Projects

Mary Blair/The American Museum of Natural History
Expanding Wallace Biodiversity Modeling Software to Support National Biodiversity Change Indicator Calculations for GEO BON Assessment and Reporting
16-GEO16-0027

This project is linked to the National BON: Colombia BON
More information: wallaceecomod.github.io

Background

Effective policy responses to changes in biodiversity are only possible with adaptable analytic tools that leverage the influx of data from biodiversity observation systems. Such analytic tools must also be streamlined and readily mastered by researchers making scientific recommendations. In this project, we will create software to assess biodiversity change indicators by building on the recently developed software Wallace as a new GEO BON in a Box tool. Wallace is an R-based application with a graphical user interface that supports species distribution modeling (SDM) in a reproducible, flexible and extensible platform to facilitate a wide range of ecological analyses. These models characterize environmental suitability for species and can be used to estimate species’ geographic distributions — a GEO BON Essential Biodiversity Variable (EBV). Wallace harnesses biodiversity data from online databases or user input, assembles a variety of tools for building and evaluating models, and offers guidance for important conceptual and methodological issues. Critically, its modular nature facilitates the addition of new, cutting-edge innovations by other programmers, as we propose here. To expand Wallace as a new BON in a Box, we will develop two new R packages to be integrated as new modules for Wallace’s workflow to facilitate biodiversity change indicator calculations for GEO BON assessment and reporting, in collaboration with the Colombia BON as a model.

The first package will use single species range maps in conjunction with remotely sensed (RS) products to estimate the species’ current range. It will harness RS products derived from NASA (and other) satellites, including percent forest cover, vegetation classes, and NDVI, which can be spatio-temporally matched with recent in situ observations of species’ occurrence to determine habitat tolerances (e.g., to land use change). The second package will allow calculation of key indicators including single-species biodiversity change indicators such as range size, extent of occurrence, percent suitable land cover, or projected trends under future scenarios at multiple spatial extents. Calculations will also include multi-species indicators, such as the EBV taxonomic diversity, and others not currently implemented through other BON in a Box tools, with a focus on ones most useful at regional scales.

Module outputs from Wallace will be designed to feed directly into other BON in a Box tools, such as our international collaborator Instituto Humboldt’s BioModelos. BioModelos is a web application enabling expert validation of species range estimates in support of the Colombia BON. We will also develop training materials for Wallace targeting conservation practitioners that reflect best practices for using SDM for biodiversity change indicator calculations, resource management and biodiversity conservation decision-making. Our project results will address challenges to overcoming the gap between best modeling practices and biodiversity conservation decision-making and will facilitate responsible reporting on biodiversity by national BONs.

Via focus on the development and enhancement of open-source, user-friendly software to contribute to BON in a Box, our proposal will advance broad application, exploration, visualization, and understanding of EBVs and generation of EBV data. Furthermore, developing a connection between Wallace and BioModelos, another BON in a Box tool, will be valuable for future interfacing with efforts at IUCN, BIEN (the Botanical Information and Ecology Network), and other similar projects that use SDM predictions as inputs. Finally, although we will focus development of our BON in a Box with the needs of the Colombia BON and with NASA Earth Science data, we will incorporate flexibility to address the needs of diverse users at multiple scales from national to global, and to work with remotely sensed data from any space agency in GEO.

Progress Update, March 2019

• Held a formal end-user consultation workshop in Bogotá, Colombia to consult with invited biodiversity experts for their input on how to best develop the new software with their needs in mind.
• Learned that less than half of participants currently use species distribution models (SDMs) in their work, and we aim to increase the proportion of our end user base using SDMs to contribute to biodiversity assessments for resource management and conservation decision-making through this project.
• Based upon end user input from the workshop, our first R package MaskRangeR has been developed. The package spatiotemporally matches in-situ observations of a species’ occurrence to remote sensing (e.g. MODIS-derived) products to derive accurate metrics important to species’ tolerances, and use them to refine species distribution model predictions and estimate the species’ current range.

Progress Update, October 2019

• We held our second formal end-user consultation workshop in Bogotá, Colombia from June 11-12, 2019 to capture end users’ feedback on the functioning prototype of the first phase of the expanded Wallace interface that includes our new R package MaskRangeR.
• With the BioModelos team, we completed a data architecture and software integration plan and a workplan to achieve it, focused on our goal of integrating Wallace with BioModelos.
• At AmeriGEO Week from Aug 19-23, 2019 in Lima, Peru we collaborated with the other NASA-funded projects working with the Colombia BON and held a workshop (in Spanish) on “Operationalization of essential biodiversity variables (EBV): Species Distribution: Workflows, best practices and principles, global initiatives and national needs, available tools” including a practical exercise on BioModelos and Wallace.

Progress Update, April 2020

• Based on end user input from our second consultation workshop in 2019, our team has begun development of our second of two new software packages. This package will calculate biodiversity change metrics related to species’ ranges such as range loss over time and will help users identify and compare areas based on the metrics.
• We have acquired all of the necessary data for a suite of use cases that will be used for software testing, and to detail user decision making processes. The use cases also help ensure that our software development reflects the needs of Colombia BON users, including their desired functionalities as identified in the consultation workshop.
• A major milestone was achieved with an update to BioModelos’ API.
• Regular calls among integration team members continue; the BioModelos-Wallace integration workplan is on schedule.

Progress Update, October 2020

• The software package changeRangeR has now been beta tested by project participants. This package calculates biodiversity change metrics related to species’ ranges such as range loss over time and will help users identify and compare areas based on the metrics.
• The new integration functions developed to connect the BioModelos and Wallace tools have been beta tested; these enhance the functionality of both as BON in a Box tools.
• The changeRangeR and BioModelos API functions are being integrated into the Wallace GUI software platform. These will be launched and tested by project partners and participants at the third annual end-user consultation (virtual) workshop in mid-October.
• Former graduate trainee Dr. Jamie Kass led a publication detailing a use case for one of key functions in the first new R package developed for the project, maskRangeR:

Kass, JM, SI Meenan, N Tinoco, SF Burneo, RP Anderson 2020. Improving area of occupancy estimates for parapatric species using distribution models and support vector machines. Ecological Applications. https://doi.org/10.1002/eap.2228

• PI Blair presented on project progress at the North American Congress on Conservation Biology in late July and also held a virtual Wallace training workshop for conservation practitioners, which was well attended and received highly positive evaluations.
• A Group on Earth Observations–Google Earth Engine proposal was funded to project collaborators at BioModelos to produce free maps and protocols using GEE data to identify near real-time loss habitat alerts.

Project Update, April 2021

• Completed and beta-tested a prototype expanded Wallace GUI software platform with new features to measure biodiversity change and to pull and push data to the BioModelos database.
• Expanded Wallace was launched and tested by project partners and participants at 3rd annual end-user consultation workshop. Because of the COVID-19 pandemic the workshop was transformed into a virtual format and held online 13-15 October, 2020. Testing showed demonstrated improvements in the performance of users’ abilities to estimate species’ ranges and calculate biodiversity change metrics. The workshop also debuted training materials to help users understand the new Wallace features including R markdown vignettes and demo videos, now available on YouTube.
• Project team members held two additional virtual species distribution modeling training workshops: 10th Student Conference on Conservation Science – New York (October 2020)
• Pontificia Universidad Javeriana in Bogotá, Colombia for their master’s program in conservation (January, 2021).
• Participants across the three workshops provided valuable feedback to finalize software development; they found some software bugs and described other issues that the Wallace development team has since addressed.

Walter Jetz/Yale University
Activities to Advance, Build, and Deliver Remote-Sensing Supported Species Distribution and Species Abundance EBVs
16-GEO16-0078

This project is linked to the Working Group: Species Populations
More information: mol.org

Background

Biodiversity and the many ecosystem functions and services it underpins are undergoing significant changes worldwide scientifically rigorous information is needed to monitor their status and trends. Essential Biodiversity Variables (EBVs) capture key constituent components of biodiversity change. Facilitated by the GEO Biodiversity Observation Network (GEO BON) efforts are now underway to conceptualize, develop and deliver these essential components to enable a more focused, integrated and effective biodiversity monitoring and provision of national indicators in support of assessment and policy. The EBV class ‘Species Populations addresses the most fundamental aspects of biodiversity change, the variation of species geographic distributions and abundances in space and time. A GEO BON Working Group is currently advancing the concept frameworks, methods and infrastructure for a generalizable capture of spatial and temporal variation in species populations and support the delivery of EBV and indicator products for policy and management at national and global scales. Proposal PI Jetz and Co-PI McGeoch are the Co-Leads of the GEO BON Species Populations Working group, and the goal of the proposed work is to explore, demonstrate, and deliver key advances for species population EBVs and indicators enabled by remote sensing coupled to novel in-situ data sources.

