NASA is hunting for plastic waste in the oceans from space: how EMIT and a new spectral library are changing the fight against marine litter
At the end of 2025, a team of scientists demonstrated for the first time that concentrations of plastic waste on land can be spotted from the International Space Station using NASA's EMIT instrument. Although this sensor system was originally developed to measure mineral dust from deserts, it surprisingly proved capable of recognizing the "fingerprint" of plastic in landfills, agricultural areas, and in large plastic structures like greenhouses. This opens a new chapter in the idea of applying the same approach one day to monitoring waste in the sea.
Simultaneously, a group of researchers led by young marine scientist Ashley Ohall published an open spectral library called MADLib – a database of nearly 25,000 precisely measured "molecular fingerprints" of different types of marine debris. In combination with EMIT and other hyperspectral satellites, this database could become a key link in the development of algorithms that will track the movement of plastic waste across the entire planet from space.
A plastic wave threatening the oceans
Plastic waste is now considered one of the greatest environmental threats to the world's oceans. According to estimates by the United Nations Environment Programme and a series of recent studies, between 8 and 11 million tons of plastic enter the sea every year – from large fishing nets and packaging to microscopic particles created by the breakdown of larger pieces. Most of this waste originates from land, washing away from landfills, inadequately managed municipal waste, or through rivers that carry plastic to the coasts and into the open ocean.
Plastic in the sea creates multiple problems. Marine species swallow it or become entangled in it, it damages fisheries and tourism, and plastic fragments enter the human food chain. Analyzes of global scenarios warn that, without changes in policies and consumption habits, the amount of plastic ending up in the oceans each year could nearly triple by 2040. This means that, along with traditional waste reduction measures, significantly better monitoring of what is already floating in the sea is needed – at the level of the entire planet, and not just individual beaches or coves.
This is where space technology enters the scene. Satellites monitoring the Earth have been measuring temperature, ocean color, clouds, and greenhouse gases for decades. Now, the same infrastructure is being leveraged to map the "flows" of marine litter, and NASA's EMIT and the new MADLib library are one of the first concrete steps in that direction.
EMIT: a dust instrument that started "seeing" plastic
EMIT, short for Earth Surface Mineral Dust Source Investigation, was launched on July 14, 2022, to the International Space Station platform. The instrument was developed at NASA's Jet Propulsion Laboratory as part of the Earth Venture program and was envisioned as a key tool for studying mineral dust from deserts and dry areas. This dust, carried by winds for thousands of kilometers, can heat or cool the atmosphere depending on its composition, and EMIT's task is to precisely measure the "mineral map" of these source areas.
Unlike classic satellite cameras that capture only a few wide spectral channels, EMIT is a hyperspectral spectrometer. It measures reflected sunlight in a continuous series of wavelengths from the visible to the short-wave infrared range, obtaining a detailed "spectrum" from 380 to 2500 nanometers for each pixel on the Earth's surface. In practice, this means that for every square of about 60 meters on the ground, the instrument "sees" not only the color but also the fine structure of absorption lines that depend on the chemical composition of the observed material.
When researchers began systematically reviewing EMIT data, they noticed that the same instrument that distinguishes minerals like calcite or hematite can also recognize the very specific absorption features of some plastic polymers. In a paper published in 2024, it was shown that EMIT can detect traces of high-density polyethylene (HDPE) and polyvinyl chloride (PVC) from orbit in large landfills and in agricultural areas where plastic is used as film or in greenhouse structures. Scientists have managed to extract these signals on a global scale, revealing sources of plastic waste on multiple continents.
These results confirmed that hyperspectral technology has the potential for monitoring plastic from space – at least on land, where the signal is relatively clean and the background is simpler than at sea. The next logical step was the question: can the same approach be applied one day to monitoring waste in the oceans?
How hyperspectral "fingerprint reading" works
The basis of EMIT's success lies in a technique known as hyperspectral imaging or imaging spectroscopy. Instead of describing each point on the planet's surface with three basic colors (red, green, blue), the instrument measures light reflection in hundreds of narrow "strips" of wavelengths. Every mineral, plastic, or organic material absorbs and reflects light in a unique way, creating a recognizable spectrum that experts often call a "molecular fingerprint."
Plastic, for example, has a series of characteristic absorption lines in the infrared range associated with carbon–hydrogen bonds in polymer chains. Different types of plastic – polyethylene, polypropylene, polystyrene, polyester, and others – have nuances in these spectra, depending on the chemical structure, color, and any additives. In this way, at least in theory, not only the presence of plastic but also the approximate type of polymer can be distinguished from orbit.
