Latest / Stage Talks with Bellingcat / "Monitoring Violence In Myanmar" with Bellingcat's Pooja Chaudhuri
Transcript
- 0:04You're listening to a stage talk titled Monitoring
- 0:07Violence in Myanmar with Bellingcat's Pooja Chaudhuri.
- 0:11Pooja reminded us of the complexity of reporting
- 0:13on a region such as Myanmar and gave advice based
- 0:16on her own experiences researching Rakhine states.
- 0:20You can find links to all the resources mentioned
- 0:22on the talk in the podcast description. This
- 0:25talk was hosted by me, Charlotte Maher, on Thursday
- 0:27the 28th of May, 2026, in the Bellingcat Discord
- 0:31server. I think we should get started. Welcome
- 0:48to this week's State Talk. Today we're joined
- 0:50by my wonderful colleague and talented researcher,
- 0:53Pooja Chaudhuri. Pooja specialises in human rights
- 0:57and conflict reporting. For the last three years,
- 1:00she's been monitoring developments in Myanmar
- 1:02via open sources, plotting evidence of the destruction
- 1:06of villages and reporting on airstrikes and violence.
- 1:09In her latest, published yesterday, she identified
- 1:12115 villages in Rakhine State as partially or
- 1:16completely destroyed since the February 2021
- 1:18military coup that overthrew Myanmar's democratically
- 1:22elected government. Today we'll hear more about
- 1:25that work, the tools, methods and skills involved
- 1:28in ongoing monitoring and why open source research
- 1:32can be so valuable in regions facing conflict
- 1:34and upheaval like Myanmar. We will have time
- 1:38for audience questions. You can place your questions
- 1:40in the chat accessible on the right -hand corner
- 1:43of your screen. But please remember, especially
- 1:46if this is your first stage talk, that this is
- 1:49being audio recorded for our podcast series.
- 1:52So if you don't want me to mention your username,
- 1:54please add that as a note in your question and
- 1:57I will absolutely avoid doing that. Okay. Pooja,
- 2:01tell us a little bit more about your investigation,
- 2:04please. Like Charlie said, I've been looking
- 2:08at Myanmar for a few years now. But just to clarify,
- 2:13it's a big country. Different regions, different
- 2:16states have their own issues. And I am not an
- 2:22expert on Myanmar in any way, but I have focused
- 2:26the past few months looking at Rakhine. And the
- 2:30issues that Rakhine faces would be different
- 2:32from other states in Myanmar. because the country
- 2:37has a lot of armed ethnic groups trying to fight
- 2:40the military and trying to gain control of their
- 2:45respective states. The focus of my story was
- 2:50looking at what happened in the last five years
- 2:54in Rakhine since the 2021 coup. And it's already
- 2:59reported a lot that the military does airstrikes.
- 3:03and tries to bomb civilian areas as well. But
- 3:07what I wanted to see is what happens to villages
- 3:11and towns once they are attacked? Is there redevelopment?
- 3:16If the armed groups are taking control of a certain
- 3:19area, what happens to that area? Are people able
- 3:23to return back? And also, do settlements get
- 3:29completely erased once they are bombed? lots
- 3:33of questions I had while trying to do this research.
- 3:37It's very difficult to find answers to most of
- 3:40these questions because of the tight grip of
- 3:44the military in Myanmar and the tight grip of
- 3:48the Arakanami in areas that it controls. To briefly
- 3:53tell everyone about what is happening in Rakhine
- 3:56is... So the military is fighting this ethnic
- 4:02armed group called the Arakan Army. Ethnic armed
- 4:05groups are based in different states in the whole
- 4:09of the country, and Arakan Army is the one that
- 4:12is based in Rakhine. Why I looked specifically
- 4:17at Rakhine is for two reasons. One is that Arakan
- 4:23Army has been able to take over most of the state
- 4:25from the military. Rakhine has 17 townships and
- 4:29the Arakanami right now controls 14 of those
- 4:32townships. So there are three which are not in
- 4:35control of Arakanami yet. And the armed group
- 4:40has vowed to take over the state by next year.
- 4:44Whether or not that will happen, time will tell.
- 4:47But a lot of the fighting currently is concentrated
- 4:50in the townships that the Arakanami or AA hasn't
- 4:55been able to take over. So because of the progress
- 4:59that the AA has made in Rakhine, I wanted to
- 5:02look at it. And the second was the civilians
- 5:05in Rakhine, mainly the Rohingya, face attacks
- 5:09from both sides. So both from the military and
- 5:12from the AA. And why that is, is because the
- 5:19military does forced recruitment of Rohingya
- 5:23to fight on its behalf. And what happens is that
- 5:28if an area is attacked by the military, along
- 5:32with Rohingya militia, there are retaliation
- 5:35attacks from the Arakanami. And then they target
- 5:38Rohingya villages and Rohingya towns. So the
- 5:42civilians face attacks from both sides. And this
- 5:46has also been reported since the military started
- 5:50losing a lot of territory to the Arakanami. they
- 5:53increased the force conscription of Rohingya.
- 5:57And in 2024, which we also reported on this attack
- 6:03that happened in a township called Bhutidong.
- 6:07And Bhutidong is one of the two townships that
- 6:11has the largest Rohingya population of the country.
