts. We set up the map. We choose our start and end dates for the compositing period. We select the imagery. We can load that directly through Earth Engine by loading it from the catalog. We add some additional features such as vegetation indices and elevation data.

39:35 Speaker 5: We create polygons of training and testing data. We use this to run a random forest classification algorithm in Part E. and we load Global Mangrove Watch data to compare the results, generate the sampling to assess the accuracy. So, just going to show you what that looks like in the script, and also tell you a bit more about different parts of this script that you would need to change if you wanted to adapt this to run in your area.

40:11 Speaker 5: So there are different ways of defining your area of interest. You can load a polygon, you can draw a polygon in Earth Engine. You can also find different publicly available data sets through the catalog. Here we've loaded protected areas from the World Database of Protected Areas, and from that we filtered it down to Saloom Delta. We're defining our center, the center of our map, the boundaries of the project area. In part B, we have the years of the reference period that we're using to create our multi-temporal composite.

40:54 Speaker 5: This helps us generate a cloud-free input image for our supervised classification algorithm. In part C, we prepare our input imagery by running a cloud masking function. And then we also prepare some additional features such as normalized difference vegetation index, normalized difference mangrove index, a simple ratio, green chlorophyll vegetation index. These are essentially calculations that use the bands and Landsat to run simple calculations that make it easier for vegetation health to pop out and for mangrove features to appear more distinctive than other features in the landscape.

41:41 Speaker 5: Because the cloud masking isn't always 100% effective, we are also preparing, we do a median composite to pull the median value and to have a more robust cloud-free image. We add elevation data. Mangroves tend to appear at lower elevations, so this will be a useful feature that can help with running the classification algorithm. And we define our classes, mangrove and non-mangrove classes. And to generate some simple training data, we've drawn some polygons down here in the mapping panel. This is a quick and easy way to get started, but one challenge with generating training and testing data this way is that you don't always know what the time period is of this high-res base map data.

42:38 Speaker 5: So it's good to generate data perhaps in Google Earth Pro or generate data in the field, but this is just a quick and easy way to get started. We have a part of the script that tells us how well each of those features are performing in terms of generating the classification algorithm, to what extent they're important. We can see those results here in the console. We can also review the size of our training and testing data, the extent to which they fall within non-mangrove areas, which is zero, and mangrove areas, which is one.

43:18 Speaker 5: And at the end of all of this, we also have an estimation of mangrove extent. The first value is from our supervised classification, that's around 503 kilometers square. And the second value is the estimate from Global Mangrove Watch, which is around 621 kilometers square. And in the layers panel, you can turn these on and off to view the results. It takes a minute to load. In pink, you'll see mangrove extent. And in yellow, you'll see Global Mangrove Watch data. I'll give this a second to load.

44:06 Speaker 5: And then we will transition to the next part of our.

44:13 Speaker 2: Module.

44:14 Speaker 5: In the next part, Dr. Lola Fatayimbo will tell you more about how to take this mangrove extent data and go deeper into detail to help you estimate mangrove height, canopy height, understand mangrove structure, and use that information to estimate mangrove carbon stocks. So basically in pink, you have the results of our classification algorithm, and in yellow, you have the results of Global Mangrove Watch. There are a lot of similarities, but if you zoom in, you can see where those two results differ.

44:54 Speaker 3: So in this next section, we will be focusing on how we use Earth observing data to map mangrove height, structure, and biomass with the ultimate goal of being able to estimate carbon stocks. So as we mentioned before, there are different tiers by which you would report your mangrove ecosystem carbon stocks. And there are several factors that influence how you might calculate your Tier 2 estimates of mangrove carbon stocks. And these are the same factors that influence the above the biomass and the soil carbon in mangroves.

45:37 Speaker 3: So when we are interested in understanding and estimating above-ground biomass, in particular in mangroves, we have a range of factors and indicators and measurements that we can use to estimate the above-ground biomass. so some of these factors are the height of the vegetation or the structure essentially this we mean you know how tall is the canopy of a tree how tall are the trees what type of tree is it the condition of the forest is it densely vegetated or is it sparse We're also looking at tree species diversity and composition, so knowing what the main species are that you will find on the ground is very important because that affects your allometric models that you will use to go from your field measurement or your remote sensing based measurement to above ground biomass.