But for satellites to automatically recognize such signals in real-world conditions, it is necessary to have an extensive and standardized reference spectral database for a wide range of waste found in the environment. This is where Ashley Ohall and the new MADLib data library enter the story.
MADLib: a global library of marine debris fingerprints
During an internship at NASA, marine scientist Ashley Ohall gathered an international team of researchers to solve one of the key problems in remote sensing of marine debris: the lack of a unified, well-curated library of spectral "fingerprints" of different types of trash. For years, laboratories around the world had measured the reflection of plastics and other materials with handheld spectrometers, but the data remained scattered across different databases, tables, and publications, often in inconsistent formats.
MADLib – MArine Debris hyperspectral reference Library collection – was created precisely as an attempt to unify this data. In a paper published in late 2025 in the journal Earth System Science Data, the authors state that the library contains 24,889 hyperspectral measurements taken from 3,032 waste samples, collected from 13 different data sources. It includes both plastic and non-plastic materials: pieces of rope, fishing nets, tires, metals, foam materials, cork, glass, as well as typical pieces of packaging such as bottles, caps, and films.
Special attention was paid to plastic, given that it makes up the largest share of marine waste. The library covers the spectra of 19 different polymers, with samples recorded in various states: brand new, partially "weathered" in the sun, overgrown with algae, wet and dry, and even in a submerged state in water. Researchers strictly standardized the method of data collection and processing so that later algorithms could easily compare laboratory spectra with those recorded from the air or space.
In practice, MADLib functions like a catalog: when a satellite or aircraft captures a hyperspectral image of the sea surface, an algorithm can compare every suspicious pixel with thousands of known "fingerprints" from the library. If the signal matches a certain type of plastic or other waste, the system can mark with a certain degree of confidence that a specific type of material is present on the ocean surface.
Why plastic in the ocean is harder to find than in a landfill
If plastic can be seen in landfills and greenhouses, it is logical to ask why it cannot be found just as easily on the sea surface. The answer lies in the physics of light and the specifics of the aquatic environment.
Seawater strongly absorbs infrared radiation, precisely the region of the spectrum where plastic has its most pronounced characteristics. While plastic objects are clearly visible to EMIT and similar instruments on dry land, their infrared signal drops sharply as soon as they end up in the water. Additional problems are created by waves, foam, and the reflection of sunlight ("sun glint") on the surface, which can mask subtle differences in the spectrum between plastic and, for example, marine algae or natural organic material.
Another difficulty is the scale of the problem. Although "garbage patches" like the Great Pacific Garbage Patch are often highlighted in the media, most plastic is in the form of relatively small fragments scattered over a vast space. A hyperspectral satellite observing the sea surface from an altitude of several hundred kilometers must simultaneously:
– resolve very small objects in pixels tens of meters in size
– distinguish their spectrum from often very similar natural materials
– deal with changes in lighting, water turbidity, and atmospheric conditions.
Because of this, experts emphasize that EMIT and similar instruments currently work best on land-based sources of plastic pollution – for example, along rivers, landfills, or industrial zones near the coast – where the background is relatively "clean" and where it is possible to map hotspots before the waste even reaches the sea.
But at the same time, experiments are being conducted that combine data from multiple sensors and different platforms – from satellites and research aircraft to drones and ship-based measurements – to find the optimal strategy for detecting plastic in the ocean itself.
Satellites, aircraft, and artificial intelligence: a new generation of tools
NASA and partner institutions have made significant efforts in recent years to develop algorithms that can automatically recognize "islands" of marine debris from satellite images. For this purpose, publicly available images from the European Sentinel-2 system and commercial satellites with high spatial resolution are used. In one of the projects, published through NASA's IMPACT program, scientists are training deep neural networks to spot patches of trash on the sea surface by comparing thousands of labeled examples.
Reference databases such as MADLib play a key role in such systems. They allow algorithms to rely not only on the "appearance" of an object in the visible spectrum but also on finer differences in the hyperspectral signal. In the future, these models are expected to be used in combination with new missions like NASA's PACE (Plankton, Aerosol, Cloud, ocean Ecosystem), launched in 2024, which brings another advanced hyperspectral instrument to orbit for measuring ocean color. Although PACE was not designed exclusively for plastic, its sensitivity to changes in the optical properties of the surface layer of the sea could help in distinguishing areas with an increased concentration of marine debris.