Species occurrence and abundance data are spatially and temporally sparse and biased, limiting appropriate EBV development based on just this ‘in situ’ information to small spatial extents and few species. Remote sensing, especially global products at high resolution and now extensive temporal coverage such as MODIS and Landsat, provide a powerful complement. Remotely sensed products, when coupled with growing in situ species data, and with models increasingly capable of handling disparate input data, are enabling predictions about species population change for unsampled locations at national and global scales and for an increasing number of species.

In the proposed work, we will undertake select activities to advance the use of remote-sensing data for species distribution and abundance modelling. We are strategically including partners (as Co-Is and named Collaborators) with ongoing projects in this area that allow us to combine a range of activities and achieve synergies. We are also able to benefit from GEO BON partner Map of Life (PI Jetz, Co-I Guralnick) as modelling and web- infrastructure that ensure the effective and continued delivery of results to GEO BON and national and global stakeholders. First, we will advance the use of inventory datasets such as camera trapping and plant plot survey data combined with remote sensing data, as these (in combination with broadly-available incidental point records) enable the use of more sophisticated models for assessing change in occurrence and abundance. Second, using these data we will improve and demonstrate modelling methods that are particularly well-suited for assessing population change when combined with remote sensing. Third, we will build on these two efforts to deliver a range of spatial and, as possible, spatio-temporal EBV products for select species in North America (birds, mammals), South Africa (Proteas), and globally (select terrestrial vertebrates), including in particular select invasive species. Fourth, we will use these EBVs to provide improved information for indicators that have been endorsed by the CBD and (as core indicator) by IPBES.

Our proposal thus responds to the key GEO BON call elements, including advancement of EBV methodology, EBV and indicator production, and provision of information relevant for BONs. The proposed project is endorsed by the Global Biodiversity Information Facility, the Convention on Biological Diversity, the IUCN Invasive Species Specialist Group, and the South African Environmental Observation Network.

Progress Update, April 2019

Published a study that provides the definition and analysis framework for the Species Population Essential Biodiversity Variables and sets the basis for the work planned: Jetz, W., J. Beck, M. J. Costello, M. Fernandez, S. Ferrier, G. N. Geller, M. Gill, R. P. Guralnick, P. Keil, M. A. McGeoch, C. Meyer, H. M. Pereira, E. C. Regan, D. Schmeller, & E. Turak. (2019). Essential information for monitoring species distributions. Nature Ecology & Evolution, 3, 539-551. https://doi.org/10.1038/s41559-019-0826-1

• Hosted a workshop at Yale University in May 2018 to bring together 25 members of the GEO BON, camera trapping, remote sensing and spatial biodiversity modelling community to advance remote sensing-supported Species Population Essential Biodiversity Variables (SP EBVs).
• Completed the biodiversity ingest workflows and metadata so data is now moving into Map of Life. Work is progressing for developing spatio-taxonomic cleaning routines to reconcile records to the Map of Life master taxonomies.
• Tested deployment of integrative modeling frameworks on Map of Life, with focus on data integration across data types, and scale-up examples in North America and South Africa
• Set up Earth Engine to perform on-demand integrated modeling, focusing on deductive modeling, but with tests of inductive modeling, and change assessment approaches. Deductive models of habitat suitability change (supporting the GEO BON – Map of Life Species Habitat Index) run at global scale.
• Established and tested process for integrating data from the Global Register of Introduced and Invasive Species (GRIIS) into Map of Life, as a data input for the production of SD EBV’s

Progress Update, October 2019

Published a study that provides the definition and analysis framework for the Species Population Essential Biodiversity Variables and sets the basis for the work planned: Jetz, W., J. Beck, M. J. Costello, M. Fernandez, S. Ferrier, G. N. Geller, M. Gill, R. P. Guralnick, P. Keil, M. A. McGeoch, C. Meyer, H. M. Pereira, E. C. Regan, D. Schmeller, & E. Turak. (2019). Essential information for monitoring species distributions. Nature Ecology & Evolution, 3, 539-551. https://doi.org/10.1038/s41559-019-0826-1

• Following initial workshop at Yale in Year 1 convened working group on the model-based integration of different spatio-temporal biodiversity data types (manuscript close to submission). In collaboration with Wildlife Insights (camera trapping) and Movebank (tracking data) partners, advanced empirical reporting tools and metrics addressing data complementarity.
• Completed and published group work defining the Spatial Populations EBVs (Nature Ecology & Evolution 2019; https://doi.org/10.1038/s41559-019-0826-1). As part of the sTwist working group (https://www.idiv.de/en/stwist.html) advanced visualizations, workflows and tools in support of invasive EBVs.
• Developed modelling tools for the integration of remote sensing data with expert-based lateral and elevational constraints for species distributions. Applied them to South American hummingbirds and all African birds.
• Advanced data fusion modelling approaches and workflows for S African plants and global terrestrial vertebrates in support of Species Population EBVs and an indicator derived from them, the Species Habitat Index.

Progress Update, April 2020

• Following the first year workshop, our working group has completed a manuscript and R package exemplifying the integration of five data types (counts, presence only, presence/absence, detection-non-detection, movement) in one single point process modelling (PPM) framework.
• Integration of eMammal data with other data sources in Wildlife Insights (https://www.wildlifeinsights.org/home) is near complete. We concluded an initial coverage assessment of camera trapping data soon available through Wildlife Insights.
• A paper describing Wildlife Insights is now published: Ahumada, J. A., E. Fegraus, T. Birch, N. Flores, R. Kays, T. G. O’Brien, J. Palmer, S. Schuttler, J. Y. Zhao, and W. Jetz. 2020. Wildlife insights: A platform to maximize the potential of camera trap and other passive sensor wildlife data for the planet. Environmental Conservation 47:1-6. doi.org/10.1017/S0376892919000298
• A commentary piece on an EBV-supported invasion monitoring, including recommendations for post-2020 biodiversity framework is now published: McGeoch, M., and W. Jetz. 2019. Measure and Reduce the Harm Caused by Biological Invasions. One Earth 1:171-174; doi.org/10.1016/j.oneear.2019.10.003
• We have extended and updated the Species-Population-EBV-driven Species Status Information Index and Species Protection Index for use in the Environmental Performance Index and CBD process. These EBV-based indices provide support for policy and management.

Progress Update, October 2020

• Held virtual two-day workshop on ”Borrowing strength across species to address the Wallacean shortfall”, including fifteen international experts in machine learning, Bayesian models, point process models, and remote sensing. Developed draft figures and outline of conceptual working group manuscript on the topic, led by grant-supported postdoc Kevin Winner.
• Completed the production of validated global-scale Species Distribution EBV products, covering 2001-2018, for 20,000 terrestrial vertebrate species. Built online infrastructure for the sharing and visualization of the products.
• Ran updates of the Species Habitat Index (SHI) and Species Protection Index (SPI), both developed under the auspices of GEOBON and endorsed by the Convention on Biological Diversity and IPBES.

Project Final Update, April 2021

Here, a selection of final outcomes for the project, which has now ended. However, many related activities continue, cf Map of Life.

Biodiversity data integration in support of EBV development:

• Advanced metadata and data standards for inventory (e.g., survey-based) data integration, Humboldt Core.
• Supported adoption of Humboldt Core through formal Biodiversity Information Standards (TDWG) process: https://www.tdwg.org/community/osr/humboldt-extension/
• Through engagement and collaboration with Movebank and Wildlife Insights initiatives integrated GPS-tracking movement data and camera trapping data with other spatial biodiversity data with future availability through Map of Life
• Developed new Map of Life species map user interface supporting advanced display, filtering and download of different biodiversity data, e.g. https://mol.org/species/map/Ciconia_ciconia
• Mobilized a range of new data sources and data types in Map of Life for visualization and further model-based integration (see https://mol.org/datasets/)

Methods development and engagement around model-based data integration for EBV production

• Community engagement. Held three in-person / virtual workshops with experts on multi-species approaches for species distribution EBV prediction. Working group is ongoing and group MS is in prep.
• Developed methods to combine expert lateral and elevational distribution knowledge with point records and remote sensing to predict species distributions

Species distribution and EBV predictions:

• Produced initial 1km global distribution predictions for thousands of terrestrial species based on the above approach.
• Produced validated, global-scale Species Distribution EBV products, covering 2001-2018, for 20,000 terrestrial vertebrate species.
• Built online infrastructure for visualization and use of these range products.

