Rising Waste Levels Observed at Samarco Prior to Disaster

Satellite imagery collected in the months leading up to the catastrophic Samarco mine disaster on November 5 in Minas Gerais, Brazil reveal a substantial increase in the amount of water and mine waste being stored behind the now failed Fundão Dam. Images taken by the satellite Earth-imaging company Planet Labs two months before the dam collapse show that Samarco, co-owned by BHP Billiton and Vale SA, were acting on their plans to raise the height of the dam. Compared to 2013 Astrium imagery in Google Earth, additional structures appear at the top of the dam, trees have been cleared and roads have been cut to accommodate the heightened level of waste in the reportedly 55 million cubic meter impoundment:

On the left side of the image you can see that by September 2015 the fluid level had risen substantially since May 2013, filling valleys upstream of the dam. In the center, you can see the growth of the dam as new contours are added, presumably to raise the crest of the dam. According to our calculations, between May 2013 and September 2015 the surface area of the impoundment increased by approximately 100 acres (406,000 sq. meters).

Though the comparison is not nearly so stark, here is another image collected by Planet Labs on October 2, side-by-side with the same September 25 image seen above. The images were collected at different times of day, so features that were in the shadows on one image will be visible in the other:

Mining and Civil Engineers – See anything notable about developments at the Fundão Dam? Leave a comment below. 

Oily globs close SoCal beaches – Where did they come from?

Yesterday afternoon oily globs from an unknown source began washing ashore in Southern California, prompting officials to close Manhattan Beach, Hermosa Beach, and Redondo Beach. Reports saytar-like globs ranging from the size of golf balls to footballs began washing ashore along a six-mile stretch of coastline.” The U.S. Coast Guard has yet to identify a source, but did confirm the material was a petroleum-based product. Last week a pipeline spill of over 100,000 gallons of crude oil west of Santa Barbara dominated the headlines, so there has been some speculation that this oil came from that spill. However, thanks to satellite imagery, we believe we have found a more likely source. 


This image was collected at 6:52 am PDT on May 26 by Sentinel 1A, a radar satellite operated by the European Space Agency. Lifeguards at Manhattan Beach first reported the “tar-balls” washing ashore around 12:30 pm PDT, approximately 30 hours later. The image appears to show an oily slick only six miles west of Manhattan Beach and covering approximately 1,000 acres of Santa Monica Bay.    

Sometimes we see bilge dumps from passing vessels (like this 90-mile spill off the coast of Angola), but this slick doesn’t appear to have any vessel associated with it and it isn’t the right shape. So we started looking for any infrastructure around Santa Monica that could be the source of this pollution event. That is when we found this map of the five-mile sewage outfall from the Hyperion Wastewater Treatment plant at Playa del Ray. Take a look at what the outfall pipe looks like on the bathymetry layer in Google Earth. 


To see if this pipe might have anything to do with our suspected oil slick we digitized the outline of the Hyperion 5-Mile Outfall and brought that into QGIS to lay on top of the imagery from Sentinel 1A.



The southern diffuser leg of the Hyperion 5-Mile Outfall is less than 900 meters from the northern edge of the suspected oil slick.

Given how close the slick is to the Hyperion 5-Mile Outfall, we believe this could be a source for the contaminants washing up on Manhattan, Hermosa, and Redondo Beaches. It is not impossible that tarballs from the pipeline spill at Refugio Beach State Park travelled over 100 miles to wash up en mass on these beaches, but it seems less likely than a discharge of some kind from the Hyperion Wastewater Treatment plant. 

Furthermore, we found more evidence which supports the notion that waste from the outfalls could reach shore given the right conditions. Back in 2006 the City of Los Angeles shut down Hyperion’s 5-mile outfall for inspection, diverting the wastewater to the shorter 1-mile outfall. Scientists at the Southern California Coastal Ocean Observing System (SCCOOS) tracked the waste from the shorter outflow pipe and modeled the behavior of the plume. Bear in mind that the animation below shows the behavior of a pollution plume back in Winter 2006 from a point-source four miles closer to shore, but you can see how it is possible that waste from the outfalls could turn toward shore and contaminate the beaches.

See more about the Hyperion outfalls at the SCCOOS website, and let us know if you have any additional information about this pollution event or the Hyperion Wastewater Treatment Plant.

National Response Center (NRC) Alerts Are Back!

This week we restored a nationwide data source to the SkyTruth Alerts system, our free email-based pollution and incident notification tool. For over a year, the public interface for the U.S. Coast Guard’s National Response Center (NRC) has been offline, and we have not been able to send out notifications when new pollution reports are reported. The NRC receives approximately 30,000 reports a year, and about 10,000 of those are in the Gulf of Mexico. This outage substantially limited our ability to research pollution incidents and notify the public about oil and hazardous chemical releases, so we are pleased to announce that we can now provide weekly alerts from this important data source.

The Coast Guard has been posting a weekly update of all the reports they receive, in the form of an excel spreadsheet. It has taken us a while to do the necessary data wrangling and re-coding, but we have finally found a way to incorporate this unwieldy file into our Alerts system so we can resume providing you the easy-to-understand email notifications that you’ve been missing.