Parallel to this, specialized datasets for machine learning are being developed, combining satellite imagery, field measurements, and manually labeled positions of marine debris. The goal is to create a robust ecosystem where new algorithms can be tested and compared, and the results used in real operational systems for maritime surveillance.
From global maps to local decisions
Why is it even important to know exactly where the strips of marine debris are and which types of plastic dominate? The answer is simple: without spatially and temporally precise data, it is difficult to design effective measures.
If satellite observations show that certain types of plastic accumulate most at the outflows of specific rivers, it is a signal to local authorities that monitoring of landfills and waste collection systems in those basins needs to be strengthened. If it is discovered that fishing or aquaculture equipment makes up a significant part of the waste in a certain area, regulators can direct policies toward better labeling and return standards for equipment, deposit systems, or incentives for alternatives.
Mapping plastic "hotspots" near tourist regions helps in planning beach cleanups and educational campaigns, and information about material types can be crucial for the recycling industry developing technologies for processing specific polymers. For scientists, such data opens the possibility of linking the movement of waste with ocean currents, climate patterns, and ecosystem changes.
Therefore, experts point out that today's work on EMIT and MADLib is only the initial phase: for hyperspectral technology to truly be integrated into daily marine management, it is necessary to link satellite data with local monitoring, the legal framework, and economic decisions.
The people behind the technology: motivation and hope
Behind the numbers and technical terms are real people. Ashley Ohall, originally from Florida, grew up along a coast where traces of plastic are becoming increasingly obvious. In statements following the publication of MADLib, she emphasized that her goal is to show how remote sensing can become a reliable tool for monitoring marine debris – and that the fact that something was not possible before does not mean it cannot be achieved in the future.
At the level of NASA headquarters in Washington, programs supporting such research are coordinated within the oceanography and biogeochemistry departments. Program scientist Kelsey Bisson points out that people have an intuitive, almost emotional connection with the sea and its health, and that is precisely why it is the agency's task to use advanced technology to respond to this global challenge. In her perspective, monitoring marine debris is a natural continuation of NASA's tradition of using satellites to solve socially important problems – from air quality to food security.
A similar attitude is shared by the wider scientific community gathered around international working groups for remote sensing of marine debris. Open access to data, transparent methodology, and cooperation between oceanographers, remote sensing experts, programmers, and decision-makers are prerequisites for techniques like hyperspectral imaging to be applicable outside the laboratory.
What follows: from prototype to an operational system for ocean preservation
Currently, EMIT's ability to recognize plastic on land and the MADLib library, which standardizes nearly 25,000 spectral records, represent the foundations of a future system for monitoring marine debris from the air and space. Prototype algorithms are already being built on these foundations that will merge various satellite missions, meteorological models, and field measurements.
Next steps include:
- further expansion of spectral databases, especially for plastic in different conditions in the sea (algae coverage, biofilm, various thicknesses of the water layer above the waste)
- development of methods that combine hyperspectral data from multiple spectra and high-resolution visual imagery from commercial satellites
- harmonization of new satellite missions – such as PACE and planned specialized missions for monitoring marine debris – with the needs of coastal states and international organizations
- construction of operational centers that will turn data into practical maps and warnings for competent authorities in real or near-real time
Although the path to this goal is long, the combination of precise spectral data, advanced algorithms, and increasingly powerful satellite sensors offers a new type of "radar" for plastic – one that does not depend on occasional expeditions and manual sample collection, but can continuously monitor the entire planet. In a world where plastic is increasingly shaping ecosystems, the ability to track from space where it is located and where it is traveling could become one of the key tools in efforts to preserve ocean health for future generations.
Sources:- NASA / Phys.org – report on the use of the EMIT sensor for plastic waste detection and the development of the spectral library MADLib link- NASA – official EMIT mission page with a description of the instrument, objectives, and mission status on the International Space Station link- Earth System Science Data – scientific paper "The MArine Debris hyperspectral reference Library collection (MADLib)" (Ohall et al., 2025), description of the database of 24,889 spectra from 3,032 waste samples link- 4TU Research Data – repository with the complete set of MADLib spectra and associated metadata for the development of remote sensing algorithms for marine debris link- NASA Earthdata / NASA IMPACT – projects and blog posts on using artificial intelligence and commercial satellites for marine debris detection and the broader context of remote sensing of marine debris link- UNEP – overview of marine plastic pollution and estimates that more than 11 million tons of plastic enter the oceans annually, highlighting the need for a global response link
Find accommodation nearby
Creation time: 5 hours ago