Biodiversity Indicator development:

• Production of global, national and species-level values of the Species Habitat Index (SHI) and Species Protection Index (SPI), both developed under the auspices of GEOBON and endorsed by the Convention on Biological Diversity and IPBES. SPI for example featured here: https://www.half-earthproject.org/maps/ (via explore countries)
• Production of global, national and species-level values of the Species Status Information Index and a newly developed Species Sampling Effectiveness Index.

Gretchen Daily/Stanford University
Improving Linkages Between Earth Observations and Ecosystem Service Models with Essential Biodiversity Variables
16-GEO16-0031

This project is linked to the Working Group: Ecosystem Services

Background

Biodiversity dynamics underpin the regulation and provisioning of many ES, but ES models used commonly in decision-support tools do not adequately represent the relationships between different levels biodiversity (genes, species, ecosystems) and ES. Most ES tools use land-use/land cover (LULC) as the sole input representing ecosystem contribution to ES, and therefore neglect the potential impacts of within-habitat variation or changes in diversity. The Essential Biodiversity Variable (EBV) framework establishes metrics for spatially-explicit representation of biodiversity change over time and addresses these multiple dimensions of biodiversity. In theory, EBVs can feed into models for ecosystem services to represent the benefits of biodiversity to people. However, few analyses demonstrate the extent to which EBVs measured by Earth Observations (EO) represent biodiversity dynamics across multiple taxa, and no tools currently translate EBVs into ES and/or their contribution to human well-being.

We propose a two-stage project to fill these gaps: the first stage links EO to the EBVs most relevant to ES, and the second stage links EBVs to ES and their benefits to people. We will develop and test models that use EO to calculate ecosystem-level EBVs (E-EBVs) and predict biodiversity patterns at ecosystem and species level (S-EBVs). These EBVs will then serve as inputs to new models that predict three focal ES that exemplify three categories of services: regulating (water regulation), supporting of provisioning (pollination for crop production), and cultural (wildlife for intrinsic value, and wildlife-based tourism). We will test these new EBV-driven ES models against those driven purely by LULC using independent data on ES production.

These models will demonstrate several pathways for using EOs to link biodiversity to ES. For water regulation, E-EBVs calculated solely from EO may be used instead of LULC in process-based models that include other inputs (e.g., precipitation, soil depth and bulk density, etc.) to produce ES supply (seasonal water availability). For pollination and tourism, S-EBVs can be derived from EO with additional predictor variables, ideally from globally-available data (e.g., microclimate, soil type, aspect, etc.). These spatially-explicit S-EBVs can then be used individually or in combination to serve as proxies for ES providers, like pollinators or wildlife, that are essentially the origin of supply for these ES. In both cases these models of ES supply, when combined with models or data for ES demand (e.g., infrastructure, access, location of beneficiaries, etc.), will produce spatially-explicit estimates of ES benefits to people.

We plan to analyze and compare EO-EBVs derived from multiple EO mission data sources in Costa Rica, including Landsat, MODIS, Hyperion, and GLAS, as well as airborne data from the 2005 LVIS mission and 2005 CARTA mission. To link EO to E-EBVs to S-EBVs, we will develop predictive models based on species observations collected in Costa Rica, augmented with additional field collection. To link EBVs to ES, we will use a variety of techniques suited to the specific service, including sophisticated hydrologic modeling (water regulation), field exclusion experiments (pollination), and regression analysis with social media data (tourism). For each of these linkages we will: 1) collect or process data for model development, 2) develop model, 3) generate nationwide maps for Costa Rica, 4) test model against independent data sets, and given testing results, 5) produce generalized software tools.
We expect this project to fill key gaps in our knowledge of how EO can predict patterns of biodiversity at different levels, and in our understanding of the relationship between biodiversity and ecosystem functions and services. This is an important step forward for the fields of EO and ES, and for support of decisions on conservation of biodiversity and sustainable development.

Progress Update, March 2019

• Met with the Costa Rican Minister of Environment and Energy, Carlos Manuel Rodriguez, his staff, and staff of the Central Bank of Costa Rica in February to develop a needs assessment for geospatial data that could inform National Accounts for Ecosystem Services, Sustainable Development Goal working groups, and their national monitoring programs. Also held public lectures about our project attended by > 100 representatives of the government, civil society, and academia. (https://semanariouniversidad.com/pais/minae-presenta-alianza-con-universidad-de-stanford-y-the-natural-project/)
• Hosted the President of Costa Rica, Carlos Alvarado Quesada, at Stanford in March, including a private reception with leaders in Silicon Valley, and a discussion with students, faculty, and the Stanford community on Costa Rica’s decarbonization strategy and the transformation to green inclusive growth. (https://news.stanford.edu/2019/03/14/costa-rican-president-speaks-stanford/)
• Gathered EO products including fractional cover of bare ground, vegetation, impervious surfaces, tree cover, and temporal variance in vegetation dynamics. These are used along with GBIF data to develop species distribution models for reptiles, amphibians, birds, bats, non-flying mammals, and pollinators, early results of which we presented at the Fall meeting for AGU in December. (https://agu.confex.com/agu/fm18/meetingapp.cgi/Paper/451509)

Progress Update, October 2019

• Leveraged this project and the partnership with MINAE to secure $60k in cloud credits on Amazon AWS; using this remote machine we are running models globally at 300 m resolution and for Costa Rica at 10 m resolution.
• Added water quality regulation (sediment retention), even though it was not part of our original scope (because there is not currently an ecosystem service account for water quality regulation; there is a separate water account and at some point in the future the Bank is interested in merging this accounts); it can still be improved through EO data.
• Defined a policy-relevant scenario to illustrate the value of EO information in refining how governments target conservation or restoration interventions for ecosystem services, given limited funds to enforce everywhere. We are simulating reforestation of all riparian habitat according to the forest code specifications: 10 m in rural areas, 50 m on steep (>45 degree) slopes.
• Project team traveled to San Jose, Costa Rica, in July, to meet with MINAE and BCCR staff again, and discuss data needs to support the project. Useful datasets were identified to help validate model output: within SINAC (the department overseeing the protected areas network, which manages the forest inventories), IMN (the Meteorological Institute, which manages a wide array of remote-sensing data), and ICE (national electricity provider, with hydrologic data); carbon data have been obtained from SINAC and data-sharing agreements are in the works with IMN and ICE.

Progress Update, April 2020

• Developed a new hybrid approach to modeling bird biodiversity that bridged global (GBIF occurrence and broad climate variables) and local (long-term transect observation and fine-scale habitat variables) data. While general patterns of climate are still apparent, spatial projections now also show clear signals of local land-use characteristics. In particular, the traditional species distribution modeling approach (using only global data) undervalued the importance of Amistad International Park and the Nicoya Peninsula in terms of their ability to support species richness.
• Collaborating with Amanda Armstrong and Howie Epstein, we have applied their Ecosystem Functional Type (EFT) approach to map EFT diversity across Costa Rica. We are using this in pollination modeling to parameterize nesting and floral resources in habitat surrounding croplands.
• Mapped tourism (predicted by bird biodiversity) and above ground carbon storage (estimated with lidar) for Costa Rica at the national scale; both approaches were refined using improved local data. The use of lidar for the carbon map reduced error by 50% compared to the more common approach of using LULC type.
• Formed collaboration with local consultancy charged with developing new financial mechanisms for the national Payments for Ecosystem Services (PSA) system. Our products will be used to inform that process.

Progress Update, October 2020

• Now have model outputs for carbon, sediment retention, pollination and nature-based tourism utilizing advanced EO/EBV data. Main focus now is on determining whether these EO/EBV-based model outputs outperform models using LULC.
• Obtained (after two years of waiting) sediment data from the Costa Rican water company, ICE, for 59 gauges across the country. This enables comparison of the project’s NDVI-based sediment modeling approach to the more conventional LULC-based approach to see which is more accurate.
• Received coffee maps from i-CAFE and are now testing an EFD-based (ecosystem functional diversity) pollination model compared to the LULC-based approach in anticipation of validation data from a contract with CATIE who will collect pollination data will next year starting in January (during the next coffee flowering period)
• Validation data for carbon storage that we obtained from the Forest Inventory suggest that satellite-derived biomass is far more accurate than a LULC-based approach to modeling carbon

Project Update, April 2021

From Becky Chaplin-Kramer:

Project results presented in October in a Virtual Workshop:

More than 150 participants, from Costa Rica and 30 other countries
• Technical session included staff from MINAE and the Central Bank
• Many other organizations and Ministries attended
• More information, including the complete recordings in English and Spanish, can be accessed at https://simocute.go.cr/events/riquezasnaturales2/

Sediment retention model:

• Using EVI instead of LULC class vastly improved performance, cutting the mean square error nearly in half
• Discussions with ICE (the water company who provided the data) about how they can use this model to improve their decision-making are underway

