 9 a.m. sharp. So we should begin for 9 a.m. sharp in Costa Rica that is thank you so much everyone for being present at this web seminar, which is natural wealth of Costa Rica, mapping and modeling of biodiversity and its contributions to society using Earth observation. Since 2018, the Ministry of Environment and Energy in Costa Rica with the natural captain project have been working on a project using new Earth observation tools to reform analyses on biodiversity indicators and modeling to quantify and model ecosystem services. This project is a partnership and it's based at Woods Institute Stanford University, the United States, and at the me nine Costa Rica. We've been working with the Center back off Costa Rica the National Information Center in the Prius lab at the high technology lab in the framework of environmental information framework and monitoring of land service and use of ecosystem. This initiative is also received support from NASA and a joint program between the observation Earth Observatory view and this is the second workshop that we hold here in Costa Rica in February this project began so we're continuing it and seeing what results we have. And preliminary results because we are working still. So it's important again I repeat to go to the right interpretation channel to be able to hear the interpretation and hear the activity. Thank you. Another another caveat. Please go to mentee.com mentee.com. And the code is 46 58 417. That's the code. So you can tell us what you think about this conference and that way you can participate in the polls. Or in the in the activities. Right. So today, these are our speakers, Becky Kevlin creamer. She's going to give us context about the project. Then we have a file Schmidt will speak about mapping of ecosystem services and Alejandra Chevarria and we will be hearing about the project in Costa Rica with Gilberto Mara and how it's positioned internationally and he's from geo. And I am Rafael Monge director of the geo environmental services at the Ministry of Environment and Energy of Costa Rica. So without further ado, Miss Becky Chaplin creamer will take the floor from Stanford. Hello. Okay, I'm just going to start off. Actually, Jeff, why don't we switch because I think I made some updates to my slide so I'll just share my screen really quick. And before we begin, I will just recommend that if you so that we can make sure to save your questions for the Q&A panel at the end. If you have questions related to the topics, please put them in the Q&A. If you have any logistical needs, go ahead and put them in the chat. But if you put your questions in the chat, we might miss them for the panel and we do want to make sure to get all your good questions into our panel discussion. So thank you today. We're very excited to be here. Our whole team to speak with you about Earth observations and ecosystem service modeling. And before I get into the work I first wanted to introduce you to our group. The natural capital project is based at Stanford University is pioneering science technology and partnerships that enable people in nature to thrive. Through our cutting edge science, we map, measure and value nature's benefits to people in order to provide actionable information to decision makers. Our diverse portfolio of projects demonstrates where and how natural capital science is encouraging long term strategic investments in nature. From direct engagements in over 60 countries to support decisions related to sustainable development planning, securing fresh water, fostering sustainable livable cities and resilient coastal communities, and creating standards for the private sector. We've developed and refined our flagship tool, Invest, with uptake in an additional 125 countries and application now at the global scale. Invest, the integrated valuation of ecosystem services and trade offs is a suite of free open source software models used to map and value the goods and services from nature that sustain and fulfill human life. Enabling exploration of how changes in ecosystems lead to changes in the flows of many different benefits to people. And for example, this information can provide the basis for making sound investments to safeguard nature and its benefits to people in things like payments for ecosystem services, which now total more than 440 billion US dollars annually in more than 550 programs globally. For example, over 50 major cities in Latin America have launched payments to upstream landholders to secure water supply. And China is paying 200 million people to restore ecosystems across 50% of the country, both guided by integrated ecosystem service modeling. And multilateral development banks, the UN system of environmental economic accounts and the UN sustainable development goals aim to scale and systematize these efforts. Most recently, we've used this type of information in an international collaboration to feed into the convention on biodiversity's next set of conservation targets, which will be negotiated next year. To identify critical natural assets, the most important areas on land and in the ocean for providing an array of benefits to people. In this case, we optimized over 12 ecosystem services. However, we still face major limitations that prevent broader uptake of this information. Ease of use is maybe the biggest of these. It's difficult and time consuming to parameterize models. Relevance is another. So proxies that are often used in ecosystem service modeling like land cover can mask important functions. And such a reductionist way of seeing the world obviously oversimplifies complex relationships between the diversity and productivity and condition of ecosystems and their benefits to people. And this often means we might be getting it wrong in our modeling. So accuracy is also a concern. It means we might be missing important heterogeneity or prioritizing in the wrong places. So we think that Earth observations can address all three of these limitations simultaneously, providing public, free, easily replicable and globally available data and approaches with values clearly and convincingly attributed to ecosystems and resulting in information with lower uncertainty or better validation. Addressing those limitations to improve ecosystem service modeling through Earth observations is exactly the aim of this NASA and geo funded project that we're presenting on today. And our collaboration as well with Manai, Prius and the Banco Central. This modeling we're undertaking in Costa Rica can feed into a broad array of policy opportunities at the national scale. From the creation of a national system of environmental information to support the Costa Rica's national payments for ecosystem services program and monitoring for the CBD and SDGs. Through the strategic use of this information and efforts with UNDP and UNCA as well, for example, Costa Rica can serve as an example for other countries to follow. So Rafa will say more about many of these opportunities after we share some of the science advances before we move into the panel. And now before I turn it over to the team to get into the specifics on each service, I just wanted to show this overarching figure of how we see Earth observations factoring into each of these models for pollination, tourism, ecotourism and water purification. And in many cases we see over in this area usually land cover being used as a proxy for these important function ecosystem function and species abundance and diversity that really are determining the supply of ecosystem services. But with better information about fractional cover or vegetation indices telling us about the productivity of the system, we can understand better about the diversity of habitats that may provide, for instance, habitat for pollinators. While at the same time creating species distribution models that would understand about where pollinators are and lead us to the supply, the abundance of pollinators, which when we combine with the demand in terms of the types of different crops and their pollination dependence gets us to the service of crop pollination as one example. We might take a different path when we're thinking about erosion control. Again, we