Site icon Conservation news

Radar returns to remote sensing through free, near-real-time global imagery

  • The European Space Agency’s launch of the Sentinel-1 satellite has made 20-meter resolution radar imagery of the whole planet freely available.
  • The “all-weather, day-and-night supply of imagery of Earth’s surface” complements standard optical satellite imagery in detecting forest loss, even under heavy cloud cover.
  • The Monitoring of the Andean Amazon Project (MAAP) demonstrates the benefits of analyzing free radar imagery to accurately quantify wet season loss of rainforest in the Peruvian Amazon.

Satellite imagery has transformed how we assess changes in forest cover.

The standard optical sensors carried by Earth-orbiting satellites capture the energy from sunlight reflected off objects on the Earth’s surface. These sensors recognize levels of brightness and color in the reflectance, which enables users to distinguish between greener and browner vegetation, or row crops and savannah, as well as among objects, such as tree species, with different chemical properties.

They cannot, however, see though clouds. Clouds block the reflectance data from reaching the sensor, which limits the effectiveness of even high-resolution (3-5 meter) satellite data, such as Planet or DigitalGlobe, in rainy regions like the Amazon.

The Monitoring of the Andean Amazon Project (MAAP), which uses remote sensing data to highlight deforestation hotspots in the western Amazon and activities that cause forest loss, has addressed this challenge with an alternative, 80-year-old technology—radar.

MAAP now combines medium- and high-resolution optical data with the power of radar imagery to enable its team to monitor deforestation in the Peruvian Amazon year-round, in near real-time.

The rainforest along a meandering river in southeastern Peru. Clouds block the “view” that satellites’ optical sensors have of the Earth below. Photo credit: Sue Palminteri

An all-weather, day-and-night source of imagery

Radar technology uses radio waves to determine the location, distance, or velocity of objects.

Originally RAdio Detection And Ranging, or RAdio Direction And Ranging, radar was developed for military use during the First and Second World Wars. It remained effective but expensive. That barrier changed in 2014, when the European Space Agency (ESA) launched its new Sentinel-1 satellite into space.

Similar to the now-freely available 30-meter resolution Landsat optical imagery, the Sentinel-1 project provides a constant supply of free, 20-meter radar data for the entire planet.

“With Sentinel-1, regardless of weather conditions, there is a crisp new image every 12 days at 20 meters resolution,” MAAP’s director Matthew Finer told Mongabay-Wildtech. “That is pretty incredible for monitoring in the Amazon.”

ESA developed Sentinel-1—actually a pair of satellites, orbiting on opposite sides of the Earth— to monitor a wide range of natural and human activity, from the Arctic sea-ice extent to forest, earthquakes, oil spills, ships, and volcanoes.

How can radar “see” changes on the Earth, such as deforestation, across landscapes that are shrouded in cloud cover for much of the year?

How satellite-carried radar sensors emit radiation down to Earth and receive some of the reflected energy. Knowing the height and speed of the satellite and other details allows users to delineate the size, structure, and movement of objects on the Earth’s surface. Image credit: MAAP

Unlike optical sensors, that only capture reflected energy and thus require the sun’s radiation, radar emits its own radiation, which can pass through the clouds. This energy, typically in the microwave portion of the electromagnetic spectrum, hits the Earth and bounces back to its sensor, even at night or in cloudy conditions.

The radar sensor detects, measures, and times the backscattered microwave radiation. The time required for the radiation to travel to the object and return back to the sensor determines the object’s distance from the sensor.

Radar cannot see color, but this time differential allows it to determine the structure and surface roughness of objects, such as buildings or trees of different height, width, and density. This process is repeated for all objects as the satellite passes by, enabling it to produce a two-dimensional image of the surface below.

Quantifying site-level deforestation

In the Peruvian Amazon, Sentinel-1 obtains imagery every 12 days with a resolution of  approximately 20 meters (66 feet), Finer said in an email.