Ecosystem Functional Type (EFT) map for Costa Rica:

• • Applied methodologies developed by Howie Epstein & colleagues (another NASA project)

Compared different sensors (MODIS and Landsat) at different scales (Costa Rica-wide and for the Nicoya Peninsula

• Mansucript in preparation

Pollination model:

• Using EFT diversity instead of LULC class improved performance
• Overall error was reduced by a third (21% to 14%) and the underestimation error was reduced by nearly three times (44% to 17%)

Tourism model:

• Addition of biodiversity improved performance. Biodiversity was derived from species distribution models utilizing Landsat-based fractional cover
• Without considering biodiversity, the tourism value of large, intact tracts of forests such as La Amistad National Park are underestimated, and their easier to access (but less diverse) park edges are overestimated
• Manuscript submitted to PNAS

Maria Kavanaugh/Oregon State University
Dynamic Seascapes to Support a Biogeographic Framework for a Global Marine Biodiversity Observing Network
16-GEO16-0113

This project is linked to the Thematic BON: Marine Biodiversity Observation Network (MBON)
More information: Ocean Observing System OL3

Background

The GEO-BON Thematic Marine Biodiversity Observing Network (MBON) supports the operational monitoring of marine biodiversity, defined broadly as the variety of life at the gene, species and ecosystem levels. Understanding how biodiversity changes across these levels provides insight to ecosystems’ resiliency or vulnerability to global change, and their capacity to maintain the vital ecosystem services on which humans depend. The systematic observation of life in the global ocean requires both standardized in-situ (field observations) and remote sensing methods, and a means for stakeholders to compare status and trends between locations. In the US, three pilot programs are sponsored in a partnership established between the NASA, NOAA, the Bureau of Ocean Energy Management (BOEM), the Smithsonian Institution’s Tennenbaum Marine Observatories Network, and the Ocean Biogeographic Information System (OBIS). We propose to link these demonstration projects with efforts worldwide via a common biogeographic framework that characterizes the four dimensional variability of ocean dynamics. This framework will be relevant at global and regional scales, can serve as the basis of scaling local observations of biodiversity to regional responses to climate, and provide an objective means to intercompare biodiversity-habitat relationships in advective environments.

Proposed work: Objective classification and validation of remotely-sensed dynamic seascapes using high resolution (1 km) satellite data has occurred at regional to local scales in temperate-upwelling and tropical reef environments as part of the US MBON pilot program. Seascapes provide a framework to assess and scale up patterns of biodiversity and effects of environmental change on pelagic community structure, ranging from microbes to fish. With partners at the Arctic MBON and the Distributed Biophysical Observatory (DBO), we will add a polar ecosystem case study to demonstrate transferability and track changes in multi-trophic level diversity in response to changing sea-ice, temperature and nutrient delivery, and ocean chemistry.

Concurrent to the US MBON effort, the US Geological Survey and the Environmental Systems Research Institute (ESRI) embarked on a global 3-D static classification of ocean volumes using multi-year climatologies of physical and chemical data archived in the World Ocean Atlas (WOA); future efforts may classify based on seasonal climatologies. We will classify dynamic seascapes globally using readily available satellite products, and inter-compare the boundaries associated with each classification scheme as a means of cross validation. Intercomparisons will highlight variability at depth not directly observed by satellites but important to pelagic species. Conversely, dynamic seascapes will illuminate changes in habitat quality and extent not apparent from climatologies.

Deliverables: 1) A global seascape classification scheme that reveals multiscale ocean dynamics, using ocean color, winds, temperature, sea surface height and sea-ice from NASA repositories; 2) An intercomparison with ESRI /USGS partners to determine boundary co-location across methods; 3) Extension of the US-MBON high resolution seascapes to polar seas; 4) Examination of biodiversity patterns across polar seascapes using AMBON and DBO data.

Partners include GEO members NASA, NOAA, USGS, ESRI, Arctic DBO and Blue Planet with the goal to provide a sustainable, standardized, and validated biogeographic framework to monitor and intercompare biodiversity patterns and trends, accessible to stakeholders worldwide. This proposal is complementary to that of Montes et al. that seeks to unify in situ sampling methods across sites in the Americas contributing to a Pole-to-Pole or global MBON.

Progress Update, March 2019

• Continued the routine production of global and regional seascape maps through NOAA CoastWatch and the US MBON Portal.
• Established an engaged stakeholder community within the US and globally, including the US-MBON pilot projects, NOAA IOOS nodes, and the AmeriGEOSS Pole-to-Pole program participants.
• Continued compilation of local validation data sets including chl-a, particulate organic carbon, SST, salinity, phytoplankton and zooplankton species counts and forage fish surveys.

Progress Update, October 2019

• Seascape identity maps and probability distributions continue to be served in near real time on NOAA CoastWatch at 9 km spatial scales and 8 day and monthly time resolution.
• Provided satellite and seascape training at MBON Pole to Pole workshop (Mexico, April 2019).
• Downscaling, seascape validation efforts, and community structure comparisons are now underway at three locations in the California Current (Santa Barbara Channel, Monterey Bay, Oregon and Washington Coasts), in the Arctic (Chukchi and Beaufort Seas), in the Antarctic (with Palmer LTER) and in the Florida Keys.
• Began discussion with the GEOBON Ecosystem Structure Working group to utilize Seascapes as an Ecosystem Extent Essential Biodiversity Variable.
• Synergistic effort funded in the UK to determine applications of dynamic seascape classifications for marine spatial planning. (July 2019).

Progress Update, April 2020

• Developing downscaling methodology in conjunction with NOAA CoastWatch and the US MBON; the Northern California Current and South Florida regions are being used as test areas. Downscaling will provide greater spatial resolution and thus support a wider range of users and applications.
• A sensitivity study is underway for the Arctic to address the impact of cloud cover on our ability to produce accurate seascapes. Because microwaves penetrate clouds the goal is to incorporate microwave-based seascapes (these are at a coarse, 25 km resolution) to interpolate between high resolution seascapes using ocean color spectra (9 ,4 and 1 km).
• Working with the Gulf of Maine MBON to create regional seascapes. Also exploring extending these to the Massachusetts and New Jersey shelves.
• Specific products (and likelihood of adoption) under consideration for EBVs include: dynamic seascape extent (high likelihood), seascape uncertainty/novelty (experimental), seascape diversity (moderate likelihood). These are being developed in part with MBON Pole to Pole in the Americas (E. Montes, Lead PI).

Project Update, April 2021

Updated global seascapes are freely available every 8 days on the NOAA CoastWatch site (https://coastwatch.noaa.gov/cw/index.html)

• Applied for funded extension associated with COVID to support further stakeholder engagement with US MBON, MBON Pole to Pole, and NGO partners including Nature Conservancy in the Gulf of Mexico.
• Conducted Global Seascape overview for US MBON during February All Hands Meeting with Workshop devoted to training and sharing access tools for seascapes on NOAA CoastWatch: https://coastwatch.noaa.gov/cw/satellite-data-products/multi-parameter-models/seascape-pelagic-habitat-classification.html
• Development of R Package and RShiny App to facilitate access and analysis of seascapes developed by Ben Best (Ecoquants), E. Montes, and M. Kavanaugh. https://marinebon.org/seascapeR/

Seascape related publications submitted, accepted, or published in last 6 months.
GEO BON

• Woodill, J, M.T. Kavanaugh, M. Harte, and J. Watson. 2021. Ocean seascapes predict distant-water fishing vessel incursions into exclusive economic zones. In press, Fish and Fisheries
• Estes M., C. Anderson; W. Appeltans; N. Bax; N. Bednarsek; E.Boss; G. Canonico; S. Djavidnia; P. Fietzek; M. Gregoire; E.Hazen; M.T. Kavanaugh; F.Lejzerowicz; F. Lombard; P. Miloslavich; K.Moller; J. Monk; E. Montes; H. Moustahfid; M.Muelbert; F. Mueller-Karger; L. Peavey Reeves; E. Satterthwaite; J. Schmidt; A. Sequeira; W. Turner; L. Weatherdon. Monitoring Life in the Sea is a Critical Component of Sustainable Economic Growth. Marine Policy Submitted

Case studies through MBON pole 2 pole (*=direct result of PI-led training/engagement demonstrating sustained stakeholder base.)

• Mazzuco* et al., 2021. Coastal reef larval recruitment in response to seascape dynamics in the SW Atlantic. Submitted to Limnology and Oceanography
• Livore*, J.P., Mendez, M.M., Klein, E., Arribas, L. and Bigatti, G., 2021. Application of a Simple, Low-Cost, Low-Tech Method to Monitor Intertidal Rocky Shore Assemblages on a Broad Geographic Scale. Frontiers in Marine Science, 8, p.294.