might use something like EVI, the vegetation index to get us to ecosystem structure or function. But in this case we're thinking about a cover factor, the action of habitat to retain sediment that leads us to this supply of ecosystem service. And for that model, we would need to know things about infrastructure like dams to get at a demand for the supply of sediment retention toward the ecosystem service of water purification. And then for a cultural service like ecotourism, we also care a lot about the individual species, not just not necessarily just the land cover. So in this case, again, we need species distribution models to tell us maybe in this case more about vertebrate diversity or specifically maybe bird diversity that create the supply. Oh, I'm sorry for the ecosystem service and the demand is dependent on access and amenities and other ways of accessing that. So we're going to be going into detail on the science advances for each of these services and then would love to discuss with you the ways that this information can be used. So just briefly we'll start with my colleague Raphael Schmidt who will speak about water quality regulation. And then Jeffrey Smith will speak on pollination crop pollination and Alejandra Eteveri will speak on ecotourism. Again, just want to remind you if you have questions for the panelists, please put them in the Q&A so we don't miss them because we would love to get around to answering every question. If you have technical needs just go ahead and pop those questions in the chat. Thank you so much. And I will turn this now over to Raphael. You can all see my screen now. My name is Buenas Diaz, my name is Raphael Schmidt and I'm a senior scientist at the natural capital project at Stanford and today I will be speaking about how we can use remotely sense vegetation indices to improve erosion on a large scale. And I would also like to acknowledge my collaborators here at Stanford and Costa Rica. And with that, let us start first, I'd like to say erosion is a natural process that is driven by many different factors such as climate, topography and vegetation. However, erosion is often also increased by human activities that is for example land conversion as you see here in the background picture and increased erosion will lead to first of all loss of fertile soil and fertile land. But it will also typically increase sediment loads in rivers and increase sediment loads in rivers has number has a number of negative impacts, for example on downstream dams and reservoirs, or on the natural beauty of rivers and in a country like Costa Rica that relies heavily on hydropower for its energy. And the beauty of its nature for tourism, human made erosion can have a major impact on the economy. However, in Costa Rica, as in other places worldwide, we often lack models to predict from their sediment originate. And today I'm going to speak about two topics. First, how to improve the representation of habitat and erosion models by using remote sensing data and second, I'm going to answer the question, if and how remote sensing data can help our ability to model sediment transport and thus our ability to manage erosion exo sediment in river. And first, let me say to represent erosion and sediment transport from a watershed. We usually apply specially distributed numerical models and numerical models, last to understand from bear sediment originates the troll vegetation and natural hybrid place to retain sediment, and where we should invest in conservation to minimize sediment export. So numerical models use typically created input data layers that you can imagine. I'm sorry, Becky. I'm sorry. I'm sorry, Mr. Smith. You need to put your microphone closer. It's the interpreter. I can't hear you. Maybe you can hold your microphone closer to your mouth. I'm so sorry to interrupt. No. So much better. Okay. Yes. Okay, thank you. Sorry about that. Sorry. Okay, so yeah, so basically use numerical models and maybe should go back because we missed the last slide. I'm sorry. I'm sorry. Maybe just backtrack a bit. I won't interrupt. Okay, so let's go back to the question. So basically today is going to be about two different questions. The first about how to improve the representation of vegetation in sediment models using remote sensing data. And second, does this use of remote sensing data improve our ability to model sediment transport and erosion. And to represent erosion and sediment transport from a watershed, we usually apply specially distributed numerical models. The models allow us to understand from where sediment originates, which role vegetation and natural habitat place to retain sediment and where we should invest in conservation to minimize sediment export. And these models use typically created data layers which we can imagine as basically being digital maps to represent the spatial distribution of most important factors behind erosion and sediment transports. For example, digital maps of topography of precipitation of soil type and very importantly of vegetation. However, often these data are highly simplified for example, the vegetation covering the land might be represented in attack in a categorical form for example for small sample watershed here. We have a mixture of pastos and light gray and Bosque and dark gray. And the values of the so called C factor that represents the role of vegetation for erosion control is then derived from tabulated global values from literature basically and together the land cover map and the lookup table can then be combined what you see here to the right into a spatial distributed representation of that C factor. You see that this map is categorical basically the C factor is low where there's forest and high where there's past also forest contributes less to erosion than pastos or other human land users. However, landscapes are very complex and thus the categorical representation of the C factor does not highlight the benefits of intact vegetation or the impact of small scale environmental degradation. For example, if you go back to the picture that I showed at the beginning, everything might be classified as forest while indeed we see that there's significant agricultural encroachment. If instead we classified everything as agriculture, we would not be able to capture the effect of the existing forests for retaining sediment on hillslopes. Thus, this leads me to the first question, how can we improve the representation of vegetation and natural habitat and sediment models through remotely sense data. And for that we do not really propose to replace categorical land cover maps. Completely rather what we propose is to supplement the categorical and well pasted land cover maps with an index called evi enhanced vegetation index that's derived for example from the Landsat satellites. And this index provides high resolution information on how healthy vegetation is. We then combine this information with land cover maps. And by using the evi information within each land cover class, we determine very vegetation is more intact within one class. As a result, a pixel that is classified as pastos in our top categorical map will always have a higher C factor than it picks the classified as Bosque. But as you see in the small map at the bottom right, if we then derive this continuous C factor, we greatly increase the spatial variability within each of these classes. And I'm not going to go into all the methodological details here, but please feel free to post more questions in the Q&A or be in touch with email. And I just want to highlight some results here. Two different C factor maps for Costa Rica you see to the left the categorical map and to the right the continuous C factor map. And you see the pattern is overall similar. However, the continuous map to the left to the right shows a much greater diversity in the continuous C factor. And we then use both of these C factor maps in the invest sediment delivery model or in short SDR, which is a numerical model for both erosion and sediment transfer and hill slopes. And the map to the right shows the difference between the model results where the map is blue. It means that the export is higher when we use