46:36 Speaker 3: And some other important measurements that you might need are measurements of tree density, base area, knowing the solidity, and knowing the age of the forest patch that you're looking at. But of course, we don't always have access to all this data. This would be very detailed measurement that you might get from measurements in the field when you're doing in-situ measurements of forest plots. When we're comparing or trying to estimate above ground biomass and carbon stocks using remote sensing data or using Earth observation data, we don't have access to these really detailed measurements that we could use to map and model and estimate above ground biomass and carbon stone mangroves.

47:25 Speaker 3: And so here on the right of these models, it really gives a nice overview of what the structure of a mangrove forest is and why it might be challenging or different in a mangrove to estimate above ground biomass from a more typical forested ecosystem. So as you see here, mangroves are really characterized by these complex above-ground root systems, and those systems also allow you to know which species you are primarily working with. But they also grow in tidally inundated areas, so you might have a presence of water that can change essentially the height of the trees depending on whether it is high tide or low tide.

48:11 Speaker 3: But when we're looking at estimating above ground biomass and canopy structure using space-borne or remotely sensed data, we essentially ignore a lot of these detailed measurements and really look primarily at the canopy height. Because if you know the height of a tree, that is a very good indicator of how much biomass is actually stored in that tree. Generally speaking, when we're working in the field, we would work with a measurement called the DBH, or diameter at breast height, which is essentially the diameter of the tree, which is an even better indicator of what the biomass might be of the tree.

48:53 Speaker 3: However, we cannot directly measure the diameter of individual trees using space-borne data. We can measure the height of the trees using several different space-borne data sets. And that's what I'll be talking about in the next few slides. So, as mentioned before, there are a range of different Mangrove datasets. Some of them have

49:21 Speaker 3: datasets use very similar methods, and then some of the datasets use different methods that are similar but with different results. And so, one very important thing to take into account when looking at this whole range of datasets that might exist is really which dataset works best in your area of interest. So linking again to the mangrove data comparison tool that was mentioned earlier, you can see here that we also have in there different different data sets of ground biomass of soil carbon that can be used to estimate a range of above ground biomass measurements or get a range of soil carbon measurements.

50:01 Speaker 3: And they're also very useful to compare to with in situ or field based measurements of both above ground biomass or soil carbon. So when you're clicking on this app, you can also look not only on the extent data sets, but also at above-ground biomass and soil carbon. And in the next few slides, I will also describe an above-ground biomass data set that you can use to look at above-ground carbon and above-ground biomass measurements. So here I'd like to provide an overview of some of the existing mangrove data sets that are available for measurements of above-ground biomass.

50:41 Speaker 3: One thing to note is that here we are really only showing two different data sets, one that is produced by our lab and another one that was published about 10 years ago. Since the publication of both of these data sets and both of these papers, there have been several other regional and global estimates of above-ground biomass, of above-ground carbon, and of soil carbon that have been published with various ranges of accuracy, of resolution, and covering different timeframes. So it is very important when looking at these datasets to make sure that the methodology used is really appropriate for the objective of the use of that dataset.

51:27 Speaker 3: One other thing to note is that oftentimes these global data sets are calibrated on a very limited amount of field measurements of above-ground biomass, soil carbon, or total carbon. So, another really important consideration when looking at these data is to make sure that what you're seeing on the ground or what measurement or site that you're selecting is actually well represented in this data set. So, for example, when looking at the Hutchinson et al. map, you'll see that there isn't any data available from the Atlantic coast of Africa.

52:06 Speaker 3: Similarly, when you look at the Samart et al. map, you see that there is a wider distribution of sites from which we were able to get in-situ data to calibrate the map. However, it is still limited, especially looking at Asia, for example. Every single map has its qualities as well as some of the limitations, but generally speaking, provide a good overview of what the trends are and what the total above-ground biomass and carbon stocks are for the timeframe selected. Looking at the mangrove data comparison tool, we can also see the wide range that might be present when looking at these different maps.

