How Citizen Science can help SIDS reach SDGs
With the detailed metrics of the Sustainable Development Goals (SDGs), many small communities lack data to measure their baseline status and progress. Small Island Developing States (SIDS) occupy a unique space among UN member states, geographically isolated, small in population, and ecologically unique. With regard to the environment, many SIDS lack the resources to collect reliable data using traditional methods. This project explores the potential of open source, low cost, and automatic technologies to assist in this effort. Here we test options for three SDG indicators (airborne particulate matter, ocean acidity, and green coverage) and one other (seafloor coverage), to track SDG progress in Aruba using citizen science. Preliminary designs for replicable sensor stations, protocols for GIS mapping, and public data that is some of the first for the country have resulted. With further development, this approach can help communities around the world, to track SDG progress and better understand their natural environments.
The Sustainable Development Goals (SDGs) are an ambitious set of goals that look at 17 broad areas of development, each of which is defined by several specific targets, and in turn quantified by one or more indicators. At this basic level, the indicators are the way we can measure our progress towards the SDGs and realize “peace and prosperity for people and the planet”.
The official Aruba SDG Roadmap groups the goals into nine “accelerators” divided among five “pillars” each of which has a number of associated SDGs. The planet pillar contains seven different goals, connected to one accelerator called Natural Resource Management. In 2018, the national SDG commission published a baseline measurement for Aruba’s progress on the SDGs which includes only one of the 45 targets that fall under the planet pillar, and considers two others to be possible to measure in the future.
After conversations with members of various government departments and other relevant actors, it was confirmed that, indeed, there is no investment towards collecting this data. This leaves 93% of the environment related SDG targets for Aruba with no official plan. At the same time, people and countries around the globe are increasingly recognizing the severity of human impacts on the planet, and governments in 26 countries have officially declared a state of emergency over climate change (9 countries nationally, 1333 jurisdictions in 26 countries covering 803 million people).
In Aruba, there is also increasing awareness and concern over degrading environmental health among the community. Here we present an active response to these concerns, facilitated by a small team working out of a local makerspace called Brenchie’s Lab, to leverage open source techniques to facilitate the independent collection of some of the environmental indicators omitted from the country’s national SDG plan. Specifically, the approach focuses on indicators that can be measured physically, so the results not only measure SDG progress but also contribute to a better understanding of the changing state of Aruba’s environment. The goal is to openly develop replicable and affordable methods to monitor environmental health that are not labor intensive.
This situation is not unique to Aruba. Many other small islands, less developed countries, and rural areas face similar challenges with unavailable data and resources, due to small size or limited funding. This is especially unfortunate considering that many of these locations face the most risk from the impacts of climate change. This project seeks to facilitate access for communities to collect environmental data independently, with available resources, so that they might initiate the research process and bring issues to light to be included in the global conversation.
In the interest of shedding some light on the mindspace and perspective from which these data collection tools are being developed, we can place it within a context that motivates the need for this type of research. Issues of lacking environmental research, specifically in areas facing significant and existential threats from climate change and pollution, highlight a clear case for the empowerment of small, remote, or low-resource actors to establish a basic scientific understanding of their environment.
Environmental justice is based on a set of principles that demand equitable access to the benefits of a healthy environment, and a universal responsibility to maintain such an environment, especially when it concerns underrepresented communities. It is through this lens that we can analyze the role that citizen science, i.e. science carried out by members of the public, can have in how SIDS pursue the environmental SDGs as well as climate change mitigation and adaptation, because at the core this is a clear case of environmental injustice.
While SIDS have a relatively small impact on climate change, they will likely carry a disproportionate weight of its impacts. Already we have seen several islands, including just recently Puerto Rico, Barbuda, Grenada, Dominica, The Bahamas, Sint Maarten, and British Virgin Islands, facing severe impacts from hurricanes, in some cases making entire islands unlivable. While these impacts directly affect relatively small groups of people, they are devastating to those groups, especially on islands where there is no option for relocation. This same problem also faces many pacific islanders, who face typhoons and sea level rise, wondering “where do we go when our islands disappear due to the impacts of climate change?”
On the political side, many SIDS are not even part of the discourse on climate change, even while some islands including Micronesia build up grassroots initiatives to raise their local issues to be included in the global conversation. Additionally, many SIDS lack tools to monitor and predict the local impacts of climate change, especially smaller islands without resources, universities, or environmental protection agencies to research these issues. Even islands that are legally part of regions like the European Union, United States or the Commonwealth are often ignored or used as case studies for disconnected research that might not even make it back to the islands. Even when this research is shared, it generally lacks follow-up and connection to other local work, and is done without the cultural understanding necessary to be able to understand and communicate the issues in the local context.