the continuous C factor and red means that the model sediment export is higher when we use the categorical C factor. You see that there are some spatial patterns that emerge. For example, sediment export is always higher along the central mountains when we use the categorical C factor. However, on the pixel level if we zoom in, we see that there's a complex pattern where one approach might yield higher results in one pixel and another approach might yield higher results. And this suggests that the choice of the land use representation has also great impacts on the representation of sediment transfer. So for example, if we wanted to ask a question where we should maintain or restore habitat for sediment retention, results would greatly vary as a function of which land use maps we use. However, if we don't have any data to compare our model results to we cannot really make statements if any, if either of the two approaches better than the other. Nor can we really answer to our second question if remote sensing data does improve sediment models. Fortunately, for Costa Rica, ISE, the Instituto Costa Ricanza de Electricidad provided us long term sediment data from 49 hydrologic stations that you see on the map to the right with the small water drops. And the red colors indicate the sediment that was observed at each of these stations. And in general, more red means higher sediment and you see that the highest sediment was observed in history in the center of the country where the north and south have lower sediment yields. And we then used the watershed area of each gauge, which are shown in black and calculated the sediment yield from both models from the watershed area of each gauge and we then calculated the error between the observed and the model results for each watershed. First, I'd like to show you the results using the categorical C factor map, the standard approach. The map shows the model error in percent and you see that there are lots of dark browns, which indicates that the model overestimates for a specific watershed. And in statistical terms, we found that the mean square error of the standard approach is 11.5 tons per hectare and the maximum error is leading to an overestimation of up to 3,000%. We found that the model error was significantly lower when we used the continuous C factor as evident from the much lighter colors in the lower map. And those load lighter colors basically mean that the positive and negative errors are closer to zero. The mean square model error is also only 5.8 tons per hectare, which is significantly less than when we use the categorical land use maps. And the results provide evidence that using an EVI based vegetation factor can indeed improve the quality of erosion models. And lastly, I would like to make the point that using the continuous C factor does not just decrease model sediment export everywhere. But as the zoom map to the left shows, there is a heterogeneous difference between the two approaches on the pixel level. However, if you look cumulatively, the continuous C factor helps us basically to better estimate sediment export and to reduce the error on the watershed level. So basically in the same watershed, we can have sometimes higher and sometimes lower results using the continuous C factor. But together, the better representation of ecosystem diversity decreases the uncertainty in the numerical models. And this decrease in the industry enabled by remote sensing. To conclude, I would like to point out that the hybrid approach developed at the natural capital project improves how vegetation is represented in sediment models. One of the key findings is that this continuous representation of the C factor significantly decreases the model error and thus increases our ability to model sediment transport processes. In the future, we also hope to obtain more sediment data that would allow us to better calibrate the models and thus to further reduce model errors and build models that are as useful as possible for managing erosion and highlighting the role of vegetation for managing sediment in Costa Rica. And with that, I would like to thank all of our collaborators in Costa Rica and at Netcap. And if you want to learn more about how we use the sediment model for a management application in Costa Rica, my colleague Kelly Langhans will do a deep dive tomorrow morning. And also, please be in touch if you're interested in this approach or would like to contribute to our studies. Yeah, thank you for your attention to that. Back to Becky. Thank you so much, Rafa. And I just realized we forgot to do your poll, but I think you might have given it away, so maybe we'll skip that one. And we'll see if Jeff, when you share yours, would you like to start off with your poll? We've introduced some fun little things to try to keep this lively since we're not able to be in person. Do you want to start or do you want to wait till you get partway through? You're muted. Sorry, we can start with that. Start with it? Okay. So as Jeff is getting his slides ready, feel free to weigh in on what is your favorite drink or food that is dependent on pollinators. You can share your slides, Jeff, if you want to give them. Great participation. I wish that we had, I wish we had this rate of voting in our country. We already are up to 73%. Okay, so this is not a surprise to me, but I will share the results with you. And it is coffee is our winner, which is very appropriate for your talk, Jeff. It is, yes. Okay. So once again, just please be dropping your questions into the question and answer box will be answering them as soon as we can. And look forward to making sure we get all of those answered either live or via text, or possibly over email afterwards. And as Becky said, today I'm going to be talking about pollination. And so here we actually have a bee pollinating a little coffee flower. And specifically, I'd like to talk about how pollination across Costa Rican landscapes can be impacted by the, not only what's happening on the farm, but then also what's happening around the farm. And I apologize, I live very close to a train, which you probably just heard go by. So as many of you probably are already aware, pollination is incredibly important for a wide number of reasons, including the maintenance of natural ecosystems, providing economic returns. And for our own nutrition in terms of the generation of vitamins and calories and other types of important substances. And what we typically do when we think about modeling pollination for ecosystem services is we use a model that considers the nesting and floral resources available to bees across the entirety of the landscape. So in this landscape, we have a patch of crops here. And for some parts of the year these crops are really great at providing resources. So if we think about coffee, for example, during March or April when there's a lot of flowers, this can be a really great place for bees to acquire nectar and pollen. However, those flowers are only on the plants for one or two weeks at most. So for the rest of the year, these bees are going to be needing to acquire floral resources and other places. So they might be going to nearby natural habitats to get floral resources. They might also be utilizing these natural habitats for nesting and creating their homes. And all of these interdependencies of the bees, the orientation and allocation of the natural habitats and the crops, as well as how dependent the crop is on pollination influence the amount of yield that you're going to get as a result of pollination. So if we think about this last parameter, for example, when Becky put up that poll, we showed coffee, which only has about 25% dependence on pollination. So if there were no pollinators coffee would still produce some fruit it just wouldn't produce as much fruit, whereas chocolate is almost entirely dependent on pollination. Depending on which crop you're dealing with the pollination services maybe more or less important and that's just something that's important to keep in mind as it will ultimately influence the ways in which we think about the values that we can derive from pollination. And so this is the basic setup of the model