52:51 Speaker 3: Again, focusing on Senegal here, we'll see, for example, that when we look at this Hutchinson map from 2014, the mean above-ground biomass density for Senegal was found to be 142 tons per hectare, while using the Samard map, the mean above-ground biomass density was only 28. megagrams per hectare. So this is a very large range, and this might be influenced by the type of data that was used as an input. So essentially looking at what data sets were used to create this model. It could be due to the resolution, it could be due to the timing of the measurement, and it also could be due to the type of measurement that was used.

53:38 Speaker 3: In our case, we were working with global space-borne SAR data, as well as global space-borne LIDAR data that allowed us to estimate the canopy height, biomass, and carbon stocks across a wide range of sites. But there were some limitations in areas of very short mangroves, which may lead to a skewing of the actual measurement to a lower. to a lower biomass. Similarly, in other datasets, you might find that they underestimate how many short trees there are, leading to a skewed or a biased estimate of the total biomass or the biomass density towards the higher numbers.

54:35 Speaker 2: Following up and.

54:36 Speaker 3: Going now into the soil organic carbon data sets, we're now providing an overview of existing mangrove data sets specifically that are focused on the measurement of soil organic carbon. Similarly, there have been now several global and regional data sets of soil organic carbon that are available for download and for analysis. Specifically, the data set produced by Sanderman et al. in 2018, which provides 30-meter resolution estimates of the first meter of soil carbon for all mangrove areas worldwide.

55:12 Speaker 3: It is based on the Global Mangrove Watch extent. and uses a wide distribution of field measurements of soil carbon to model what the global estimates of soil carbon are. Other data sets that were produced or that can be used are mentioned here as well. They were published by Atwood et al, Rovi, or Jardine and Sigamaki. But as mentioned before, some of these older data sets may have been based on a more limited amount of in situ measurements to train their models and therefore keeping an eye out for more updated data, more updated versions of the data is always really important because there might be a better estimate of soil carbon or above ground biomass available.

56:06 Speaker 3: One other thing to mention is that it's very important when looking at soil carbon to note the depth of the measurement. So in the Sanderman et al. paper, for example, all of the measurements are... published are for the first meter of depth of soil. Now in some areas you might have peat soils or mangrove organic soils that are really, really deep and that go beyond the first meter. So to this date there is not a data set that allows us to know what the depth of the soil is, which is also a very important measurement.

56:43 Speaker 3: We also don't have many data sets that allow us to go past one meter of soil depth to know what the soil carbon is. Finally, I'd like to provide you with a quick overview of the first global map of mangrove canopy height and mangrove biomass measured at 30 meter resolution. This is a data set that was produced by our team back in 2019 and is heavily based on data from the Shuttle Radar Topography Mission, which is mission that allowed us to have the first global digital elevation model and also allowed us to then get estimates of mangroves canopy height measurements on a global scale.

57:35 Speaker 3: When you look at the map and at figure one here in the top inset number A, you'll see there are Several numbers on those maps highlighting the areas where there were field measurements taken to estimate the above-ground biomass canopy height and also soil carbon in these mangroves. These are all the sites that were used to train the models to estimate above-ground biomass. These are also the sites that were used to validate the estimates of canopy height in these areas. And one of the reasons we're so interested in better understanding what the distribution of canopy height is and what the range of canopy heights is in mangroves is because this really allows us to get a more constrained measurement of the environmental conditions that mangroves grow in, but also of the total carbon estimates in terms of both above ground biomass and below ground soil carbon.

58:37 Speaker 3: So one of the big discoveries that came out of this work was to actually be able to directly measure what the maximum height is of a mangrove tree. And what we found when looking at this map is that there are patches on the planet, especially looking here at Gabon, which I'm showing in the inset on the right, where there are mangrove trees that are able to grow up to 65 meters in height. So there are patches of mangroves where almost every tree is over 60 meters and individual trees that grow up to 65 meters in height.

59:14 Speaker 3: These are patches that would not have been found most likely if we did not have access to this amazing data set that gives us these measurements of topography and can be used to estimate canopy height. Finally, the SRTM data set is now quite old. SRTM was flown back in 2001, so that data set is now over 20 years old, and it's really important to be able to update these data as possible. So, we've recently worked on a new updated version of the global canopy height model. This data is now available for download and for use, including in Google Earth Engine, and that is based on a dataset called TandemX, which stands for Terasar X add-on for digital elevation measurements.

