Myanmar Witness Profile picture
Feb 11, 2022 35 tweets 15 min read Read on X
🔥 Documenting fires in Myanmar
#WhatsHappeningInMyanmar

Today we look at methods for documenting the burning of civilian structures.

There are several methods we can use to verify that a fire occurred, and then to map that damage.

#OSINT #THREAD

1/35 👇
Following reports of a village being burnt, the first thing that we try to do is verify the attack. For that we look for thermal anomaly detections from several satellites that orbit the globe twice per day. These are the VIIRS and Aqua/Terra Earth Observation satellites.
The data from these satellites regularly gets uploaded to a NASA website called FIRMS:
firms.modaps.eosdis.nasa.gov/map/#d:24hrs;@…

There are a lot of fires (mostly burning agricultural fields after harvest), but if you have specific info it is very useful to verify.
Some twitter accounts are dedicated to automatically detecting likely fires within likely settlements. Such as @FIRIS_FireAlert, which has recently branched out into monitoring Myanmar along with Africa.
This method has limitations however. It relies on clear skies, and fire actively burning at around 11:30am or 11:30pm, local time, which is when the satellites pass over.

Otherwise, it won't show any detections. This means the absence of a detection =/= no fire.
In these cases, we rely on classic geolocation to verify the fires: (1)Matching features to locate the place photographed and (2) verifying that it is new footage.

Often these photos are from far away, but still allow for verification of a fire having taken place.
Next, we often try to map the extent of the damage, to quantify how many structures, or how much of the village, was destroyed.

Image shows a recent example (more here: ).

But there are several methods of varying precision that we can use.
Let's start with the most basic, classic geolocation of ground footage.

Media coming out of Myanmar is not always particularly useful, showing very close up photos of the damage that are impossible to accurately geolocate.

Here's an example.
Other footage, is more useful.

This image was taken in Thantlang during the 18 September arson attack.

We geolocated it by matching the distinctive building with other footage we had of the area: google.com/maps/place/22%….

This lets us map at least 10 burnt structures.
But these photos, and even videos, are far from able to provide a complete picture of the area that has been burnt.

For this we go back to satellites.

But the ideal, high-resolution, high-quality satellite imagery is rarely what we get...
The higher resolution a satellite is, the smaller area it can image daily, and the longer you generally need to wait for the imagery. What we generally have is lower resolution - either open-source scientific imagery, or what a company called @PlanetLabs can provide.
Let's first look at the open source satellites. The best are the Sentinel-2 satellites, run by @ESA_EO. This pair of satellites images the entire globe at 10m resolution every five days. When most of a village has been burnt it can generally be detected here.
The general rule of thumb I go by is that something is clearly detectable by a satellite if its sides are more than 4x4 pixels, so for this satellite 40x40m.

However, for stuff like the burning of villages, the change is often more subtle, more textural than anything else.
Large parts of the village of Ke Bar were burnt in mid-Dec 2021. We can verify that a fire occurred using geolocation or FIRMS, but we can't clearly quantify the scale of the fire (other than that it was big).
We can look at an RGB-stretch of the Sentinel imagery from Dec 1st (pre-fire) and from Dec 16th (post-fire) and see that patches of the village near the centre look different, along with a small area closer to the North.
The RGB stretch here approximates what the human-eye would see, but shifting into other wave-lengths provides better insight. Here you can clearly see burn scars if we look at infrared false-colour imagery, with red showing healthy vegetation and black fire scars.
Here is a comparison of an approximated map of the fire scars comparing what is detectable from natural-colour imagery (green lines) and false-colour infrared imagery (red lines). We will come back to this village later, to compare it to a 'ground truth'.