today. What we would typically do is we would take maps of the current land use as Becky was saying, and try to use some type of modeling such as the invest pollination model to predict what the pollination or the pollinator abundance would look like as a function of these ecosystem types. As Becky's pointed out in her earlier talk, there's some problems with this first of all it's very hard to parameterize this model. We do have a lot of information and in a lot of cases we're making sort of educated guesses about what the right parameters are that relate a given type of land use to a pollinator abundance. It also is challenging to update these maps regularly as things change. So as many of you in Costa Rica might be aware, creating a land cover map like this one that integrates data from lots of different agencies is often a multi year process. In remote sensing and other proxies for ecosystem structure, we may be able to update these much more regularly to track what's actually changing on the landscape through time. And that's what I basically want to get into today about how we can use advances in remote sensing to better inform these models to make them more accurate to make them more easy to update and replicate. And hopefully just make this a more seamless workflow from satellites all the way to ecosystem service models. And the first adaptation that I'm going to talk about today is adding ecosystem diversity and this is getting to the point that I just mentioned where it's very challenging to convert a land use and land cover map into parameters for how bees are likely to respond to these different types of ecosystems. One thing that we can think about it instead is based on this idea of species functional diversity. So many of you might be familiar with the story of Darwin's Finches on the Galapagos Island where there are all of these different adaptations that these birds have developed to eat different kinds of seeds. So depending on what types of seeds the bird specialize in their beaks have been modified to fit that niche. So for example, this bird here might eat much larger seeds while this bird here eats much smaller seeds. And we can apply the same principle to trees and as a result, the forests that they make up. So we can think about these trees having different growth forms. They have different amounts of biomass. They have different phenologies and life cycles. So when we look at that over the course of the year, it might look something like this. And so you have to excuse me, this is a graph made for the temperate zone, but you could imagine one for the tropics as well, where we can use a remote sensing metric such as EVI, which we've already heard a bit about today that we can derive from MODIS or Landsat to track basically how green the vegetation is over the course of the year. We can track sort of the total greenness, the seasonal variation in greenness, the date at which it is the most alive and most vibrant and the date at which it is most dormant. And these are important parameters to describe how these forests change over the course of a year. And we basically can make this type of analysis for each pixel across the landscape and break them into categories. And that's what we've done here is we've essentially broken Costa Rica into five distinct categories for ecosystem functional types. And so this is a map that will probably look fairly similar to a lot of other zonation schemes that we've seen in Costa Rica, where the mountains have this greenish color, the coasts in blue, Guanacosta and the Nacoya in yellows and purples, where we can use remote sensing to essentially say that these ecosystem functional types are different from these blue ecosystem functional types. And we can update this each year as climate change and other factors change the distribution of vegetation across the landscape. And basically what this map looks like is a combination of all of these different categorical classes so the numbers here just represent ecosystem functional type one, 10, seven and 45 so in this map. We can put it down into a bunch of different categorical classes that we can then summarize to create some sort of ecosystem functional diversity. So for this map what we've basically said asked is for each pixel to look around and ask its neighbors how many different types of ecosystem functional types are around me. So if you're in a pixel that is surrounded by a lot of different types of forest you'll have a much higher level of ecosystem functional diversity. Whereas if you're in an urban or agricultural setting where a lot of the pixels around you are identical to you, the ecosystem functional diversity will be quite low. And the reason that we're using ecosystem functional diversity here is it essentially says how different is the phenology of all of the pixels surrounding you. So we think back to the be example from earlier. If you have a high ecosystem functional diversity. It means you're much more likely to have power to have pollen and other floral resources available for bees throughout the course of the year. So we can use this map of ecosystem functional diversity, as well as ecosystem functional types to predict all sorts of ecosystem services such as carbon sequestration pollination and ecotourism. So that we take this map of ecosystem functional diversity and apply similar principles to what is in the invest model to again create pollinator abundance maps that we can compare to the standard invest model outputs. And we can see that while some of the main patterns may still show through there are a lot of places in the country where the models predict different things where according to remote sensing maybe over or underestimating the amount of pollinators that are available across the landscape. So for example on Amistad here, our new model predicts that there should be fewer pollinators and that's because Amistad is a relatively homogenous forest so there's a lot of really great natural habitat available to bees but it may be largely the same types of resources available throughout the whole year. So this is one way in which we can change these two parameters of the model where we can switch up how we're dictating the amount of nesting and floral resources that are available to bees. But it's also quite possible that we want to change these bee population dynamics. What I haven't mentioned yet, which many of you may be thinking is that bees aren't uniformly distributed across Costa Rica. There's lots of places where bee density is higher or lower. So we might expect bees to be much more common say by the coasts than they are in the mountains because they prefer warmer climates, and you'd be exactly right. And that's where we're trying to use ecosystem earth observations again at an ecosystem levels. So things like temperature and precipitation paired with remote sensing of the amount of photosynthetic vegetation and the impervious surfaces, along with where we know bees occur across the entirety of Costa Rica to make predictive maps using species distribution models to say where should be abundance be higher or lower. And once again, we see that bee abundance in this model is expected to be higher along the coasts than it is in some of these mountainous areas. And this allows us to add in some of these large scale macro ecology or landscape level patterns to our understanding of how bees vary across the country of Costa Rica or indeed any country around the world. And what we can do with this map of bee abundance is combine it with other pollinator abundance maps that we've made from invest to come up with this new pollinator abundance map where we can once again see this reflection of the large scale landscape level patterns with bee density being higher along the coasts and lower in certain parts of the country. And we can once again map this on to where this deviates from our standard models that just use land cover, and again see relatively significant changes in terms of where bees are likely to be distributed across the landscape. This of course begs the question, why do we necessarily care about the total pollinator abundance. Obviously pollinators are important to