1:00:15 Speaker 3: The beauty of this dataset is that it is more recent, and it also is at much higher resolution than the SRTM dataset that is now over 20 years old. So the TANM-X model is actually available at 12 meters spatial resolution, and the vertical accuracy is much better than that of SRTM as well. And so we can now look at almost canopy level or at least patch level variation in canopy height in mangroves. This also allows us to do a better job at deniating areas that are dense mangrove versus areas that are maybe not as dense mangrove because of the higher resolution and getting, hopefully, much improved estimates of above-ground biomass and carbon stocks in mangrove forests worldwide.

1:01:12 Speaker 3: Now, finally, all this information on canopy height is, of course, very useful to be able to generate your own map of mangrove height and biomass. As I mentioned earlier, one of the really important factors in estimating the above-ground biomass and soil carbon is really the availability of in-situ datasets. So here we have provided a script that can be used to generate a localized estimate of above-ground biomass. It will allow you to estimate above-ground biomass on existing mangrove canopy height datasets.

1:01:54 Speaker 3: To do that, we use a generic allometric equation that goes from the height measurement directly to above-ground biomass. This is using a generic global above ground biomass allometry that was developed using those 330 bot data sets that I showed in the previous slides. And there are two different models that can be used to estimate above-ground biomass, either using a metric that is called basal area weighted height or using a metric called maximum canopy height. But generally speaking, this is a script that allows you to go from canopy height to above-ground biomass, and then from above-ground biomass to carbon essentially is a division by So generally speaking, when we estimate above ground biomass in a forest, the assumption is that about 50% of that biomass or that value of biomass is the carbon.

1:02:59 Speaker 3: So to move from above ground biomass to carbon stocks, you would essentially multiply by 0.5 or in the case of mangroves, we use the value of 0.49 as was recommended by publications.

1:03:12 Speaker 3: Thank you. And next, we'll go to the summary part of the slides.

1:03:21 Speaker 1: Thank you very much, Lola. So, let's take a minute to review the key points from Part 1 of our training. We hope that you became familiar with global datasets that show mangrove extent, canopy height, and biomass, as well as understand how these datasets were produced, as well as some of their limitations. We introduced basic criteria for assessing the suitability of existing datasets for your purposes and how to use Google Earth Engine to generate your own mangrove extent data. We also explored how to estimate mangrove canopy height, biomass, and carbon stocks in your area of interest and know what data is necessary for more precise mangrove ecosystem carbon stock estimates.

1:04:04 Speaker 1: In part two on December 5th, we'll see demonstrations of mapping salt marsh and seagrass with Earth observations. Specifically, demonstrations on how to map the extent of salt marsh and seagrass ecosystems using satellite observations, calculate the carbon stocks of mapped salt marsh and seagrass ecosystems, and explore synthesis methods to estimate blue carbon across ecosystems. Quick reminder, This training series will have one homework assignment. You can access the homework from the training web page starting December 5th, and the answers should be submitted via Google Forms by December 19th.

1:04:42 Speaker 1: Certificates of completion will be issued to participants who attended both live trainings and submit the homework before that deadline. Certificates will be emailed about two months after the training. This slide shows the contact information for myself and today's speakers, as well as links to the RSET website, YouTube channel, and Twitter, or X, and our sister programs, Develop and Survear. Thank you very much. We'll now be moving over to the question and answer session.

1:05:13 Speaker 6: Hi, everyone.

1:05:27 Speaker 5: We are still in the process of answering your great questions, but I'll start with one that I've already answered. It's question number five. Mangrove species have varying capacity to store carbon. Therefore, a finer classification into mangrove species may be important to have a more accurate estimation of mangrove ecosystem biomass and carbon storage capacity. Can Google Earth Engine be used for mapping mangrove extent? to provide a finer classification of different mangrove species? The short answer is yes.

1:06:05 Speaker 5: But the longer answer is it really depends on the training data that you have and the type of satellite imagery that you use. So Google Earth Engine, there's a difference between the software or the remote sensing software that you're using or tool and the input data that you're using. and so really the limitation is about the data. Stratifying based on species is a very complex way to stratify mangrove areas into different strata that may have different carbon stocks. An easier way to begin to do that would be to stratify based on vegetation health.