A basic gold-mining operation along the Madre de Dios River in Peru on left. Photo credit: Sue Palminteri Aerial image of the fine-scale but devastating impact of gold mining to the southwestern Amazon’s rivers, wetlands, and forests. Photo credit: Rhett A. Butler


A recent MAAP analysis demonstrated the power of radar imagery to enable near real-time deforestation monitoring under cloud cover. The analysis focused on the Madre de Dios region in southeastern Peru, an area experiencing ongoing deforestation driven in large part by gold mining along the rivers and wetlands. Using radar, the project detected the loss of 3,260 acres [1,320 hectares] of forest over the past year.

Time series of deforestation in the Madre de Dios / Tambopata region of southeastern Peru. This animated series of Sentinel-1 radar images shows the spread of deforestation (shown in purple) from gold mining operations between January 2017 and February 2018, within a generally forested landscape, shown in yellowish-green. Image credit: MAAP, Data: ESA, SERNANP (Peru)

Deforestation by gold miners expanded rapidly within study areas A, B, and C. Study area D is the Tambopata National Reserve, where illegal gold mining-related deforestation intensified in 2016 but was effectively halted in 2017.

Detecting land use change that takes place during dark or rainy periods can offer radar data greater speed and accuracy in assessing total deforestation than is possible using just optical image data.

This second image indicates where radar detected deforestation between January 2017 and February 2018 in these same study areas. Around half of this loss took place between October 2017 and February 2018, the rainy season, when the availability of good optical images was more limited due to persistent cloud cover.

“We documented the illegal gold mining deforestation of over 1,600 acres [650 ha] between October and February, a time of year when there is little cloud-free optical imagery due to the beginning of the rainy season,” Finer said.

Radar-detected deforestation between January 2017 and February 2018 (indicated in yellow and red) in the same study areas. Red indicates the 1,609 acres (650 ha) of forest lost since October 2017, during the rainy season. Image credit: MAAP, Data: ESA, SERNANP (Peru)

An optical image of the same area misses some of the rainy-season deforestation detected by the radar imagery, suggesting that at this spatial extent and time period, radar more accurately detected the gold mining deforestation.

An optical image of the same area did not capture all wet-season deforestation as well as the radar image. Each technology has advantages; freely available data allows monitoring teams to integrate findings from the two systems. Image credit: MAAP, Data from Planet Team (2017). Planet Application Program Interface: In Space for Life on Earth. San Francisco, CA.

New tools to analyze new data types

Finer said that although the Sentinel-1 imagery is free, it can be challenging to analyze because it uses just one band in the microwave part of the electromagnetic spectrum, “so it looks very different than an optical image and may be difficult for non-experts to analyze.” In contrast, Landsat has 8 or 11 different bands, and tools exist to analyze those bands to visualize where forest loss has occurred.

“However, there are techniques, such as multiple polarizations, that provide richer data for [radar] analysis,” Finer said. “For example, in MAAP #79, we present dual polarization images, so there is a bit of color to them instead of just black and white of the single polarization image.”

Intact palm wetland along a rainforest lake in southeastern Peru. Miners seek gold in the sediments, cutting trees and inputting harmful mercury into waters to identify the tiny gold pieces in the soil. Photo credit: Sue Palminteri

The MAAP project and others aim to spark more interest in radar monitoring among forest monitoring agencies and to develop tools for analyzing radar data. “Just like with optical imagery years ago,” Finer explained, “the transition from expensive to freely available data is sure to spark much innovation.”

“There are initiatives to develop an automated forest loss detection system, similar to what [Global Forest Watch] GLAD alerts do with Landsat imagery,” Finer explained. “We plan on using radar much more this year in Peru and hope to show its power and utility to government officials working to improve the country’s near real-time monitoring and response capabilities.”


FEEDBACK: Use this form to send a message to the editor of this post. If you want to post a public comment, you can do that at the bottom of the page.