maintain for the sake of biodiversity but they provide key ecosystem services such as improving yields of different crops. So we think about coffee farms which as I mentioned earlier are about 25% dependent on palanation. We can take this map of coffee farms across the entirety of Costa Rica and zoom in on different areas and we can essentially model for this small region for example I've blown up here. The expected amount of yield of coffee so here in darker green, we see higher yields and in light green we see lower yields. And so perhaps this is a place that is closer to a forest edge and has a wide variety of resources available to it, whereas this area here is likely more interior to the farm, and it's more homogenous in terms of the types of floral and nesting resources available to those bees. We're currently working with colleagues at Katia to verify this with in situ data or on the ground data, because this is obviously going to be essential as Rafa said in his talk earlier to ensure that our models are not just mathematical, but have some relation to these field plot data. And so just to summarize what we went over today, I showed the current state of palanation models and invest in other types of ecosystem service models and suggested ways that by adding ecosystem diversity, or by adding species diversity we can improve these models. And if we think about what this means for management or decision context, depending on the allocation or depending on the parameters that you have for a given crop system, you can infer how likely it is that by adding natural habitat you'll see increases in palanation. So if you're in a very, if you're working with a crop that is very dependent on palanation, it's much more likely that you should be adding forests nearby. Similarly, if you're working on a landscape that already has a lot of forest, by adding more forest you might not necessarily increase palanation as much as if you had added that forest to an area where there's not a significant amount of forest already. And with that, I'd like to thank a lot of great colleagues and co-authors on this talk, as well as lots of different agencies that have supported this work and turn it back over to Becky. Thank you, Jeff. And we're getting lots of great questions in the Q&A so please keep that up everyone. I'm going to launch another chat or another poll I think while Ale loads her slides. So here we go. This one is, which animals would you rather see in the wild? And there will be some of these that are featured in this talk. I'm rooting for sea, I'm going to be honest. Ooh, I just swung the boat. Marine animals, good point. That will be next talk when we're back next year, we promised to feature some marine animals. Okay. So, looks like maybe it was biased by the photo on the front when it is the kids' live. Okay, very well. I hope everyone can hear me. Okay, so we can hear. Good morning to all. Thank you for attending. I'm Alejandra. I'm a foreign ethologist. So, I'm so happy the kids' lives one during for this survey. I'm biased for terrestrial species. So, sorry about that. Sorry to the marine biologists out there. So, I'm going to speak about what variables explain nature-based tourism. So, I'm going to refer specifically to Costa Rica's eco-tourism. First, about the economic importance of Costa Rica's tourism, eco-tourism sector. This is information from the Ministry of Tourism and Costa Rica. In 2016, data revealed that tourism is approximately 7% or makes up 7% of the GDP. Last year, it created more than 70,000 jobs in the country. So, that's tourism in general. Sorry, not eco-tourism. But when we specifically speak about nature-based tourism, surveys ranging from 2013 and 2017 stated that 45% of tourists stated that nature or viewing nature was the main reason for visiting Costa Rica. So, keeping this in mind, we asked some research questions that include three main questions. Question one, how important is biodiversity to justify or predict nature-based tourism in protected areas? How similar are patterns of birdwatching tourism compared to those of nature-based tourism in protected areas? Two patterns persist at a national scale, not only in protected areas but outside of those areas as well. So, how did we go about this? We use different data sources and there are three ways of gauging tourism. The first one was using Flickr, a website where lovers of photography that are not professional will take your reference to photographs. Orange, the darker areas, is where there are more pictures taken by tourists that have uploaded this into the website. And we had a density map for pictures, basically. Flickr. And then to measure bird tourism, we have a platform called eBird. And for my colleagues in the audience and all bird lovers, you are all very learned and I'm sure you know eBird. It's a platform where bird watchers will make lists as to where they went to view birds. And orange means areas where there are more lists or visits by bird watchers created by the actual bird watchers. And we also use data from SINAC about the number of visitors to protected areas. And we have data from 41 protected regions. The three variables are all, they all correlate among each other by approximately 50%. So, generally where people like to take pictures are, they like to go to national parks to do that. SINAC is a national conservation system areas of Costa Rica. So, we're going to speak about the variables that we analyzed by diversity variables. And we mapped, we came up with distribution maps for 463 species for vertebrates. Out of these, 256 were bird species. And we also used water, sorry, data from the human footprint index or how much human impact was in pixels and the hotel density, as well as distance to roads. We analyzed environmental variables, temperature, force cover, distance to water. We used the following methods to answer our questions regarding protected areas. And they're called a multi-modeling averaging methods. They analyze all possible combinations of variables in linear regressions. So, given the variables I showed you by diversity, human footprint, temperature, all of these. I predicted where the Flickr pictures were with 14 variable. And then I analyzed all the possible variable combinations. I analyzed more than 16,000 possible models. And then given the models that best explained variants, I created averages of the coefficients I'm going to show you in the next slide. To answer the first question, how important is biodiversity in predicting nature-based tourism in protected areas? This graph shows a coefficient of data graph. The points to the right of the zero are positive predictors and the left negative. And if they don't cross, that means they're significant, as you can see. Out of the 635 best models that explained the variants, my data, we found that the most important data to for tourism in protected areas was human footprint, biodiversity, vertebrae, biodiversity and temperature. The rest were not important. In protected areas where people are visiting, they have the most presence of animals, but they're also more accessible. They're closer to roads and areas that are warmer. The second question. How similar are patterns of bird watching tourism compared to those of nature-based tourism in protected areas? We use the same data to answer this question, but instead of vertebrae species, diversity, we used only bird diversity, avian diversity. So here we see with that same graph that the only variable that mattered in predicting the bird watching patterns was the diversity of birds. So out of the 8,500 approximately best models, we found that bird diversity was the only factor that explained or that justified this pattern. Question three, do patterns persist across the tail, not only in protected areas? And we created these map distribution map. This one's called the Toledo, the long-tailed mannequin. They're found in the green areas. They're not in the white areas. So this is a species from the North Pacific area of Costa Rica, Guanacaste. And we created a map here. We superimposed one map over the other to create a map showing the total wealth of the species. Now we crossed these maps at