Predicting global biodiversity patterns in Costa Rica with ecosystem modeling on AWS

Though it accounts for only 0.03% of the Earth’s surface, Costa Rica is home to about 6% of the world’s biodiversity. Leveraging Costa Rica’s rich biodiversity and advancements in ecosystem modeling, leaders in sustainability are making geospatial data and analysis of Costa Rican landscapes publicly accessible. The data and analysis help people understand how different species influence their ecosystems, while also informing important policy decisions that support the natural world.

As part of the Amazon Sustainability Data Initiative (ASDI)—a program that seeks to accelerate sustainability research and innovation by minimizing the cost and time required to acquire and analyze large sustainability datasets—Amazon Web Services (AWS) invited Rafael Monge Vargas, director of the National Center of GeoEnvironmental Information (CENIGA) at the Costa Rica’s Ministry of Environment and Energy (MINAE), to share how his team is helping advance conservation and economic development in Costa Rica and how they utilize ASDI and AWS to support these efforts.

Integrating Earth observations and ecosystem service modeling

CENIGA is a technical intelligence unit that leverages mapping and modeling tools to integrate, standardize, and communicate Costa Rica’s environmental data. It convenes with governments and scientific institutions worldwide to develop methods that identify biodiversity patterns, helping leaders make better decisions for sustainable development in Costa Rica and around the world. CENIGA partnered with Stanford University’s Natural Capital Project, the Central Bank of Costa Rica (BCCR), and PRIAS lab to leverage ecosystem modeling to help predict biodiversity patterns.

The process begins with Earth observations (EO), or data about Earth’s physical, chemical, and biological systems. Examples of this EO data include recording rainwater trends, monitoring temperature, and mapping forest cover. Satellite technology makes it possible to capture this information via remote sensing in near real-time, providing a more comprehensive picture than with data collection on the ground. Currently, there are more than 60 environmental satellites surveying the Earth from space; however, standalone EO cannot fully assess ecosystem properties like biodiversity, its value, and how ecosystems change over time.

This is where ecosystem service modeling comes in. Ecosystem service modeling uses geospatial data to predict patterns about a real ecosystem. To make sense of geospatial data for real-world applications, CENIGA and its collaborators integrate EO of ecosystem-level essential biodiversity variables (EBVs) with ecosystem service modeling, which produces representations of relationships between variables—like tree cover, vegetation cover, and bare ground. These models are used to make predictions about the real ecosystem. Pairing robust EO datasets with modeling leads to more accurate, accessible, and relevant predictions of how ecosystems behave and how they might evolve. The work starts with accessing the EO datasets available on AWS, and then using that data to conduct the ecosystem service modeling with cloud infrastructure built on AWS.

Predicting ecosystem services through open datasets

CENIGA and its collaborators create, test, and improve models that use EBVs to predict ecosystem services and assess the value of those services. Below are examples of models currently adopted by the groups for pollination, tourism, and sediment retention.

Pollination: Assessing pollinator availability and coffee pollination in relation to floral and nesting resources

Many crops depend on pollinators—such as birds, bees, and butterflies—and suffer 5–95% in yield reductions without adequate pollination. The partnering organizations found that coffee production is up to 30% higher when its flowers are pollinated by bees; however, coffee flowers alone are not enough to sustain healthy bee populations, as they need nectar and pollen from flowers year-round. When there are other types of flowering plants nearby, it boosts the number of bees in the environment and influences their availability to pollinate crops. The same goes for nesting resources. Honeybees nest in hives, but many wild bees nest in the ground or in stem cavities. Farms tend to be too disruptive to allow bees to nest on them, so making sure appropriate nesting resources exist close to farms is important for maintaining pollinator populations.

Tourism: Leveraging social media data to predict visitation rates based on different ecosystem types and bird diversity

Tourism is measured in three ways: 1) the number of photos taken by unique users per day from social media (Flickr), 2) the number of bird checklists uploaded from a specific location (eBird), and 3) the number of visitors per day to a national park. Together, these data points capture the popularity of a destination and serve as a proxy of its tourism value. The groups predicted that greater biodiversity would correlate to higher tourism rates in Costa Rica, as visitors tend to gravitate to parks where they can see the most different types of wildlife. The hypothesis proved true across all three measures. In particular, threatened and endemic bird diversity was most strongly correlated with visitation of all the types of diversity, including all birds and all vertebrates.

Sediment retention: Parameterizing soil erosion modeling through integrating remotely sensed vegetation indices

Soil erosion occurs when water or wind causes soil to deteriorate and run off, potentially affecting agricultural land productivity and downstream water quality. Vegetation creates sediment retention and traps eroded soil on land before it enters a stream, helping the water stay clean. Controlling sediment affects the quality of drinking water, dams and reservoirs, clarity for recreation, fish populations and fishing, and irrigation. As such, assessing vegetation indices—which help monitor vegetation presence and its health—is an integral part of natural hazard planning and preventing climate risks.

Leveraging ASDI and the AWS Cloud to enhance ecosystem services

CENIGA built its infrastructure on the AWS Cloud with AWS Promotional Credit from the Earth Observation Cloud Credit Program, a collaboration between ASDI and the Group on Earth Observations (GEO). CENIGA uses AWS Simple Storage Service (Amazon S3) and Amazon Elastic Compute Cloud (Amazon EC2) to host, process, and analyze large geospatial datasets, which are hosted through ASDI at no cost for anyone in the world to access. Additionally, CENIGA has standardized its key offerings including its map viewer, web map service, metadata manager, and web feature service protocols. To date, the restructured infrastructure contributes to a 90% cost optimization.

Ecosystem service modeling in the real world

In 2019, CENIGA published over 40 spatial data layers for ecosystem level EBVs (e-EBVs) like tree cover, vegetation cover, and bare ground; species level EBVs (s-EBVs) like climate-based species distribution models (SDMs) for individual pollinator species and more; and ecosystem services like s-EBV improved tourism, s-EBV improved pollination, e-EBV improved carbon storage (above-ground biomass), and e-EBV improved sediment retention. Each product includes detailed metadata in Spanish and is available in Costa Rica’s National System of Environmental Information portal.

In subsequent years, CENIGA improved on the models’ performance on AWS and worked to make the modeling accessible to technical staff. The modeling results are used by leaders to develop Costa Rica’s National Adaptation Plan, which identifies medium- and long-term adaptation needs, and implements strategies to address those needs. The goal is to identify priority areas for four nature-based actions: protection, sustainable management, restoration, and urban greening. A recent research article examines how biodiversity and infrastructure interact to drive tourism within Costa Rica.

Understanding the relationship between biodiversity and ecosystem services provides valuable insight in support of decisions related to conserving biodiversity, climate adaptation, and sustainable development. CENIGA organizes technical exchanges with local experts and works with key stakeholders—like Central Bank and National System of Conservation Areas—to leverage the modeling results and optimize the benefits of biodiversity across Costa Rica, and other parts of the world.

Learn more about the Amazon Sustainability Data Initiative.

AWS Public Sector Blog