Citizen science is a tool to democratize the capacity to understand one’s own environment by providing the tools to collect and share the data, and talk about it. In a sense this is about “participatory justice — speaking for ourselves, or a seat at the table” (Schlosberg, 2013). This brings islands into the conversation of environmental management and the impacts and causes of climate change on the ground. Expanding access to the tools to monitor the environment depoliticizes the resulting data. “By altering power dynamics in material, literary, and social technologies used for scientific research, [citizen science] enables citizens to question expert knowledge production through critical making tactics, and creates opportunities to generate credible public science” (Wylie et. al, 2014).
Through this process, this project aims to raise the role of citizens from mere subjects, or victims of climate catastrophes, to actors in adaptation and mitigation of climate change and its impacts.
Our first step in identifying potential opportunities for citizen science to monitor SDG progress was to go through the list of indicators to identify those that could be measured physically. In Aruba, none of these were being measured at the national level, and there were no plans to do so, making any of them suitable for citizen science efforts.
The full list of physical indicators includes:
- 6.3.2 Proportion of bodies of water with good ambient water quality
- 6.6.1 Change in the extent of water-related ecosystems over time
- 11.3.1 Ratio of land consumption rate to population growth rate
- 11.6.2 Annual mean levels of fine particulate matter (e.g.PM2.5 and PM10) in cities (population weighted)
- 14.1.1 Index of coastal eutrophication and floating plastic debris density
- 14.3.1 Average marine acidity (pH) measured at agreed suite of representative sampling stations
- 15.1.1 Forest area as a proportion of total land area
- 15.3.1 Proportion of land that is degraded over total land area
- 15.4.2 Mountain Green Cover Index
The list was then narrowed down to a selection that could potentially be measured using low cost, open source tools, and that could help initiate monitoring of several neglected SDGs within the scope of a one year exploratory project.
The selected indicators include:
- 11.6.2 Annual mean level of particulate matter (PM2.5 & PM10)
- 14.3.1 Average marine acidity (pH)
- 15.1.1 Forest as a proportion of total land area
In the context of Aruba, a fourth parameter was added to the list as an indicator of marine biodiversity, a parameter particularly relevant to SIDS and with global impacts: seafloor coverage. Between them, coral reefs and seagrass make up two of the most important marine ecosystems on the planet. With the high levels of biodiversity and intricate and robust structures of coral reefs, they provide habitat for many species and support surrounding ecosystems including mangroves, seagrass, and fish populations. Seagrass similarly provide habitat and structure to collect nutrients to support a a variety of species and act as an important global carbon sink. For people, including most tropical SIDS, coral reefs provide additional value protecting the coast from storm surges and erosion, supporting livelihoods based on fishing and tourism, and enduring natural heritage.
Coral reef coverage is also linked to many of the environmental indicators as an important water-related ecosystem with high biodiversity, and sensitivity to water quality due to pollution, and changes in nutrients, temperature, light, and pH.
The four selected indicators span the two most basic categories of environment, land and sea, each with one parameter that can be continuously monitored using sensors, and one that can be mapped periodically.
Air pollution, specifically the concentration of airborne particulates, has been recognized by the WHO as one of the major risk factors contributing to noncommunicable diseases. Two specific measures, PM2.5 and PM10, correspond to the concentration of particles in the air below a certain size in micrograms per cubic meter (μg/m^3), below 2.5μm and 10μm in diameter, respectively. Both of these parameters have internationally recognized target exposure levels, and a third parameter, PM1.0, is now being researched as an important measure of air quality.
While particulate matter itself is not a measure of carbon dioxide, it is often a side effect of the same carbon emitting activities that drive climate change, particularly combustion and deforestation. But unlike carbon dioxide levels in the atmosphere, which have a global effect, particulate levels in the air can have very localized effects, and are closely related to local behavior. At the same time it has a large and direct impact on human health and life expectancy, making it easier for people to relate to and organize around.
In Aruba we have seen an increasing concern about air quality, especially as it relates to waste management and energy production on the island, and with dust produced by off-road vehicles. This discussion is often heavily politicized, and being able to discuss these issues with data and to monitor the success of potential interventions allows it to be de-politicized and made more transparent.