1:06:49 Speaker 5: You could use potentially vegetation indices, like normalized difference vegetation index, or there's normalized difference mangrove index, which I think both of those are provided within the script, the first part, the first script. A second way to stratify mangroves into different groups, which will have different above-ground biomass values and carbon stocks, could be to stratify based on height. And so you could use some of the mangrove canopy height and structure techniques that Lola Fatimbo just presented to stratify mangroves in this way.

1:07:33 Speaker 5: This can be very useful because mangrove height depends on a variety of things. But for example, you'll find that in areas with more fresh water, mangroves will grow taller. and mangroves where the salinity is lower will be shorter. So, for example, like with Rhizophora mangroves, you're more likely to find these taller mangroves in areas that are less saline, and they will have more carbon stock, more above-ground biomass values than, for example, shorter mangroves like Avicennia, which tend to be sparse.

1:08:10 Speaker 5: They will also have a higher soil-adjusted vegetation index, and lower carbon stocks. So those are just a few approaches, but yeah, your ability to apply these is not limited by Google Earth Engine. It's more limited by the data that you have. And even though some of the types of data you might wish to use for the more complex methods like hyperspectral imagery, even though that may not necessarily be publicly available in Google Earth Engine, you can always download that sort of data from other platforms and import it into your own workspace in Google Earth Engine as a GeoTIFF, as an asset.

1:08:52 Speaker 5: Okay, let's see. How do you define training data? How did I define training data for the module model? Do you acquire them from the field? And what about areas that were difficult to reach? So for the training model, in the first part, and potentially in the second part of part one as well, We just generated polygons on the fly in Google Earth Engine, and this was in mangrove areas, which are relatively easy to identify with the base imagery that's in Google Earth Engine. However, that is not recommended for a more robust study.

1:09:41 Speaker 5: because the image acquisition date of the base map imagery within Google Earth Engine, there's no way to identify it. Well, I shouldn't say that there's no way. It's not readily identifiable. You'd have to go to Google Earth and then find that exact image, and then you would know the image acquisition date. This is important because when you generate your training data, whether it's points or polygons, you want to be really sure that mangroves are there or that some other forest type is actually there.

1:10:15 Speaker 5: And because land cover can change at any point in time due to natural causes or anthropogenic disturbances, it's good to be sure to match the date of your training data to, you know, some sort of reality that you know to be true in the field or from very high spatial resolution satellite data. So, gosh, sorry, that's sort of a roundabout answer, but we did not acquire our training data in the field. We used a very high resolution satellite imagery that is in the base map of Google Earth Engine, but a better way to do that would be to use the very high spatial resolution imagery that's within Google Earth Pro, where you know the image acquisition date and can essentially train your algorithm to classify an image with the date that matches your training data, or if you have a lot of time and resources, you can also go to the field and collect the training data in situ, which is really ideal.

1:11:19 Speaker 5: Of course, the challenge is that, as you mentioned in your question, some areas will be difficult to reach, and it may be challenging to acquire a sufficient number of samples in order to train an algorithm well. Can tools and data sets used for mangroves such as Global Mangrove Watch be adapted to evaluate ecosystem services like carbon sequestration and biodiversity in support of non-mangrove wetlands? That's a really good question. We may have to come back to this. I'm not sure.

1:11:54 Speaker 6: Okay.

1:12:02 Speaker 5: I know that the IPCC wetlands supplement that was linked in the first part of part one, that includes coastal wetlands as well as inland wetlands. So that will have some good methodological guidance for assessing flooded forests, inland flooded forests. But my understanding is that Global Mangrove Watch is really focused on mangroves. This is kind of, sometimes the techniques that are used to map mangroves, people may put a filter on it and limit the scope of image, the geographic scope of imagery that's assessed to be restricted to within a certain number of kilometers from the coast.