A couple of details and cautions about how the restored system will work as long as we receive these weekly updates from the NRC:

  • Weekly Updates: We will issue Alerts every time the file is updated at www.nrc.uscg.mil, typically on Tuesday or Wednesday for everything that has been received in the preceding week.
  • >50 Alerts: Because you will be getting a weeks worth of reports all at once, there is a good chance your weekly update from the NRC may include more than 50 incidents, so we won’t able to show everything in your daily email. There are a few simple solutions to this…
    • Break up your subscription into smaller areas. Just unsubscribe from your current subscription and re-subscribe to a several smaller areas.
    • When you get your weekly email, just click through to the Alerts map (below) to see everything that has happened in your area of interest.

    NOTE: This map is dynamically updated, but for performance reasons we can only show up to 50 incidents at a time, and we only display Alerts that have occurred in the last 30 days. Be sure to zoom in to the city or county level and pan around to make sure you are seeing everything.
    Unfortunately, this is best that we can do with the available data, and we won’t be able to restore daily Alerts from NRC data until further upgrades are made on the government’s end. In the meantime, we continue working on a general overhaul that will incorporate more data sources, new features, and improve the overall user experience. We currently have over 2,200 subscribers, which is far more than we ever imagined we would have when we built this tool for our own internal use. We hope you’ll continue to find this service helpful, and while Alerts are free, your donation will certainly help us maintain and improve this resource.

    To sign up for alerts, or to resubscribe to new areas, visit – alerts.skytruth.org/.

[Updated] Bangladesh – Oil Spill in the Sundarbans National Park

Updated Dec 16, 2014 at 6:00 PM with new information on the location of the Southern Star 7, as well as new and updated satellite images.

Posted Dec. 15, 2014 at 11:00 PM: On the morning of Dec. 9, 2014, a tanker carrying heavy furnace oil to a powerplant in Bangladesh was struck in the fog by a cargo vessel and partially sank, releasing thousands of gallons of oil into the Sundarbans, the world’s largest continuous mangrove forest and a UNESCO World Heritage Site. This tidal river delta, already threatened by climate change, is home to incredible biodiversity including rare Irrawaddy and Ganges dolphins and what is believed to be one of the largest populations of the very endangered Bengal Tiger

Details, including the exact location* of the incident, remain vague, even though nearly a week has passed since the accident. The Times of India reports that while the accident occurred in a commonly travelled shipping lane, the collision occurred within one of the Sundarban’s three dolphin reserves.

*SkyTruth has now received the coordinates of the incident from representatives on the ground. See below.

The “Southern Star 7” was carrying somewhere between 66,000 and 92,000 gallons (250,000-350,000) of furnace oil, but how much was actually spilled into the river remains unknown.

Residents have been seen collecting the oil by hand and with buckets to sell for a small reward to the state run Padma Oil company, while fishermen attempt to use their nets to contain the spill. Regional officials just announced they are hiring 100 boats and 200 workers to expand the clean-up effort. These tedious and messy clean up methods are a stark reminder that even after the BP disaster in the Gulf of Mexico in 2010, the Exxon Valdez spill off Alaska in 1989, the Ixtoc 1 spill in the Gulf of Mexico in 1979, and even the Santa Barbara Oil Spill off California all the way back in 1969, we haven’t really made any major improvements in how we clean up spilled oil.

SkyTruth has been monitoring satellite images of Sundarbans National Park, and we believe we can see evidence of the oil on imagery from the European Space Agency’s new radar satellite: Sentinel 1 – A.

According to sources on the ground, the Southern Star 7 sank into the river at 22°21’14.33″ N, 89°40’17.66″ E, about four kilometers from the confluence with the Passhur River. On December 12th it was lifted from the river floor and moved up to the riverbank to 22°22’1.44″ N, 89°38’30.91″ E.Because this region is a tidal river delta, water sloshes in and out of the mangrove forest twice a day. There are reports that the oil is continuing to spread up and down the river, and throughout the canals and channels that crisscross the region. On radar satellite imagery, we have observed what appears to be ropy strands of oil along 30 miles of the Passur River.

Here are some more images of the area as seen by Sentinel 1 – A and Landsat 8…


To download these images for yourself, visit EarthExplorer for the latest Landsat, and create an account at the Sentinel Scientific Data Hub.

SkyTruth Releases Map of Drilling-Related Impoundments Across PA


Thanks to the efforts of several hundred citizen-scientists who participated in our FrackFinder PA projects, we are releasing a map of fluid impoundments that we believe to be associated with drilling and hydraulic fracturing (fracking) in Pennsylvania’s Marcellus Shale.

Among other public and environmental health concerns, the shale drilling boom in Pennsylvania has led to construction of hundreds of large ponds and reservoirs to hold the water needed to frack modern shale gas wells, and to contain the “flowback” or “produced water” that returns to the surface laden with toxic and often undisclosed chemicals.

Click here or above to explore a full-screen, interactive map: This map displays impoundments related to shale gas drilling and hydraulic fracturing (fracking) in Pennsylvania, as identified by SkyTruth staff and volunteers on USDA aerial survey photography from 2005, 2008, 2010 and 2013.