Ocean acidification (OA) is directly related to climate change and affects the livelihoods of many Arubans. OA is a measurement of aqueous hydrogen ions in the ocean, which result from reactions between water and carbon dioxide (CO₂), measured as pH. As atmospheric carbon dioxide levels increase, more dissolves into the ocean. This reacts with water to form carbonic acid, which releases hydrogen ions that in turn combine with carbonate in the ocean to form bicarbonate. Many species of coral and other shelled animals need carbonate to form their skeletons, which are made of calcium carbonate. The increased acidity can even lead to their skeletons dissolving, and less carbon ending up being stored in sediments and reefs in the long term.
On the other hand, some corals and shelled sea life are able to make use of bicarbonate to form their shells, and increased levels of dissolved CO₂ can lead to increased growth of some photosynthetic marine life such as seagrass and phytoplankton.
Further complicating the situation, as ocean temperatures rise, the solubility of carbon dioxide decreases, and ocean currents change, so the ability of the ocean to continue absorbing atmospheric CO₂ goes down and at a certain point the ocean will begin releasing it back into the atmosphere.
With all these changes happening at such a rapid pace, the ocean is becoming uninhabitable for coral reefs and the rich biodiversity that these ecosystems support. In Aruba, this will likely result in declining fish populations, and reduced capacity of the ecosystem to maintain the island’s famous white sand beaches, which will decline along with the fish and coral populations on which they depend.
Green coverage (GC) is a measure of how many plants there are on land, which is also directly related to climate change. Decreasing levels of vegetation directly accelerates climate change, as the capacity of the planet to capture carbon out of the air and trap it in the greenery and soil is reduced. When areas are deforested they also release previously stored carbon compounding the effect, and made even worse when increased temperatures and extreme weather events lead to droughts and physical destruction of more vegetation.
In addition to the global effects on the climate, there are a number of local benefits to higher levels of green coverage. Tree and plant cover help to decrease temperature in the surrounding area, improve air quality, and even benefit mental health. Decreasing levels of green coverage are one of the most visible environmental concerns, and can help motivate the community to take very direct and effective action by planting trees and maintaining local green spaces.
Coral reef and seagrass coverage are also directly related to climate change, with increasing temperature and carbon dioxide leading to coral bleaching events, decreased growth, and die-off, and seagrass populations globally form a carbon sink three times that of the Amazon rainforest. Unhealthy or dead reefs also have negative impacts on connected ecosystems and species, including seagrass, mangroves, and people. For many SIDS both coral reef and seagrass meadows are a source of livelihood, driving tourism and fishing industries. A healthy reef ecosystem supports perhaps the greatest density of biodiversity of any ecosystem on the planet.
And while these ecosystems form the backbone of Caribbean life, being underwater they are often overlooked. This might explain why they are left out of the SDG indicators for life under water. Globally, marine ecosystems are not well monitored. Though there are growing efforts to track changes using satellite imagery and recognize threats of coral bleaching events in near real time, they are not yet able to reach the small scale resolution of most SIDS.
The environmental justice framework aligns with the potential of open source movements and citizen science, which allow people to collect the data they need to fill the gaps they see, to trust the data, to have access to data, and to be more involved in the conversation.
The materials and technologies used in this project, which are typically low-cost tools that can be used by non-experts, are supported by literary and social technologies, open source hardware and software licenses, as well as public access websites, listservs, and face-to-face meetups that together form alternative institutions for civically engaging science.
These platforms supported the construction of two sensor stations to continuously monitor particulate matter and ocean pH levels, the calculation and mapping of Aruba’s green coverage and shallow seafloor coverage, and helped initiate a longer term plan to have automated and detailed oceanic maps based on underwater drone photographs combined with artificial intelligence for image classification. The results of this initial development and pilot measurement for each indicator are described below. Section headers can be clicked to access details and protocols hosted on the Brenchie’s Lab Wiki.
In order to measure airborne particulates, an affordable sensor (PMS5003) was connected to a WiFi-enabled Arduino based microcontroller to continuously measure air quality and transmit readings to an online database. For the pilot, four systems have been built and installed at locations around the island, and a website has been drafted to display the data, both in numerical format and in an animation.
The animation was designed through an open workshop, with community members of different interests and skillsets coming together to brainstorm and prototype a more intuitive way to communicate data. All of the collected data is available for download, with an example selection of data for the first month at one location graphed here.