1:12:52 Speaker 5: And so that's really not ideal if you want to assess inland forests, like for example in the Congo Basin or in the Amazon Basin. where there are large swaths of flooded forests that are nowhere near coasts. Yeah, it's a good question. It makes me wonder if the Global Mangrove Watch team will ever expand into that, but something that we'll revisit later. Okay, so question six. Considering the importance of diameter at breast height, DBH, of mangrove species and improving the accuracy of above-ground biomass calculations, are there databases that provide estimates of DBH of various mangrove species at various heights or stages of growth?

1:13:53 Speaker 5: The allometric models that help us relate mangrove measurements like DBH and height, etc., to estimate biomass or carbon are active areas of research. These relationships are often species and region specific. Although there's not one single database, there are some reviews on mangrove elometry that may be a good place to start. We will, I think we may have one or two links that we could add to that in the future. We will circle back to that and see if we can add a bit more.

1:14:28 Speaker 3: Hi, everyone. This is Lola. I can also jump in and answer this question. I think what this question is also asking is whether how we can measure dbH or estimate dbH based on height. And so just to follow up on what Adia said, which was correct, is that for that we would use an allometric equation or an allometric model that gets you from a height measurement to a diameter measurement. But actually we use those, that's essentially what we do when we use height as an input to measure biomass is that we use the relationship between height and diameter to estimate what the volume of the tree is and then that tells us what the biomass of the tree is as well.

1:15:20 Speaker 3: So I hope that also gave some more clarification.

1:15:32 Speaker 5: So I guess going to question eight, how can I identify restored mangrove sites by using SAR data? If your restored site has undergone structural changes detectable by SAR, you can begin with a simple two time point change analysis using one image prior to the restoration activity and one image post restoration. Another potential approach could be to use multiple images prior to restoration and sort of assess trends over time, looking at essentially building an average over multiple years and then looking for deviations from that established trend to help you see the differences between pre and post restoration.

1:16:32 Speaker 3: And I can answer question nine, which is a question about whether the generic allometric equation that uses basal area weighted height and SITM data is recommended for time series analysis. The SRTM data became available in the year 2000, but typographic conditions may change over time. And is SRTM still reliable for estimating above-ground biomass in that case? That's a very, very correct question, because the SRTM data is now outdated. It's quite old. And there have been many changes that have gone, that mangroves have undergone since then.

1:17:11 Speaker 3: So in that case, there are now newer digital elevation models, and there are newer models of global canopy height that can be used instead of SRTM. And here SRTM is really used as an example of what can be done and how to do this analysis. Some examples include JEDI data, which gives us actual canopy height, and other DEMs that you could use are the TandemX DEM or the Copernicus Glow 30 DEM. However, one thing to note is in every one of those cases, you do have to develop the relationship between mangrove canopy height and the height measurement that you get from the digital elevation model.

1:17:54 Speaker 3: It's not necessarily guaranteed that the equations we use to get from the SRTM height to mangrove height will be the same. So really you have to you have to make sure that you have data set specific models so that you are able to estimate your canopy height correctly and for the year that you are interested in. Okay, I can also try to answer question 10, which is asking about Sentinel-2 for subnational levels and how to to approach at subnational level where other data is required to support granular analysis, for example, gender roles.

1:19:00 Speaker 3: but still developing it in an African context. So I'm assuming that what this question is asking is, how do we use the Earth observation data such as Sentinel-2 at level that is more– that's at a higher resolution than, you know, what we've presented here, so not at a national level? In that case, I think you would use the data set that's at a resolution that is more appropriate. So it's correct here to say that Sentinel-2, for example, might give you a higher resolution result. And then I think the one thing that you can do for this particular analysis is just to do a site-specific analysis of what it is that you're looking at.

1:19:46 Speaker 3: And there will always be some limitations to what you can do with the remotely sensed data. So it really depends on what your application is and what your questions are and what your other data sets are that you're using in conjunction with the mangrove maps.

1:20:15 Speaker 5: Just a quick note to mention that we answered question 12 in our response to question 5. So essentially, you can stratify based on height. And this is an approach that's often used as tall, dense mangroves will tend to have a more above-ground biomass and higher carbon stock than shorter dwarf mangroves.

1:20:58 Speaker 3: Okay, and then I'll answer question 11, which is asking about species. Species-wide or species-specific mangrove standard spectral signature data available for identifying the different species of mangroves in the study area? If yes, then from which platform can we download? I don't think that that is available on a global scale, but it might be available. There might be publications that have done this on a site-specific scale, and otherwise this is something that you would do when analyzing your own data online.