temperature or superimposed them. Seasonality of rain, temperature seasonality, distance to protected areas, distance to roads, distance to water, forest covering human footprint. And then we asked which of these variables predict tourism? So here you can see a flicker picture-based tourism, green showing more pictures from flicker with the correlations and the spatial linear spatial regressions. We can see that what is predicted or what flicker photos predict is human footprint, wealth of rich vertebrate species, seasonality, rain seasonality, and these are positive predictors. For bird tourism, we saw that that result was even diversity, human footprint, temperature, and forest cover. So in this last slide, I'm going to show you here how the results of tourism that are based on flicker photos and e-bird for avian tourism. Here we see a map showing the difference when we include biodiversity models and when we don't. In these maps we see, okay, the left is for the flicker photos and right for the e-bird checklist. So blue shows the areas where we underestimate tourism if we don't include biodiversity. If biodiversity is included in these areas, they have more of a tourism potential. Here's Palo Verde. Red shows the areas where we overestimate tourism without biodiversity. The same goes for this map. In areas that are closer to the coast, including biodiversity, has much importance in explaining the patterns of as to why Palo Verde, in Palo Verde, for example, or explaining tourism in areas such as Palo Verde, protected areas. So conclusions in protected areas and at a national scale, we see a strong signal for sign of biodiversity and human footprint in predicting tourism. In protected areas, we see that only avian species wealth predicts bird watching while at the national scale, it's avian species wealth, human footprint, and temperature, I believe. So I do thank my colleagues, especially my Costa Rican colleagues that have helped us by sharing their data and with the questions, the central bank, as well as the three Bini colleagues. And I'm looking forward to working with all of you and hopefully with the other agencies that are listening in. If you have any questions, please write to us, reach out to us. You can Twitter me or email me. Thank you. Thank you, and I just want to pause really quick to remind you if you haven't taken this poll. We would love, I will, I will put this in the chat. I would love to hear what aspects of natural capital interest you most that will really help us in our future planning and I think it'll actually help in the discussion, the panel discussion as well. So I'm going to pop this in the chat and turn it over to Rafa for the next segment of this meeting. Thank you, Becky. Thanks Becky. Thanks, Jeffrey. And Rafa, and my namesake, and everyone else. Now we're going to speak about how we relate this global initiative to national initiatives. So I'm going to ask Gilberto Comara to join us. Are you there, Gilberto? Excellent. Thank you. So we're going to speak about how we can incorporate this at a national level, given the deal purview, the more global purview. I hope that you have no problems. I also speak Portuguese, so he can speak Portuguese if he wants. He's going to speak in Portugno, a little bit of Portuguese, a little bit of Spanish. We'll deal. Okay. Sorry for mixing languages like these, like this, but I hope you understand me. And, and that it makes sense with the translation. Okay. It's important to say that DO, as an international organization, is very happy to attend this seminar and very, very glad at Costa Rica's using most sensing data for their studies on biodiversity. Now let me speak in more general terms. I believe that developing countries such as Costa Rica, small country, and even though Brazil, which is large, well, even though it's large, it's also important that we leverage on our soft power. The leveraging on powers that's not traditional, like we're not using weapons, we're not using traditional potential indicators of the first world countries. We're improving using our soft weapons, say tourism, eco-tourism as weapons, let's say. It's very sad as a Brazilian with what's happening in my country, and I'm not going to go to that. You all know what's happening, but it's important to mobilize and to broaden our knowledge. And, of course, you know, we're not going to be the Stanford experts, but to understand or know that these tools are available is extremely important. So it's enormous pleasure to know that these types of studies are going on. So from a broader perspective, Costa Rica has had great scientific leadership with regards to Earth observation for biodiversity purposes, which really exceed the size of the country. So Costa Rica has significant importance in Earth observation, and it's an example. We've heard from Becky, Jeffrey, Rafa and Alejandra, and they've emphasized how science, good science, that is, can be used to improve the capacity of a country such as Costa Rica to take better care of their biodiversity. But to also understand from the microcosm from such as Costa Rica, how other countries, Brazil, Colombia can address many of their issues. Now it's not exactly the same as Costa Rica, but they all have similar problems, conserving biodiversity, deforestation. How it impacts the management of bees and other biodiversity impacts. So it's a great honor as director of geo and as a Latin American, as a Brazilian. To say that Costa Rica is, that's an example for all of us, and a virtuosical, well, scientific will and capacity and participation. We have great quality scientists such as our speakers today, it's just an enormous satisfaction. So, I'd like to thank everyone from Costa Rica, all the international cooperators on behalf of them. I just want to commend you on the great example you're setting and I wish many other countries would go about and have these types of things. Thank you so much for your kind words. And it leads to, or it leads to the following topic, how collaboration from that you and you know, Amazon Web Services and cloud credit program program that has provided much support for our national information environmental information system. Costa Rica also has another project, other geo project, and that's with Google Earth Engine on how to detect deforestation. We are developing the SDG 11 toolkit on sustainable cities and communities. And part of our know how of the fact that we're interested in these skills, because we have so many experts here that are viewing this seminar. And they really want to, some from Costa Rica, some are not, and they want to scale up these projects for the benefit of Costa Rican and other countries. And we'd like to become more involved in Earth observation for this project specifically for ecosystem accounting, which is completely tied to what we presented today. In Costa Rica, the information available at the National Environmental Information System, and based on the standard for monitoring and of land use coverage, we'll use this information to contribute to other information systems having to do with biodiversity here in the country, here in the country, other agencies. CC, which is ecological monitoring program, and the CBD of course, the central bank of Costa Rica is also partnered and will soon public ecosystem watersheds of Costa Rica will be the first country where we will provide this interesting information. And we can, and for that we need data of Earth observation, but also the work we do at a community, at communities, sorry, in communities, how this data can help management of resources at the community level is really important. How to improve how the payment programs for ecosystem services is run. We hope that provide more input so that that program will be better designed and create impact. And last but not least, to respond to what Martin said, marine life, that's something that we need to strengthen. And we need to use Earth observation for marine observation motion observation. We are a small country but 11th in ocean distribution. So we have to get up to speed with that. So just, this was just a brief introduction to some of the questions. So I ask all panelists to turn