#OSINT
Next, we should look at satellite imagery from @PlanetLabs. This company has a fleet of hundreds of satellites that capture imagery of most of the earth every day at 3m resolution, and with infrared imagery.
This allows for very rapid analysis of fire damage, here we will look at fires that took place in Pale township between Dec 30th and Feb 1st, then, one village was burnt every night. We were able to download imagery from Dec 30th and Feb 2nd.
From here we follow the same idea looking at infrared imagery, but we have one more trick up our sleeve, which is helped by the fact that imagery from Planet is only days apart, rather than weeks apart like the Sentinel imagery we were relying on. Direct vegetation health.
The NDVI index directly examines the health of vegetation and helps cut out various 'noise' from the signal. Here bright colours represent healthier vegetation. And you can VERY clearly see the fire-scars.
In fact, if we directly compare the two, the areas affected by fire literally jump out. It's then possible to basically to trace these bright spots to map in pretty solid detail, as can be seen here.
Here is a video showing that entire process.
Then it's time to look at the gold-standard of remotely sensed fire analysis, Very High Resolution (VHR) imagery. @PlanetLabs and @Maxar, along with other companies such as @Airbus, have VHR imagery which makes it possible to pick out individual structures.
This sort of imagery is what makes up the vast majority of Google Maps, and the standard that most people expect when it comes to satellite imagery. However, this high resolution means a small footprint on the world, and a longer interval between images.
There is also no set schedule for when places get imaged, so unless you're willing to fork out thousands of dollars for an image, you're at the whim of richer customers - generally governments - who often aren't interested in Myanmar - or the companies themselves.
But when it is available, it is by far the best. This is looking back at Ke Bar village, where Google Earth has bought post-fire imagery of the town.

Here we can map the exact areas of fire and see - at the scale of individual structures - exactly what was burnt.
In this map we can see the fire scars mapped by Sentinel Visible [Green] and Infrared [Red] imagery, along with the VHR scar [White] which is far higher detailed. Burnt structures are white dots.
If we overlay them all together, we can see that they all accurately capture the areas that have been burnt.

There is increasing detail, and it's not until the VHR imagery that we can accurately estimate the number of structures burnt.

Here we mapped 99 burnt structures.
Finally, the last method to talk about is drone imagery.

This is the highest resolution, the quickest response and the only method that is effective regardless of cloud-cover.

Unfortunately it does rely on actors on the ground filming.
We hope that drone footage recordings will become standard in the future as the importance of documenting incidents is fully understood by those affected.

For now, drone footage - especially filmed towards the start of the fire like this example - is still quite rare.
Method-wise, it is just a better version of the VHR imagery.

You overlay it on pre-fire satellite imagery and are able to map the fire scar in detail along with the burnt structures in detail.
Here we return to our earlier example from Bin village in Mingin township.

The fire happened on the night of Monday 31 Jan 2022.

Chindwin Chit Thu was able to film the damage with a drone the following Friday.

And from that we were able to map around 185 burnt structures.
These are some of the methods, mostly open source, that we use to map fire damage across Myanmar, these methods apply globally, but the repeated and widespread arson attacks in Myanmar make it particularly pertinent there.
*Documenting fires in Myanmar*
GUEST THREAD
@MyanmarWitness X @Nrg8000

🔍Continue exploring
Map of fire incidents in Myanmar
myanmarwitness.org/fire-map
Geolocation videos on Thantlang Fires
myanmarwitness.org/thantlang-fires

Contact
👉 info@myanmarwitness.com
#MyanmarWitness

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Myanmar Witness

Myanmar Witness Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @MyanmarWitness

Sep 8, 2022
#WhatsHappeningInMyanmar
#MWFires

Fires continue to be used to destroy civilian property and entire villages in Myanmar, with devastating consequences for those that live there.

Below are 17 SERIOUS FIRE INCIDENTS added to the MW Fire Map in August 2022.

(THREAD) 👇
🚨Village burnt following clashes
☑️ Kale, Sagaing
☑️ 23.09851, 94.01889
☑️ 05 Aug 2022

firms.modaps.eosdis.nasa.gov/map/#t:adv;d:2…

facebook.com/khitthitnews/v…

#MWFires #Myanmar
🚨Thantlang burning again
☑️ Hakha, Chin
☑️ 22.6922, 93.42308
☑️ 06 Aug 2022

firms.modaps.eosdis.nasa.gov/map/#t:adv;d:2…

DM with links to footage if safe to do so.
#MWFires #Myanmar
Read 19 tweets
Sep 2, 2022
#WhatsHappeninglnMyanmar
🔴How do we verify fire claims?
🟡Geolocation Walkthrough

On 29 August 2022, SGG News reported a fire in Taung Myo village, Sagaing.