US-MBON

• Kavanaugh MT. and D. Siegel. Satellite remote sensing and the Marine Biodiversity Observing Network: current science and future vision. 2020 Oceanography submitted
• Benson, A. T. Murray, F.M. Karger, G. Canonico, E. Montes, J. Trinanes, and M.T. Kavanaugh. Data Management and best practices for the evolving Marine Biodiversity Observing Network. 2020 Oceanography submitted

Student training

• Blaisdell, J., Thalmann, H., Klabjor, W. Zhang Y., Miller, J. and M.T. Kavanaugh. 2021. A dynamic stress-scape framework to evaluate potential effects of multiple environmental stressors on Gulf of Alaska Juvenile Pacific Cod. Accepted Frontiers of Marine Science.

Victor Gutierrrez-Velez/Temple University
Integration of Earth Observations for Decision Making on Biodiversity Management and Conservation in Colombia: Consolidation of the Colombian Biodiversity Observation Network
16-GEO16-0052

This project is linked to the National BON: Colombia BON
More information: bosproject.org/en

Background

Earth Observations data and products represent great potential for understanding, managing, and conserving biodiversity. The ability of decision makers to translate this vast amount of data into information for practical decisions is constrained by differences in data structures, sources, and formats and the complexity of operations necessary for data assimilation, merging, analysis, and interpretation. Overcoming these constraints is critical in Colombia, where recent and unprecedented achievements to end more than 50 years of internal war constitute challenges and opportunities for sustainable development and biodiversity conservation. Colombia is one of the most biodiverse countries in the world. Current high rates of deforestation and ecosystem degradation constitute sensitive pressures for biodiversity conservation. Colombia is also one of only three countries worldwide pioneering the implementation of a national Biodiversity Observation Network (BON) under the GEOBON umbrella. The Colombian BON has identified as crucial, the need to provide processing tools and analytical skills to decision-makers for integrating biodiversity information into national and regional development planning.

We propose to develop a Decision Support System (DSS) for Biodiversity Conservation and Management in Colombia. The DSS will facilitate the integration, processing and analysis of Earth observations in one platform, to inform biodiversity decision-making. The DSS will allow users to bring existing Earth Observations into compatible data structures and to develop metrics for different geographic domains and time periods. These metrics can then be integrated into modules to characterize: 1) biodiversity conservation status, 2) human and climatic drivers of biodiversity and ecosystem change, 3) future impacts of land use on biodiversity and 4) priority areas for biodiversity sampling and monitoring.

The DSS will be dynamic, interactive, open source and flexible to include new data and functionality as new products are generated or needs emerge. The toolset will be accessible at two “tiers” of complexity, suitable to different levels of user expertise. The first tier will contain user-friendly dashboards and customizable queries to enable users, with no previous experience in computer coding, the ability to parameterize, visualize, and interpret information for monitoring, reporting and verification purposes. The second tier will offer greater flexibility in customization, parameter settings and modeling outputs to users with basic to advanced skills on computer coding. A case study will be implemented in Colombia to test the functionality of the DSS. The project also includes workshops, training sessions and webinars to broaden the participation of stakeholders in the development of the DSS and to build capacity for its use for research, policy decision-making and environmental monitoring.

The DSS will inform the National Environmental Information System of Colombia and will be hosted and administered by the official institution leading the implementation of the BON in Colombia, the Alexander von Humboldt Institute for Biodiversity, with support from the other participant organizations. The functionality of the system will also be provided as a BON in a Box set and will be designed with a level of generalization to facilitate the development biodiversity support systems for other national or regional BONs.

The DSS is highly relevant to NASA’s Earth Science and Applied Sciences Program. Given the demonstrated human dependence on ecological systems, enabling the use of Earth Observations for timely and informed decisions on biodiversity planning and management will promote long-term societal benefits. The system will also add to GEOBON’s goals of contributing to effective policies for sustainable management of biodiversity and ecosystems by facilitating the delivery of Earth Observation information for research and decision-making.

Progress Update, March 2019

• Priority ecosystems and potential indicators of biodiversity change for Colombia were identified
• The first package to derive Essential Biodiversity Variables on forest extent and fragmentation from global forest change data was published and is available at:(https://cran.r-project.org/web/packages/forestChange/index.html)
• User needs and priorities for biodiversity decision-making were identified in a workshop attended by a broad array of stakeholders
• A proposal submitted to the PEER program, focused on the derivation of EBVs at the subnational level in Colombia, was accepted and has now started

Progress Update, October 2019

• Expanded the functionality of the forestChange package to facilitate the production of both EBV metrics and EBV statistics derived from Earth Observation products for areas of interest defined by ecosystem change data.
• Produced prototypes of both the front-end interface for interaction with TIER 1 users and a modular and portable API for data querying, processing and delivery. Benchmarking has helped to improve system efficiency and reliability.
• Curated national biodiversity datasets to be incorporated in the front-end user interface.
• Workshops and training events in Colombia and Peru have provided useful feedback from researchers, national and regional officials and decision-makers and generated capacity in the use of current project functionality. A tutorial is made available to trainees.

Progress Update, April 2020

• Recent software development has expanded previous functionalities by retrieving and integrating data on 1) forest and freshwater ecosystems 2) forest ecosystem structure and 3) species distribution for quantifying and reporting changes in key essential biodiversity variables.
• An online database that includes both national and global biodiversity datasets derived from Earth observations has been constructed. This database facilitates the calculation of spatial biodiversity indicators for pre-defined or user specified areas of interest in Colombia.
• A prototype of a cloud computing system for data querying and processing has been configured and is in testing mode. The system is the engine behind the decision support system
• A prototype protocol for the transmission of user queries from the front-end interface of the decision support system to the cloud computing system and for the retrieval of information back to users has been developed and is in testing mode.
• New capabilities have been developed to derive metrics associated with 18 Spatial Biodiversity Indicators (SBI). Metrics can be calculated for pre-defined areas or for user-defined regions of interest, covering the extent of Colombia. The indicators include Biome Types, Ecosystem Types, Degrees of Threatened Ecosystem, Biodiversity Offset Factors by levels of ecosystem degradation, among others. These indicators will facilitate the reporting of advances in achieving national and international biodiversity conservation and sustainable development commitments in Colombia.

Progress Update, October 2020

• An alpha version of a package named rastermapr has been created. The package implements a user friendly workflow for land cover backdating and provides a classification as well as a change assessment.
• The software functionality of the ecochange package that characterizes landscape patterns was expanded (formerly called forestchange but now includes additional ecosystems).
• Periodic meetings and workshops with the host institution were held to identify new priority datasets and metrics that can be derived from the software functionalities developed in the project and that will be made available to Tier 1 users.
• Further refinements were made to the software that harmonizes national and global datasets to derive EBV metrics on changes in forest and wetland distribution. These enhancements improve the temporal resolution of current official maps while ensuring the suitability of the data and indicators to be used for decision-making.

Project Update, April 2021

Development of Simulated Operational Environment (SOP):

• Beta-tested the system backend in a SOP using an R-markdown environment, enabling testing the tasks of the back-end infrastructure, including cloud-infrastructure set up, load and control of the cloud infrastructure, API testing, polygon definition, and output reporting.
• Improved the SOP to make the computation of spatial indicators faster and produce graphical representations of the outputs.

Derivation of ecosystem change maps:

• Finalizing first version of the harmonized forest change product for Colombia that improves the spatial coverage and temporal resolution of the official forest change maps used for decision making in Colombia by incorporating data from the global forest forest change product.

Software development:

• Developed two R packages named ecochange (published) and rasterMapR (alpha version). ecochange V1.3 constitutes a major upgrade to a previous package (forestChange) and is already published in the R CRAN repository. This ensures that the package is fully documented, reproducible, compatible with all major operating systems and fully operational.

Manuscripts:

• A manuscript is under evaluation for peer-reviewed publication that proposes an alternative probabilistic approach for change mapping that is more robust and efficient than standard methods.

Presentations:

• Contributed with online presentations to the GEOBON virtual meeting (Jul 6, 2020 – Jul 10, 2020), AGU Fall Meeting online everywhere (Dec 1, 2020 – Dec 17, 2020), and GIS day at Temple University (Nov 18, 2020).

Partner engagement:

• Continue sustained regular meetings and workshops with collaborators from partner in Colombia (Institute von Humboldt) to 1) identify priority indicators and data visualization alternatives to implement in the Tier 1 interface (biotablero) to decision makers and 2) enable the transfer of data and processing algorithms for the sustained use of future versions of the datasets as new updates in input data become available.