on their cameras and answer questions, and we will start responding the plethora of questions that we see. So, I'm not sure how we're going to organize ourselves to respond to the questions. I think we were going to kick off with maybe just a few from you first, Rafa, and just a couple to get started. Can all our panelists join us on the video. And then actually before we get started, I did see this question come up a lot. Oopsies. Sorry, wrong thing to show you. Will a recording be available? And the answer is yes. And you'll find it, we will send out an email, but you can also find it on our website. We'll put it here. And it will also be on Cinnia's Facebook page, as well as we are going to be getting into some of the data products at a workshop tomorrow that will are posted and will continue to be posted on Cinnia. And we will share that in an email going out to all attendees as well. So I just wanted to answer that question since it came up a lot. No, muchas gracias. Thank you. Thank you. That your interest, there's much interest that's been voiced. All the information that has been said. So we have a question. Wait one second. Rafa, Rafa, the first panelist. We don't see his face, but well, there we go. Has this model been used for simultaneous planning, if it's used for any simulations in Costa Rica. That were presented by our Stanford colleagues by Julie Rodriguez. She's asking the question. So the idea is to include this information incorporated into other monitoring systems we have in in our country and reinforce our systems. So Becky. Probably for all of the panelists, and it might have just been asked during Rafa's talk, because it gave some ideas, but I think, yeah, maybe we can, we can all speak a little bit to that. And I do think in the Q&A, there was already a really great exchange between Marcello, Hernandez Blanco, and I actually maybe I can unmute you, Marcello. If I'm smart enough to do that, let me see. It's an alphabetical order. Yes, there you are. Okay, let's let's invite Marcello as a honorary panelist. Are you here? Can you hear if we talk now? Yeah, hey, how are you, Becky? Hello, it's nice to hear you. Yeah, I just appreciated the exchange that was already happening on the Q&A. So maybe we can get a little back and forth here. But just as by way of background, we had spoken to Marcello early in our project. I think you attended our workshop last year that was the first one of these and already started thinking up ideas of how this work could feed into processes like the payments for ecosystem services design 2.0, which maybe you could speak a little bit about, and then we could talk about some other projects as well. Sure. Thanks for giving me the time and putting me on the spotlight. So, yeah, and thanks for the great presentations and also thanks for, because we've been talking for more than, I don't know now for more than six months, and you've been always so kind with providing all the data that you've been collecting and all the experiences that you've been learning from these processes. Thanks so much to you guys and to Gretchen as well and Rafa and everybody. So last year, at the end of last year, we started with the design of a new PES scheme, which we've been advancing a lot during this year. And I won't comment a lot of the scope of the scheme, but what I'm going to say is what we've been using invest in the results that you guys presented here and one of the main barriers or knowledge gaps and also critiques of the current scheme is that it lacks sometimes a little bit of additionality and also the targeting scheme that the program has can be enhanced in many ways. So we've been using these models to come up to areas where our investments in these payments could have the higher returns. And what we did and I can present this somewhere sometime is an index where you combine the ecosystem services potential or what you presented with also the level of threats. So combining those two would give you a priority areas to where to pay farmers and other implementers of the new scheme, which is going to be beyond forest and beyond the current ecosystem services that we're paying. So we want to integrate, for example, pollination. And I don't want to take a lot of time. Sorry. It's very helpful. It's neat to see, I think that Costa Rica has always had the will to lead on issues related to the environment even before the information was available. And so when the national payments for ecosystem services program got set up, it was just, you know, kind of all forest treated equally without necessarily being able to differentiate between areas that were providing higher value. To different beneficiaries. So that is something that I think with the 2.0 and beyond that we really hope to be able to influence. And there was also a project with IUCN a few years ago where a similar exercise was done prioritizing using invest and another related tool Rio's to prioritize the most important places for watershed interventions. So these tools are definitely being used. But as we say, the, there are limitations because they are hard to parameterize. And so that's what we're trying to address here and the models we're presenting right now are still in development. So not quite ready for action, but that's what we're trying to be transitioning toward in the next year. Would anyone else like to chime in there. Becky, if you want to continue with the next question, it's also an open question for the panelists that I think it's going to go on with the next question. It's a question for all the panelists. What are possibilities and challenges to scale up these types of methodologies and approaches and apply them to other areas. Anyone can answer. Yeah, that's I don't want to be very realizing. So I feel like I should let someone else go first, but this actually also coincides with a question that can bag sad addicts, which was a similar idea, but also not just what are the challenges but when is this something that will start to be ready in the near term medium to near term. And I think, yes, for Latin America, we're feeling pretty confident that we will have something by the end of our grant, which is a year for basically a year from now, a little bit more than a year from now. And probably that that seems, you know, relevant, not just for Costa Rica, but that we will be able to scale up to Miso America, maybe all of Latin America. I think it's harder when we start moving into substantially different biomes, especially the ecosystem functional type map that Jeff presented that it will be an input to the pollination modeling is really different from biome to biome and we derive will be deriving it in a different way using different functional attributes and a lot of testing is still needed. And so one of the biggest challenges I think is the testing is getting the testing data to know whether these approaches are valid or our improvements over the traditional approaches that are more land cover And so Rafa present Rafa Schmidt presented a great example of where we can use testing data, the sediment data really generously provided by you say can show pretty dramatically how what an improvement in accuracy it is. But those data are hard to come by. And we don't always have cooperation of local partners like that. And so, one of the biggest challenges I think is just knowing whether when we move from one place to the other. Is it still as good, or do we need to make some further tweaks to the approach. And, Jeff for I would you like to say anything about biodiversity modeling maybe. Just would chime in that earth observations are incredible at letting us sort of interpolate between observations in the field. So we have these places where people have gone out and collected data and that's absolutely essential to sort of making these maps. Because the more data you have to parameterize the models, the better the models will fit. And so the more data that you have across the country, the easier it is to sort of interpolate between the points. So for example, Costa Rica is one of the places with the richest biodiversity data anywhere in the world. So it's a lot easier to fit models of where species are