The disturbing images led MW investigators to try to verify the report, starting with geolocating the footage.

THREAD👇 Image
#OSINT #Geolocation

After initial assessment, it was possible to determine that several images showed the same location but were taken from a different angle.

When different images contain overlapping information, that can be a good place to start the geolocation work. ImageImageImageImage
Investigators often rely on satellite imagery for geolocation.

Analysis of the roofs and roads in the images can yield valuable information. Roof aspects (size, colour, shape) can be compared with road layout to make a unique pattern MW can then look for in satellite imagery. Image
Read 11 tweets
Aug 29, 2022
🔴 Lay Kay Kaw
🟡 Report out soon

Lay Kay Kaw (လေးကေ့ကော်) in Kayin (ကရင်) State, once designated a ‘town of peace’, has been the site of violent clashes between the Myanmar military and the Karen National Union (KNU), since December 2021.

#WhatsHappeningInMyanmar
THREAD 👇 ImageImage
Instead of providing a site of refuge to Internally Displaced Persons (IDPs), returning refugees and KNU families, fighting in Lay Kay Kaw (လေးကေ့ကော်) has forced many residents to flee for their lives. Image
MW have published the first in a series of reports that document events in the region. It provides a detailed chronological reconstruction of the clashes between the military and oppositional forces between 13 and 28 December 2021. Image
Read 10 tweets
Aug 24, 2022
🚨BURNT BODIES
#WhatsHappeningInMyanmar

17 claims involving the burning of victims or their bodies in Myanmar (July 2021 - August 2022).

This list is not exhaustive and includes some credible but unverified claims where further evidence is required.

A thread👇TW//GRAPHIC
#WhatsHappeningInMyanmar
🔴 Burnt Bodies in Myanmar
🟡 Claim 1/17
Unverified media report

AUNGMYAYTHAZAN, Mandalay
27 March 2021

“Armed forces burnt a Mandalay resident alive”
myanmar-now.org/en/news/help-m…

Please upload evidence
myanmarwitness.org/upload-evidence
#WhatsHappeningInMyanmar
🔴 Burnt Bodies in Myanmar
🟡 Claim 2/17
✅ Geolocated images

KANI, Sagaing
28 JULY 2021

“The 11 men were massacred and set on fire”
Letter from Ambassador Kyaw Moe Tun
myanmarmissionnewyork.org/post/letter-fr…

“Multiple bodies”
Read 19 tweets
Aug 17, 2022
🚨Vehicle Burning
...can seem insignificant when compared to other reports of #WhatsHappeningInMyanmar.

But vehicles
+ give freedom of movement
+ transport goods
+ often most-valued possession

When they are burnt it often reveals important clues to what happened.

Thread👇 Image
🚨Burnt vehicles
☑️Yinmarbin, Sagaing
☑️22.073197, 94.789316
☑️14 August 2022

A truck and motorcycles were burnt, following reports of a massacre in Yin Paung Taing village.

myanmar-now.org/en/news/the-sm… ImageImageImageImage
🚨Burnt truck
☑️Pwintbyu, Magway
☑️20.317129, 94.488443
☑️22 June 2022

We geolocated an image of a burnt truck, following reports of military activity in Pwintbyu. Image
Read 11 tweets
Feb 10, 2022
မိမိဖုန်းကိုလုံခြုံအောင်ဘယ်လိုထားမလဲ……

ဖုန်းလုံခြုံရေးအတွက် အရေးကြီးတဲ့ အချက်များကို မျှဝေပေးချင်ပါတယ် 👇👇👇

1/7

#WhatsHappeningInMyanmar
#Myanmar #securityexplained
#MyanmarWitness
#ဒစ်ဂျစ်တယ်လုံခြုံရေး
#မိုဘိုင်းဖုန်းလုံခြုံရေး
Read 7 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us!

:(