Despite frequent burning events at the open landfill located near Pilot Site 1, particulate matter levels within the first month of monitoring were measured to fall within the Good to Moderate range of the EPA Air Quality Index table (Figure 1). In the future we aim to add a second sensor to the monitoring station to have multiple measurements for higher reliability, as is the recommendation from the World Air Quality Index Project.
Another station installed closer to the national park was affected by connectivity issues and so far has not been able to report much data, and a third in the mainstreet of the capital city was positioned in front of a plastic recycling workshop, giving measurements that are frequently high but localized, and not always indicative of the air quality in the city overall. A fourth station was recently set up at the library in the town at the south end of the island, and is just starting to report data.
Ocean acidification is being measured using relatively affordable sensors from Atlas Scientific, connected to an Arduino based microcontroller to collect pH, temperature, and dissolved oxygen readings every 10 seconds, and similarly report to an online database through either WiFi or GSM connection. One such monitoring station has been constructed and programmed, is currently being tested at the lab, and will soon be installed for a field pilot test at a pier owned by a local diving company.
Data that has been collected so far is entirely bench-top, but the sensor station is actively reporting to an online database with a web interface to publish the live readings.
In order to calculate green coverage, satellite imagery was used to calculate the Normalized Difference Vegetation Index (NDVI), which can be used to assess how much photosynthesis is occurring in different areas, indicating the presence of plants. Using scripts written in Google Earth Engine, and utilizing Landsat 8 satellite imagery, monthly calculations were done for the island of Aruba to determine which percentage of the land surface has an NDVI above several different thresholds.
A threshold of 0.4 is a standard value for identifying forests in temperate regions, but thresholds of 0.3 and 0.2 were also included as Aruba has a semi-arid tropical desert climate, which means a lot of its vegetation grows more slowly, and exhibits lower levels of photosynthesis. The graph above shows the green coverage at these three thresholds over an 18 month period, plotted with temperature and rainfall data for reference. Unfortunately, national rainfall and temperature data were not available beyond April 2019.
From a visual inspection of the graph, it seems that green coverage varies seasonally, and correlates well with rainfall, indicating a potentially high sensitivity to periods of drought.
For the ocean maps we took a two pronged approach, on one end using satellite imagery to try to map shallow reef and seagrass systems, and on the other end working to use automated underwater photography coupled with a citizen-trained machine learning algorithm to classify images as different types of seafloor cover.
Using satellite data provided by open data sources and regional actors combined with the open source geographical information system QGIS, it is possible to train a model to automatically classify underwater seafloor cover in shallow areas.
A process to create reef maps based on satellite imagery was explored and documented in order to understand the feasibility of creating a preliminary map. The technique is sensitive to image resolution and training of the model, so the higher quality image you can obtain, and the more you already know about the extent of local seagrass and reef ecosystem extents, the better the resulting map will be. We tested a combination of image preprocessing and classification using the open source QGIS software and openly accessible datasets from Landsat, and combined it with direct observations of underwater seafloor cover. An initial map is currently still under construction, but aims to serve only as a starting point. With feedback, ground truthing and repeating of the process with higher resolution imagery, this process can contribute support the production of more accurate maps and better understanding of temporal changes.
Drone-based mapping has been carried out by collecting images using a Trident underwater remotely operated vehicle (ROV). After using the ROV to explore around Aruba, we modified it by attaching a GoPro to face downwards to collect images of the seafloor as the drone drives forward. We also obtained an extended tether to be able to drive it up to 100m away, and a kayak to further increase the range.
So far we have collected images in three areas around Aruba, and have developed a plan for crowdsourcing the collection of images around the rest of the island. The drone and the Kayak are being made available for citizen scientists to borrow, along with a protocol for operating the drone to cover a certain area, and a map dividing the shallow coastal sea into manageable portions.
For post-processing of the images there are several steps before we can create a map of reef coverage. The first is to identify what is in the collected images. To do this we have developed an AI model through an open workshop using Tensorflow and Keras, which can be trained to classify images as containing sand, seagrass, healthy coral, dead coral, or rock. We ran an initial test to train the model on a set of 100 classified images, and gave it the challenge of sorting 50 unclassified images into either sand or grass with good results.
The training of this model is going to be done using crowdsourced image classification through the Zooniverse platform, which allows anybody to click through the images and classify them. In addition to helping us train the model, this will also allow us to involve a broader community in the process of mapping reefs in Aruba, as well as introduce them to more opportunities on the Zooniverse platform to engage in citizen science research projects.