occurring, because we just have so much more information. So that's not to say that the models couldn't work in other places, but we have a lot more confidence in the way the models work in Costa Rica than other parts of the neotropics where data is a little bit more limited. But with so many advances in remote sensing and community science and citizen science, we're hoping that those data gaps can be filled in. I probably also just to chime in briefly, you know, like in many places worldwide there has been a tendency to basically like abandon ground measurements, ground measurement stations for example. I'm sorry, because I couldn't hear the first part. Can you just repeat the first part? I'm really sorry. Okay, so like in many places worldwide, their hydrologic stations have been abandoned, for example. So we have much less sediment measurements, for example, in the recent years than we have in the past in many countries. And I think it's really important for leveraging all these new and important satellite measures that come in that we also maintain our ability to measure properties such as sediment transport and water discharge on the ground because one new method cannot really replace the ground measurements. So it's really important that national programs focus on both and as Becky said, unfortunately, the availability of these ground data in space and in time is very variable between countries and even between regions, which is still a bit of a limitation if you want to scale up to like a continental scale, for example. Yeah, so if you do know of data sources in another region of the world, please get in touch. Thanks, go ahead. Thank you. No, Dandere, continue. Yes, just to add to the question is we're speaking about scaling up the project, and we're looking at the scientific and data aspect. As the panelists mentioned, but how about if we look at this from a viewpoint as to how this information gets to decision makers, which is why we are holding the seminar today so that all of you can be informed as to what's being developed through the different platforms through the Costa Rican, for example, national environmental system and the territorial information system. That's where this information will be made available. The idea is to interpret to state this data in late terms, so you will understand it. And we'd like to connect at an international level. How can this data be made available to other international platforms. How can we replicate it? Thank you. DO is truly aware that open science is required for this purpose. And we're embarking on our deal knowledge hub. So project results such as this one. The ecosystem services for natural capital ecosystem services can be placed in our hub. They can be uploaded and shared with others. So we're interested in using natural capital resources results. We're using this data to be able to make all scientific data available from this project. We want to confirm that it will be that these results will be conveyed through the right software. That we did help the scientific data that's in bibliographies that that's the that's the long term goal. So to make the idea is to have a limp. Sorry, the idea is to make all of your presentations or results available through a link to GitHub, for example, and be able to get comments. For example, you know, where's the model? Where can I find this model? Where can I find the description? So all of this should be contributed constantly. Because if we don't speak about things, if we don't make our information available in a single repository, for example, I think it'll be it won't be used as much. So the idea is, or all of you to reach out to Joe to be able to marry your project also with our knowledge. So, so at 5pm, I'll be attending my fifth video conference. I do this all day. So I'm sorry, but I have to leave you. It's been a great pleasure and an honor to speak to all of you today. I learned tremendously from these presentations and congratulations to your wonderful presentations. Rafael, Becky, everybody, it's been wonderful. And to you, Rafael, thank you very much. Thank you. And until next time. Thanks. Thanks so much. But no, no, no, no one else disconnected. Just Dilberto. Actually, Rafa, that last point, I was, I'm sorry, I didn't get to ask him while he was here, but it's related to another question that that someone had asked about whether it would be possible. Wouldn't, wouldn't it be possible to have a regional Central American initiative for Earth observations. And there is actually through, there's a biodiversity observation network that Geo hosts that Geo bond and there's a, there's a Central America bond that is exactly that kind of region wide initiative and I was going to ask Hilda to speak a little to that but but we can there, there is information on their webpage if you're interested in following up on that that was Martin, my summer. There are many regional initiatives. Sica, the Central American integration system is working on Earth observation. And Mary deal. Mary do we, but how do we. There's so many local initiatives how do we work together through these different platforms through better communication. And again, holding these types of activities I think helps so we can know that you're out there and we can find new options to improve our, our linkages. So speaking about people, how can this information help identify where these ecosystem services are that can provide greater benefit for people's quality of lives at their homes, businesses. So I'm going to ask the panelists to think about how to address this, how can we prove people's qualities of life at a national level. Yeah, thanks for that question Rafa it's a it's a hard one to end on in two minutes. Maybe this is the good for thought and I think I might also show you the results of our poll as we as we depart here, but maybe just to kick off for a few thoughts from from the rest of the panelists to is that we presented sort of predominantly on the ecological side today and I do think that there is a bias or there's a there's a lot there's a lot of work being done on the ecological side in Earth observations and not nearly as much being done on the social side and that's because some things are hard to see. Of course we can see things like infrastructure and you know detect human activity with nightlights. And we're getting better and better using machine learning to identify specific things we might be looking for like dams that cap just led a project with Nat Geo identifying dams using machine learning and Earth observations. But it's still it's it's much harder to come by and especially when it comes to what you were saying quality of life indicators, really depending on people's vulnerabilities and their, you know individual dependencies on nature, and what other substitutes they have for natural capital and what sorts of you know marginalization they face in society at large and a lot of times data on that kind of information is very hard to come by. Even nationally, let alone globally. And these are the kinds of partnerships we need to be able to form with local groups that do have a better handle on the people side that we could start to link with Earth observations that might let us extrapolate from patterns that we see, you know between human activities we can detect and human characteristics that are really important and always work with the social media data is a beginning of that is trying to link some Earth observations with some, you know people's preferences and and habits that is not an Earth observation but is a another form of data that can be linked to that. I'm just. So we've also been doing work with the central bank and Ministry of Tourism to not just stop at where people are going to visit them but what that means in terms of hotel revenues and where people are staying and investing money. So those places will be supporting the local economies not only through hotels but then also through restaurants and other industries as well. So that's one of the next steps with the central bank to explicitly link it to economic well being, but that's obviously something that can be more challenging as Becky said for some services.