Did we miss an impoundment, or mark a pond that isn’t related to drilling and fracking? Let us know – info@skytruth.org- and please include as much detail as possible (pictures, GPS coordinates, etc.).

These impoundments are of interest to researchers at Johns Hopkins University who are investigating the public health implications of living near drilling and hydraulic fracturing. We couldn’t find any existing public map or dataset that would do the job, so we put together the FrackFinder program and asked for your help.

We already knew living next to one of these massive impoundments could be a headache (figuratively, if not literally) because we noticed last year there were ten separate complaints to the National Response Center from five different residential addresses surrounding the 13.5 million gallon Carter Impoundment in Washington County. Since then, our partners at Earthworks published a detailed case study of this site, and Range Resources was recently fined $4.15 million by Pennsylvania’s Dept. of Environmental Protection (DEP) for violations at the Carter Impoundment and five other centralized fluid impoundments in Washington County. The penalty is the largest fine to date assessed against a shale gas driller by PADEP.

Preliminary Findings

Year     Number of Ponds      Area – Average (meters2)      Area – Median (meters2)


In 2005, drilling was just getting started, so we don’t see many ponds or large ones. But as drilling ramped up, the ponds grew progressively larger. From 2010 to 2013 the median area of drilling impoundments more than tripled, and the average area (which also includes small fluid reserve pits located right on the well pad) more than doubled. As of 2013, the total impoundment surface area measures nearly four million square meters, scattered across the Commonwealth. (New York’s Central Park measures 3.4 million square meters.)

We also observed that these impoundments are not permanent and may be reclaimed after a few years. Of the 581 ponds we delineated in 2010, only 116 of them were identified again on 2013 imagery.


To find these impoundments, we asked volunteers to look at the aerial imagery of locations where drilling permits had been issued, and respond to very simple questions about what they saw on imagery taken in 2005, 2008, 2010 and 2013. But how did we keep well-meaning volunteers from going on a wild-goose chase and incorrectly labeling all the duck ponds in Pennsylvania as possible toxic waste storage facilities? The answer is a lot of work and multiple safeguards to make sure we produce the most accurate result possible; all while engaging the public in the process.

Careful Control of Sites Reviewed

In the first phase of the project, Project Tadpole, we showed ten different volunteers locations where a shale gas well had been permitted and asked if they saw a drilling site (“well pad”) and drilling activity at the site.

Screenshot from FrackFinder PA – Project Tadpole 2013: Since the crosshairs mark the exact location of a permitted gas well, there is little opportunity to mistake a new house for a well pad. 

In the second phase, Project Moor Frog, we asked our volunteers to identify anything that looked like a pond within a one square kilometer area of a well pad (the red shaded area in the example below).

Screenshot from FrackFinder PA – Project Moor Frog: In this phase, volunteers mark anything around the well pad that looks like a pond. We obviously get a few duck ponds in this set, but not to worry, we’re going to check everything two more times before we’re done.

For the third and final public phase, Project Dart Frog, we took all the features that people thought looked like ponds (including quite a few duck ponds, shadows, manure lagoons, and other stuff) and showed them to ten different volunteers who had been trained to tell the difference between ordinary farm ponds and the large, recently constructed impoundments we’re interested in.

Screenshot from FrackFinder PA – Project Dart Frog: Once we’ve identified ponds in the immediate area around a well pad, >7 of 10 volunteers have to agree that it has the characteristics of a drilling impoundment (straight edges, visible black or white liner, large access roads, etc.). Even this isn’t 100% perfect, but it narrows down the number of a site we need to look at in the final step. 



Notice how we talked about showing these images to multiple volunteers? We had to have >70% agreement among our volunteers for a site to pass through to the next phase of each of our projects.


Internal Quality Control

After each phase of the project, we review the results and check some of the sites for ourselves to make sure we agree with our volunteer’s determination. You can review the findings of our QA/QC here and our full methodology here.

Finally, after our volunteers helped us with all those different phases of the FrackFinder project, we asked our GIS analyst Tita to outline all of these impoundments manually. If one of Farmer Brown’s duck ponds somehow managed to get through all those hurdles, we still had this last step to throw out a few lingering sites that obviously didn’t belong.




This dataset may not capture every drilling-related impoundment that has ever been built in Pennsylvania, just the ones that appeared on the USDA’s aerial photographic surveys that are taken every 2-3 years. Furthermore, we only outlined impoundments that fit our criteria for proximity to active, permitted drilling locations, and that were identified in multiple project phases by our volunteers and SkyTruth staff. However, since we are only looking at these sites from far above, it is possible that we excluded or missed a few drilling sites and impoundments, or labeled something a drilling-related impoundment when it is just a duck pond. We welcome your feedback to help us make this map as accurate as possible — and we welcome your help with projects like this shortly, as we respond to concerns from citizens in other drilling states like West Virginia and Texas.

Contact us – info@skytruth.org – to submit corrections, request access to the data, or sign up to get notified when we launch future projects!

Editors Note: Since Google Maps Engine previously hosting this data has been discontinued by Google, the map data and styling was migrated to a new Carto account in January 2017.