Training materials:

• Wrote vignettes in R markdown to help users of the system to implement both the ecochange package and the backend infrastructure. The vignettes of ecochange constitute an updated version of the previous documentation developed for describing the former package forestChange.
• Developed a video describing procedures to download remote sensing products using the ecochange package. This video will be shared on a youtube channel during the next weeks and will be followed by other videos soon.

Howard Epstein/University of Virginia
Ecosystem Functional Diversity of the Circumpolar Arctic Tundra
16-GEO16-0020

This project is linked to the Regional BON: Arctic BON

Background

The Group on Earth Observations (GEO) Biodiversity Observation Network (BON) was developed to improve the synthesis and acquisition of data on biodiversity for the suite of potential user communities, including scientists, land managers, and policy makers. GEO BON aims to identify Essential Biodiversity Variables (EBVs), of which a category is Ecosystem Function. Within the EBV class of Ecosystem Function, the currently identified variables (Net Primary Productivity, Secondary Productivity, Nutrient Retention, and Disturbance Regime) all describe aspects of ecosystem functioning, however, none of these variables directly quantifies the observed diversity in ecosystem functioning. Just as species composition (including biodiversity) has the capacity to provide information about the resistance and resilience of ecosystems in the face of environmental change, the diversity of ecosystem functioning provides a similar opportunity. With this as our basic premise, and the identification of a potential gap in the set of GEO BON Ecosystem Function EBVs, we propose to develop a classification of Ecosystem Functional Types (EFTs). EFTs are a top-down characterization of the spatial and temporal heterogeneity of ecosystem functioning (regardless of similarities or differences in ecosystem structure), based on areas of the land surface that process energy and matter in similar ways, and potentially show coordinated responses to environmental factors. There is effectively an unlimited number of ways in which ecosystem functioning can be characterized and classified; however, certain EFT classifications are more achievable than others, given limitations of data availability over large spatial extents. The development of an extensive EFT classification, using certain functional variables can be (and has been) accomplished through the use of satellite remote sensing, and this is the approach that we intend to utilize for this effort.

While EFTs can be developed globally, we will focus this particular effort on the terrestrial ecosystems of the arctic tundra. We do this because 1) the Arctic is a region that has a relatively high degree of spatial variability in ecosystem functioning, but is also one that has been changing dramatically over the past several decades (and is projected to continue to change), as a result of dynamics in climate and land use, 2) a well-defined classification of the ecosystem structure (vegetation community distribution) of the arctic tundra currently exists in the Circumpolar Arctic Vegetation Map (CAVM) – the region therefore provides an excellent opportunity to identify and evaluate the relationships between ecosystem structural diversity and functional diversity, and 3) the Arctic is currently one of two regional Biodiversity Observation Networks, facilitated by the Circumpolar Biodiversity Monitoring Program (CBMP) – within the Arctic Council, Conservation of Arctic Flora and Fauna (CAFF) Secretariat.

Progress Update, March 2019

• Presented the Ecosystem Functional Diversity concept and preliminary results for the Arctic at three conferences: The Arctic Biodiversity Congress in Rovaniemi, Finland; the International Circumpolar Remote Sensing Symposium in Potsdam, Germany; and the American Geophysical Union Meeting (GEO-BON session) in Washington, D.C.
• Held our first annual project team meeting this fall in Charlottesville, VA.
• Working in the Google Earth Engine environment to develop the first set of Ecosystem Functional Types for the Circumpolar Arctic, based on the seasonal dynamics of the MODIS NDVI.

Progress Update, October 2019

• Implemented a procedure to map Ecosystem Functional Types (EFTs) for the circumpolar Arctic tundra, based on similarities in the seasonal dynamics of the MODIS Normalized Difference Vegetation Index (NDVI), indicative of carbon uptake dynamics. Pixels (EFTs) are at the 400m resolution. Results suggest a wide range of variability in ecosystem function, even with individual Arctic tundra subzones, as defined by the Circumpolar Arctic Tundra Vegetation (CAVM) map.
• In the process of developing the procedure to go from the Ecosystem Functional Type classification to a circumpolar map of Ecosystem Functional Diversity.
• Developing relationships with two different GEO-BON end-user groups: i) the Ecosystem Function Working Group, and ii) the Arctic BON, also known as the Circumpolar Biodiversity Monitoring Program (CBMP).
• 2nd Annual Project Team Meeting held in Charlottesville, VA – September 26, 2019.

Progress Update, April 2020

• Completed the Ecosystem Functional Type (EFT) and Ecosystem Functional Diversity (EFD) products for the circumpolar arctic tundra and are developing a paper on this topic.
• Compiling environmental datasets to evaluate the controls on EFT and EFD distribution throughout the arctic tundra.
• Conducting time series analyses of EFT and EFD dynamics using the MODIS NDVI record.
• Developed EFT maps for other NASA projects, including Becky Chaplin-Kramer’s “Improving Linkages Between EO and Ecosystem Service Models with EBVs” (above) and her Gobi Desert project.
• Conducting a regional-scale analysis of the environmental controls on NDVI and EFT distribution throughout the Yamal Peninsula of northwestern Siberia. The Yamal presents an interesting case study for this analysis as an area of arctic tundra that has indigenous migratory reindeer herders and rapidly expanding natural gas extraction operations.

Progress Update, October 2020

• Project variables Ecosystem Functional Types (EFTs) and Ecosystem Functional Diversity (EFD) have been peer-reviewed and submitted to the Convention of Biological Diversity (CBD) Post-2020 Global Biodiversity Framework and for the UN System of Environmental-Economic Account. These variables are of particular interest to stakeholders and end users at GEO BON, the Conservation of Arctic Flora and Fauna, and the Circumpolar Biodiversity Monitoring Program.
• EFTs and EFD are presently the dominant essential biodiversity variables (EBVs) representing the Ecosystem Function Working Group of GEO BON. Whereas the scope of the project began with the circumpolar Arctic, the EFT and EFD framework has been adopted by other researchers for other areas globally (e.g. Costa Rica, Columbia, Mongolia).
• The GEO–Google Earth Engine proposal “Essential Biodiversity Variables – Scale Up: Harnessing the power of GEE and GEO BON to bring EBVs from concept to application-ready global solutions in the service of society” was funded for 2020-2022. Collaborators include the University of Florida and German Centre for Integrative Biodiversity Research.
• Developed a new water mask for the arctic tundra using a dynamic Landsat surface water product. The mask identifies permanent and seasonal water and is implemented for a MODIS dataset within Google Earth Engine.
• Completed Max NDVI trend analyses (calculation of Sen’s slope and Mann-Kendall trend test) for the regional scale analysis on the Yamal Peninsula, Siberia; environmental driver datasets continue to be derived.

Project Update, April 2021

• Developed a new water mask within Google Earth Engine that is useful for areas with substantial coverage of permanent and ephemeral surface water (such as the Arctic) and recalculated the Ecosystem Functional Types for the Circumpolar Arctic
• Analyzed the environmental controls on spatial patterns and temporal dynamics of Maximum NDVI and seasonally (time) integrated NDVI for the Yamal Peninsula in northwestern Siberia, a major region of gas extraction and native reindeer herding (Nenets)
• Produced a gap analysis of Arctic protected areas with respect to ecosystem functional diversity
• Collaborated with Chaplin-Kramer’s project to develop an Ecosystem Functional Type map for Costa Rica to evaluate biodiversity patterns and improve ecosystem service modeling
• Presented four papers at the American Geophysical Union Meeting in December

Patrick Jantz/Northern Arizona University
Quantifying Forest Vertical Structure Using Spaceborne Lidar: A GEO BON Essential Biodiversity Variable Application in Colombia
16-GEO16-0066

This project is linked to the National BON: Colombia BON
More information: https://www2.nau.edu/goetz-lab/wordpress/index.php/tropical (scroll down to project info)

Background

More comprehensive mapping of earth’s biodiversity is a priority to help prevent further biotic impoverishment and maintain ecosystem functions and services. Intergovernmental organizations, national governments, and non-governmental organizations have developed numerous approaches to map and monitor biodiversity priorities and to report on status and trends in biodiversity. The Group on Earth Observations Biodiversity Observation Network (GEO BON), a flagship GEO program, has been at the forefront of efforts to coordinate and deliver this biodiversity information to national governments, researchers, and other groups with an interest in conserving biodiversity. The Essential Biodiversity Variable framework (EBV) and the BON in a Box tool are two of GEO BON’s primary vehicles for organizing and delivering technical information to its participants.