Currently the connected Zooniverse project is being revised to start beta testing, and is expected to launch in February 2020.
This first effort into measuring environmental SDG indicators on a small island has made impressive progress, not only by starting to collect data, but by starting the conversation on the island and engaging people in the process. Using open source tools and citizen science approaches, the project has developed hardware and software protocols and collected sample data showing promise for the approach to continue monitoring these indicators going forward. At the same time, many improvements can be made to the devices and techniques, as well as in spreading the effort to both share the information with the local population, and share the approach with other islands.
Specific questions and plans for improvements to the project in the coming year are detailed further for each parameter, including both technical challenges and ideas for other activities.
Looking forward to collecting data over a full year, seasonal changes and wind patterns are going to be of particular interest. In Aruba, the Caribbean hurricane season is accompanied by low winds, and seasons are defined by sudden heavy rains and dry periods, both of which can have significant impacts on airborne particulate matter.
Technically, it will also be useful to confirm the data being collected by placing redundant sensors at monitoring locations. Once the data collection process is reliable, it can also be shared on other online platforms for broader access.
There is a great opportunity to couple the monitoring program with an exploration of interventions, together with the community, to see how air quality can be improved. Examples of technologies that could achieve this range from simple steps like adding particulate filters and catalytic converters to vehicles, to social efforts for moving towards cleaner energy such as solar and wind, to more natural options such as tree planting campaigns, or even designing and building experimental biofilter walls to test their impact.
This year’s plan includes sessions with public officials to be able to use the air quality data for official purposes, and to support the expansion of the sensor network, as air quality can vary widely across different areas.
With the ocean acidity monitor, the first step is to get it installed to start collecting preliminary measurements at the pilot location. Once any issues are worked out, a second version will be built to be installed further away from shore, either as a free-floating system or attached to an existing buoy, ideally without setting down an anchor or damaging the ocean environment. Such a monitor will also require a different system for communication and power, using low-power technology to send data and running off solar power. That system can then be replicated and installed around different beaches and coastal locations to collect a representative suite of measurements. This can be done in collaboration with local officials, and result in the ongoing collection of reliable data to monitor SDG progress.
Initial green coverage calculations provide the basic data of what percentage of Aruba is green during different time periods, but the methods also establish the foundation for other calculations and maps to be produced. This includes more detailed landcover mapping, to track things such as deforestation, desertification, urbanization, changes to wetland extent, and even tracking of sand accumulation and erosion of Aruba’s beaches. With enough mapping, a model can eventually be constructed to extrapolate from historical data to predict how landcover might change in the future under different circumstances, helping to inform how related SDGs might be pursued.
There are three different avenues of development going on around mapping the seafloor. The first is the underwater drone. Currently a manually controlled drone is being used to collect imagery, but an autonomous drone has been prototyped and needs some additional programming, tweaking, and testing in order to get it to the point of collecting images without needing any human control. For both drones, there is a pending challenge to connect collected images with GPS coordinates to enable more accurate mapping. The second avenue of development is the image recognition process. With a combination of programming and human feedback through Zooniverse, the AI model can be trained to improve its accuracy and learn to classify images better as time goes on. The last avenue is in satellite imagery mapping, which also benefits from other mapping efforts. As the drones collect data and as more observations are made to confirm the maps produced by this method, the new information can be fed back into the mapping algorithm to improve the maps it produces.
This not only improves resulting maps, but it builds local mapping capacity over time, and does it with the involvement of the broader community in the image sorting process, inviting more people to look at the state of our marine environment.
The traditional method for underwater mapping, using teams of divers, is both prohibitively expensive and takes a long time to complete. This is unfortunate when reef bleaching events can take place relatively fast and the impact of climate change is accelerating. The hope is for these approaches to be shared with other SIDS that might lack funding or academic resources to complete mapping through diving exercises, enabling more places to track changes in reef and seagrass ecosystems in their oceans.
In an initial exploration, within the space of a year we were able to develop prototypes to collect preliminary data contributing to three SDGs and a broader overview of environmental health in Aruba. This was made possible by a combination of community and technology. Local communities assist in research efforts and share resources and knowledge to help make studies possible. Global communities share a variety of experiences and details on projects that can be readily adapted. Technical advances in affordable environmental sensing, open source code libraries, and global datasets are making environmental monitoring increasingly cheap and accurate. In remote and varying locations such as SIDS, this combination empowers citizen scientists to start collecting data wherever they see a need.