1:21:28 Speaker 3: and you know exactly which species are located where, you could then look at the spectral signature and try to use that to predict species distribution or species presence and absence. But in the case of mangroves, one thing that you have to take into account is the fact that they are tidally inundated. So you would probably get differences in your spectral signature depending on the tidal levels, which is a very important thing to take into account.

1:22:47 Speaker 3: So there's a question about CO2 sequestration, and Adia earlier on had a really good answer to that, looking at the different sequestration rates in mangroves and other blue carbon ecosystems.

1:23:06 Speaker 5: Yes, sorry. I'm just now seeing question 13. Just give me one second.

1:23:14 Speaker 6: Okay, sorry.

1:23:15 Speaker 3: So I see.

1:23:16 Speaker 5: This question is somewhat different from the question earlier about, yeah, different CO2 sequestration rates within the blue carbon ecosystems. But

1:23:30 Speaker 5: how do you get it from mangrove mapping? I mean, this was partially covered through your presentation. We have around 30% to 40% contained within above-ground biomass and vegetation, especially in woody vegetation and in leaves and 60% within soil. And I believe that also includes CO2 within below ground vegetation. When, I mean, also what you covered in your presentation on mapping mangrove canopy height and structure, this will give us a lot of good information to help us better better understand, better estimate above ground biomass and the carbon that it contains.

1:24:21 Speaker 5: So this in conjunction with extent and strata, looking at different types of mangroves, tall, short, dense, sparse, all of this can be used to better calculate carbon stocks in mangroves.

1:24:38 Speaker 5: Do you want to add more?

1:24:40 Speaker 3: I think the one thing that I would add is that measuring CO2 sequestration using the data that we presented here directly is really challenging. What can be done and what people are working on is using the maps and field-based estimates of CO2 sequestration to build models that relate

1:25:04 Speaker 3: sequestration rates to the signal of the satellite data that you're using.

1:25:15 Speaker 5: Yeah, just to add on to that, I think CT's presentation in the introduction also helped explain this. There's part of the mangroves that you can actually detect with Earth observation data. This is what's going to be above ground. And everything that's below ground You can use equations like root-to-shoot ratios, and you can use modeling to understand how carbon stocks in soil may be distributed and may vary across a landscape. But at the end of the day, we're not directly detecting that with Earth observation data.

1:25:53 Speaker 5: We're using Earth observation data and in-situ data to begin to attack that sort of below-ground carbon stocks through proxies, essentially.

1:26:13 Speaker 3: Okay, question 14 is, how can we consolidate the difference in satellite data with ground proof data if there's a significant difference between the two and the difference is not consistent So this is where doing an accuracy assessment is really important. When comparing two models, there will always be some uncertainty and some error. You're not necessarily measuring the same thing in the field that you're measuring from space, or at the same resolution, or at the same time. So you will have some differences in your measurements.

1:26:48 Speaker 3: But this is where doing an accuracy assessment gives you a measurement of the error or gives you a measurement also of uncertainty. So you know, for example, that your map is 90% accurate. So you can say we have a 10% error on that. And almost all measurements have some error in them. Even when we measure a height of a tree in the field, we assume that there is about a 10% error on that height measurement. Because if you compare the height that's measured between two different people, it might not necessarily be exactly the same, even if they're both standing in front of the same tree using the same instrument to take the measurement.

1:27:29 Speaker 3: So there might always be some error or some difference between two measurements. And the important thing is to get the uncertainty, understand how to do the modeling or how to calculate the uncertainty or the accuracy assessment, and then take that into account when doing your analysis or your reporting.

1:28:03 Speaker 2: Okay, well, thank you, everyone.

1:28:04 Speaker 3: It seems like there are no more questions coming.

1:28:08 Speaker 2: Also, just so you know, there'll be.

1:28:09 Speaker 3: A transcript of the questions and we'll put that online so that you can have these for your reference. And if there are any more questions that come in, we'll also try to answer them and include them in there. Thank you.

1:28:24 Speaker 5: Thanks, everyone.