In addition to global efforts, GEO BON facilitates national and regional BONs. Colombia, with oversight from the Colombian Ministry of Environment and assistance from GEO BON, is currently developing its national framework for biodiversity observations. Colombia’s Instituto de Investigación de Recursos Biológicos Alexander von Humboldt (IAvH), a non-regulatory, government research institute, is primarily responsible for developing the framework. IAvH is making progress on multiple fronts, including development of the BioModelos application which informs the Species Populations EBV. However, development of a habitat structure EBV is still needed and would make a valuable addition to Colombia’s national biodiversity observation framework.

Habitat structure is one of six EBV classes defined in the EBV framework and has been highlighted as a priority for mapping from space. Although earth observations have greatly increased our understanding of earth’s ecosystems over the past several decades, especially the horizontal distribution of forests and other major vegetation types, we still lack precise measurements of vertical habitat structure and its distribution for large areas.

Lidar data from NASA’s Global Ecosystem Dynamics Investigation (GEDI), which is currently underway, will significantly enhance habitat structure information available for large fractions of the earth’s forests. Because of its sampling density (~15 billion samples over its two year lifespan) and footprint size (~25 m diameter), it will do so with high precision and at scales commensurate with in-situ biodiversity observations and moderate resolution optical sensors such as Landsat and Sentinel. Thus, GEDI can serve as a foundation for global and national level habitat structure EBVs.
To build on Colombia’s active involvement in GEO BON and leverage new habitat structure information provided by GEDI, we propose to develop a cutting edge, lidar based habitat structure metric as a GEO BON Essential Biodiversity Variable (EBV) application. Our primary objectives are to:

1) Develop a consistent and scalable workflow that uses spaceborne lidar measurements to provide high precision estimates of the extent and distribution of forest structural types, establishing baselines for subnational, national and international biodiversity targets and reporting requirements.

2) Work with IAvH to incorporate EBV data and workflows into Colombia’s biodiversity observation framework.

The proposed work will contribute to the GEO work plan by supporting the efforts of GEO BON to leverage earth observations for development and implementation of a forest structure EBV that describes and quantifies forest vertical structure across ecosystems.

Progress Update, March 2019

• Currently modeling vegetation structure in several areas in the Neotropics and assessing the potential for GEDI data to enable wall to wall mapping of forest structure in Colombia, using a variety of multispectral, radar, and lidar data sources.
• Presented an analysis at the AGU 2018 Fall meeting describing use of GLAS lidar data to map broad vegetation structure types across Colombia in the 2000’s and to understand the factors that influence their geographic distributions. Elevation, temperature, and human influence were the most important factors related to variability in vegetation cover, vegetation height, and sub-canopy height distribution.
• Held two workshops in Fall of 2018 in Bogota, Colombia in collaboration with the Humboldt Institute; worked with participants from non-profit, academic, and government sectors to identify potential lidar forest structure applications in Colombia. There is considerable potential to work with current and planned projects to calibrate and validate a vegetation structure Essential Biodiversity Variable.

Progress Update, October 2019

• Refined the approach for developing vegetation profile classes for Colombia using GLAS lidar data and detailed ecosystem maps.
• Developed a workplan for a web application that will integrate GEDI data access, analysis, and presentation steps. The application will generate statistical and visual summaries for end users and will have hooks that allow access to underlying R modules for integration with other spatial decision support tools.
• Offered a workshop in Spanish on calculating vegetation structure metrics using lidar data at AmeriGEO Week 2019.
• Developed canopy height predictions for ecoregions in Colombia and Brazil using radar and multispectral satellite data trained on aircraft lidar measurements. The approach is described in a manuscript titled “Integrating LiDAR, multispectral and SAR data to estimate and map canopy height in tropical forests”.

Progress Update, April 2020

• Published a manuscript on the use of multi-spectral and radar imagery to predict lidar derived vegetation height in different Neotropical forest types in Brazil and Colombia.Fagua, J.C., Jantz, P., Rodriguez-Buritica, S., Duncanson, L. and Goetz, S.J., 2019. Integrating LiDAR, Multispectral and SAR Data to Estimate and Map Canopy Height in Tropical Forests. Remote Sensing, 11(22), p.2697.
• Initiated a collaboration with the Sinchi Amazonic Institute of Scientific Research, a non-profit institute of the Colombian government responsible for biodiversity assessments in the Colombian Amazon. They have shared more than 30 one hectare forest inventory plots spanning the Colombian Amazon which we will use to validate vegetation structure maps derived from GEDI lidar data, multispectral, and radar imagery.
• Initiated a collaboration with the BioModelos group to investigate how GEDI vegetation structure data can inform the Humboldt Institute’s ongoing species distribution modeling efforts.
• Working on a manuscript that investigates relationships between tree diversity in Colombia’s Choco ecoregion and wall-to-wall maps of multiple vegetation structure metrics, including canopy height, height of median energy, canopy vertical heterogeneity, and standard deviation of normalized heights.
• Working on a manuscript presenting a novel approach to infer species-absences for species distribution modelling using presence-only data. We expect that this approach will improve ecological inference related to vegetation structure variables.

Progress Update, October 2020

• Developed a draft map of vegetation height for Colombia that integrates GEDI lidar, multispectral, and SAR imagery.
• Acquired forest inventory data from the Colombian Amazon, the Choco, and the east Andes to validate canopy structure maps and canopy structure-tree diversity models.
• Identified lidar structure metrics that predict tree diversity in the Choco region of Colombia and used these to develop a wall to wall map of tree diversity for the Choco. A manuscript is in revision for resubmission to Environmental Research Letters.
• Developed a partnership with the BioModelos team to use canopy height as a predictor variable in primate distribution models, with other species to follow.

Project Update, April 2021

• Presented a map of canopy height for Colombia at the AGU 2020 Fall Meeting using Landsat, Sentinel-1, and ALOS-Palsar SAR imagery to extend version 1 GEDI canopy height measurements wall-to-wall. Variable performance from ecoregion to ecoregion highlighted the need for regionally tuned models. Map was recently updated with Sentinel-1 and Sentinel-2 texture metrics, reducing errors considerably.
• Developed approach to map gradients of canopy structure as a continuous function of environmental factors and presented it at the AGU 2020 Fall Meeting. Model outputs can aid in identifying places in the landscape where canopy structure varies gradually or rapidly and why, providing an indicator of habitat heterogeneity and its drivers.
• Developing tools in Google Earth Engine and R to facilitate access to these datasets and have been in contact with partners in Colombia (Humboldt Institute, SINCHI) about potential applications for canopy structure data in environmental decision making.
• A manuscript, Mapping tree diversity in the tropical forest region of Chocó-Colombia, is in the final round of review at Environmental Research Letters. It identifies lidar structure metrics as among the most important variables for predicting tree species richness; and approach for mapping tree richness across the ecoregion is included.

Enrique Montes/University of South Florida, Tampa
Laying the Foundations of the Pole-To-Pole Marine Biodiversity Observation Network (MBON) of the Americas
16-GEO16-0073

This project is linked to the Thematic BON: Marine Biodiversity Observation Network (MBON)
More information: marinebon.github.io

Background

The Marine Biodiversity Observation Network (MBON) is a Thematic BON established under the Group on Earth Observations Biodiversity Observation Network (GEO BON). The MBON is also the highest priority program identified by the GEO AmeriGEOSS initiative under its Ecosystems and Biodiversity Sustainability Societal Benefit Area. The vision of the MBON is a network of regional observation systems that collaborate and share information, and that work jointly to understand marine biodiversity, its geographic distribution, and how it changes through time.
The goal of the MBON is to provide the knowledge needed to promote ecosystem conservation, sustainability, and good management practices. To accomplish this, MBON connects and provides support to existing national and regional marine observation programs and provides the framework for a community of practice. The framework is a partnership between national and regional programs and the Global Ocean Observing System (GOOS/IOC-UNESCO), the Ocean Biogeographic Information System (OBIS/IOC-UNESCO), and GEO BON. The Pole-to-Pole MBON of the Americas (P2P MBON) is a first step toward establishing a regional collaborative network that supports groups in the American continent.

The proposed work seeks to implement a demonstration P2P MBON to address UN Sustainable Development Goal 14 (SDG 14: “Life Below Water”). This work will contribute specifically to establish a series of Foundational Activities under AmeriGEOSS. P2P MBON seeks to develop a community of practice for marine biodiversity monitoring. The geographic scope are waters around the Americas from the Arctic to Antarctica, in the Pacific and Atlantic Oceans. The project builds on and links past and current research and monitoring initiatives. These include the Sub-Commission for the Caribbean and Adjacent Regions of the Intergovernmental Oceanographic Commission Global Ocean Observing System (IOCARIBE-GOOS), the Southeast Pacific Data and Information Network in Support of Integrated Coastal Area Management (SPINCAM), the Caribbean Coastal Marine Productivity Program (CARICOMP), the South American Research Group on Coastal Ecosystems (SARCE), and three US MBON demonstration projects (National Marine Sanctuaries, Santa Barbara Channel, and Alaska). Linking these efforts will help engage other activities and build a broader community of practice. The objective will focus on three core activities:

• Link users into the community of practice, to ensure products are responsive to societal needs, with particular emphasis on SDG 14.
• Develop tools for the standardization and sharing of marine biodiversity observations, as well as synthesis with environmental data, derived from satellite and in situ observation. This includes data management and communication capabilities according to international standards (e.g. DarwinCore and EventCore).
• Build scientific capacity throughout the Americas with training workshops and information dissemination campaigns oriented around use of tools, portals and data. These efforts will be co-designed with OBIS, following the International Oceanographic Data and Information Exchange (IODE) Data Carpentry approach.
• Distribute seascape and related MBON products that use ocean color and infrared Earth observing satellite data, via the GEONETCast Americas satellite network and other relevant means.

The efforts proposed here complement proposals targeting the GEO BON program element (GEO Work Program 3.1) with focus on the development of Essential Biodiversity Variables and BON in a Box. These will contribute to developing the AmeriGEOSS Biodiversity and Ecosystem Sustainability Societal Benefit Area.

Progress Update, March 2019

• The first Biodiversity Workshop: from the Sea to the Cloud was held in São Sebastião, Brazil, Aug 6-10, 2018. This activity gathered 38 participants from 10 countries in the Americas to set the foundations and marching orders of the MBON Pole to Pole network.
• Field data collection protocols for macro-invertebrates of sandy beaches and rocky shores have been developed and are currently being used in biodiversity surveys across the region. Protocols are available at the MBON Pole to Pole site.
• Biodiversity surveys have been completed between August 2018 and January 2019 in Argentina, Brazil, Chile, Colombia, Ecuador and Galapagos Islands, and Mexico using standard sampling protocols. Data from these surveys will be publicly available through OBIS and GBIF.
• In collaboration with researchers of the Research Centre in Biodiversity and Genetic Resources of Portugal (CIBIO-UP), 36 biomimetic temperature loggers (Robolimpets) are being deployed at 18 sites over nine countries where biodiversity surveys are being conducted. These temperature loggers register temperature exposure of sessile macro-invertebrates of rocky intertidal zones. Sensors were donated to the project by the manufacturer (ElectricBlue).
• The 2nd Biodiversity Workshop: from the Sea to the Cloud will be carried out in Puerto Morelos, Mexico, on April 1-6 this year to advance network objectives and the implementation of the MBON Pole to Pole network. Over 30 participants from 11 countries in the Americas are confirmed.

Progress Update, October 2019

• The 2nd Biodiversity Workshop: from the Sea to the Cloud was held in Puerto Morelos (Mexico; April 1-6 this year) to finalize biodiversity survey protocols for rocky intertidal zones and sandy beaches, and advance approaches for the integration of in situ observations with satellite records. Interesting questions for which such integration is particularly important include:
  1. Which seascape classes are most dominant within a particular area? How do seascapes change in space and time? …and why?
  2. What factors determine the range of seascape categories found at a particular location? Do more categories reflect more dynamic oceanographic conditions?
• The program is now contributing biodiversity surveys data across the Americas to the Ocean Biogeographic Information System (OBIS), and has added over 14,000 new records. Survey data can be accessed here: rocky shores and sandy beaches.
• The MBON Pole to Pole contributed its vision on global biodiversity monitoring to the OceanObs’19 Special Issue on Frontiers of Marine Science: Global Observational Needs and Resources for Marine Biodiversity; Front. Mar. Sci., 23 July 2019, https://doi.org/10.3389/fmars.2019.00367.
• Biodiversity monitoring protocols and data management tools will be shared via the Ocean Best Practices System of the Intergovernmental Oceanographic Commission (IOC) of UNESCO.
• The 3rd Biodiversity Workshop: from the Sea to the Cloud is in preparation for the spring of 2020.

Progress Update, April 2020

• The 3rd Marine Biodiversity Workshop: from the Sea to the Cloud has been postponed to a TBD date. This is a joint activity of the MBON Pole to Pole and the C-GRASS SCOR WG led by Emmett Duffy to coordinate biodiversity monitoring in intertidal rocky shores, sandy beaches and seagrass beds of the Americas, and facilitate collaboration between diverse coastal ecology communities of practice.
• We are finalizing a study on minimum sampling effort needed during field surveys to characterize species richness and community structure of macro-invertebrates in rocky shore habitats of the Americas, from local and regional scale. Results will be submitted to the MBON Special Issue on The Oceanography Magazine in summer.
• The team is testing machine learning techniques to collect image-based (photo-quadrats) taxonomic records of macro-invertebrates and macro-algae from surveys conducted in rocky shores of Argentina, USA, Colombia, Chile and Ecuador (Galapagos I.). The work assesses the effectiveness of image-based biodiversity surveys to expand the scope of monitoring programs.
• The project is collecting photo-quadrat imagery at monitoring sites to derive species and functional group presence/absence, % cover, community structure, etc. The idea is that in the long run we can implement this technology to survey rocky shores faster and more efficiently, and more sites.
• Surface ocean habitat diversity maps are being produced for regional domains (e.g. Large Marine Ecosystems, Exclusive Economic Zones, Pacific and Atlantic basins) at 5 km pixel resolution. The work, which is in collaboration with Maria Kavanaugh’s NASA project “Dynamic Seascapes to Support a Biogeographic Framework for a Global Marine BON” (above) includes maps of seascape alpha richness, beta diversity, Shannon-Wiener H index, and others. These products seek to support the development of EBVs for Ecosystem Structure for the marine realm.
• In situ temperature records from rocky shores across the Americas are being coupled to high-resolution SST observations collected by ECOSTRESS aboard the ISS to develop space-based thermal habitat maps in the coastal zone. This work is being carried out in collaboration with one of ECOSTRESS Early Adopter programs led by Dan Otis (USF).

Progress Update, October 2020

• Completed a case study applying machine learning techniques to extract functional group annotations automatically from imagery collected at surveyed sites in four countries: Argentina, Colombia, Ecuador (Galapagos I.) and the NE USA. The goal is to implement these tools more broadly across the region to detect changes in percent cover of key functional groups like sessile macro-invertebrates and macro-algae. Results were presented at the GEO BON OSC in July.
• Building a partnership with the Multi-Agency Rocky Intertidal Network (MARINe) managed by University of California Santa Cruz. MARINe conducts long-term monitoring and biodiversity surveys in over 160 rocky shore sites along the Pacific coast from Mexico through Alaska. The Pole to Pole site now includes several MARINe stations on the interactive map and will continue to add more.

Project Update, April 2021

• The program has engaged in the development of a thematic global biodiversity monitoring network for the rocky intertidal led by the MBON Secretariat of the Air Centre (Azores, Portugal).
• The MBON Pole to Pole website now allows tracking growth of OBIS records over time at participating monitoring sites as well as their spatial distribution per taxonomic group. An example is available here.
• The team recently released this tutorial to guide network participants through the process of transforming original datasets to Darwin Core with new tools developed for facilitating data publishing in OBIS and GBIF. The tutorial has been published in the Ocean Best Practice System of the IOC-UNESCO.
• A manuscript on machine learning applications for monitoring functional groups of the rocky intertidal using photo-quadrat imagery and CoralNet software (UCSD) will be submitted this week to the MBON Special Issue of Frontiers in Marine Science. The study includes data from Argentina, Colombia, Ecuador and the USA.
• A second manuscript on sampling effort optimization for biodiversity surveys submitted to the MBON Special Issue of Oceanography is under review. This study includes contributions from the USA, Colombia, Ecuador (Galapagos Islands), Brazil, Argentina, Chile, and Antarctica.
• MBON Pole to Pole is contributing to coordination efforts led by Victor Gutierrez-Velez (NASA A.50 awardee) to better integrate biodiversity monitoring across land and marine ecosystems in Colombia under the Colombia BON and INVEMAR (Marine and Coastal Research Institute “Jose Benito Vives de Andreis”). The goal is to identify “super users” of biodiversity data for decision-making and develop common best practices and workflows for the delivery of data and information to Colombian government agencies reporting to national and international biodiversity frameworks.
• On January 6th, the program carried out a virtual demonstration to use R code developed by the project to extract and analyze satellite data relevant to marine biodiversity monitoring at the Coastal Ocean Environment Summer School in Ghana (COESSING).