- 6:14And this township was first attacked by the military
- 6:19and Rohingya militants and then in retaliation,
- 6:23the Arakanami then attack the township. And if
- 6:25you see satellite imagery of lots of areas of
- 6:30the main town, which shares the name of the township,
- 6:33Puthidong, and neighboring villages, you'll see
- 6:36that a lot of it just looks burned. So this is
- 6:41what I wanted to focus on. I wanted to find out
- 6:43how many villages have been destroyed, both by
- 6:48the AA and the military, although Because reporting
- 6:52is so difficult in Myanmar for a lot of villages,
- 6:56you will not be able to tell who did the attack.
- 6:59You can make good assumptions based on the kind
- 7:06of attacks, because if it's a heavily civilian
- 7:08area, the military is more prone to doing attacks.
- 7:12The military also has more air power than a Rakan
- 7:16army. So there can be informed assumptions about
- 7:20who carried on an attack, depending on when it
- 7:23took place and what it looks like on satellite
- 7:25imagery. But the story does not say for a lot
- 7:30of the villages who did the attacks because there
- 7:32is no reporting. So how I found areas that have
- 7:37had large scale destruction is the methodology.
- 7:42It took me a while to figure out the methodology
- 7:44because you know, due to lack of reporting, and
- 7:47I wasn't sure where to look at. And it's very
- 7:50rural, the villages are tiny, like, you know,
- 7:53some have like 50 homes, some have 10 homes,
- 7:55some have 80 homes, and they're very spread out.
- 7:59So it's difficult to know where you should be
- 8:02looking at. And there is lack of updated satellite
- 8:05imaging. So the sources that I refer to were,
- 8:12one was Accolade, which has conflict data, and
- 8:16it's very easy to use. I can show how I used
- 8:20Acklid. So you can download data from Acklid.
- 8:25So when you download the data, it will give you
- 8:28a CSV file, which would also have coordinates.
- 8:31It will not have coordinates of the exact point
- 8:34where an attack took place, but it would have
- 8:36coordinates of the area. So if a village was
- 8:38attacked, it would have coordinates of the village
- 8:41that was attacked. And then you can, if you have,
- 8:45if the tag was big enough, you know, you'll be
- 8:48able to see the destruction and satellite imagery.
- 8:50If not, then there could be video or images of
- 8:54that village being attacked and then you can
- 8:56do geolocation. So I used Ackler, I used NASA
- 9:00firms. If the fire is big enough, then it would
- 9:04show up on NASA firms and then you get an area
- 9:07of where, you know, fire was detected and then
- 9:11you can see if the destruction was also captured,
- 9:15if it was not enough to be captured on satellite
- 9:17image. The third was looking at just news reports
- 9:23and including on social media. There are media
- 9:28organizations that just operate out of Facebook
- 9:30and just checking what they have been reporting
- 9:34on. And of course, the fourth was just lurking
- 9:39in planet and I spent hours and hours just like
- 9:43looking at different townships and just like
- 9:47seeing which one has the most updated satellite
- 9:51imagery and then just being sort of like a satellite
- 9:56yourself and trying to see what this area looks
- 9:59like versus this area looks like. So that's how
- 10:02I got the data for the story. And what I found,
- 10:06I think the most interesting finding in the story
- 10:09was from this township called Mangdao, which
- 10:13is next to Butidong. And Mangdao and Butidong
- 10:18has the most number of Rohingya population in
- 10:23the country. And I saw that there were villages
- 10:27that I could see on planet on Mangdao. And I
- 10:31could not find the name of those villages, some
- 10:33of those villages. on the township map that the
- 10:37UN produces. And the UN map is based on data
- 10:42from the Myanmar government. So technically those
- 10:46maps are from, those are government maps, those
- 10:48are official maps. And I saw that some of these
- 10:51villages, while I can see that they are marked
- 10:55on the planet and they look destroyed, they aren't
- 10:59marked on the maps. So I saw that then what is
- 11:04happening potentially is that settlements get
- 11:07destroyed and then they get officially removed
- 11:12from sources. And I spoke to experts about what
- 11:18could be the reason behind removing names of
- 11:21villages from maps. And the experts told me it
- 11:24could be to just erase the identity of the local
- 11:29population there. There has been some reporting
- 11:32on this when it happened after the 2017 violence
- 11:37in Rakhine, the Rohingya genocide. And experts
- 11:42then said that it's so that people can't go back
- 11:45to their homes. It's not just that the village
- 11:48has been completely flattened, but its trace
- 11:54has been removed officially. So where do people
- 11:57go back to? So I think that was one interesting
- 12:02finding and there could be a bigger story there.
- 12:05Like I did not spend too much time on finding
- 12:07out all the villages that have been removed officially
- 12:11from maps, but I did find six or seven of them
- 12:14that are no longer from Wangdao that are no longer
- 12:18in the latest maps that was released by the UN,
- 12:22which is from 2024. Yeah, so that was mainly
- 12:27what the focus of the story was. And it also
- 12:32speaks about what is currently happening in the
- 12:35townships that are not in the control of the
- 12:39Arakanami. Data for one of the townships wasn't
- 12:44available, but the two townships for which data
- 12:47was available is Situwe and Chaupiu. Situwe is
- 12:52also the capital. And so the name of the city
- 12:56is Sittwe city, which is the capital and it's
- 12:58in the township, which is also called Sittwe.
- 13:02And so a lot of the fighting is now focused in
- 13:06Sittwe. And what we found is that surrounding
- 13:10the city, the villages that surround the city,
- 13:12they have been cleared and a lot of defense structures
- 13:16have been built in place of these villages. and
- 13:21the expert that we spoke to said that this is
- 13:23so that it becomes difficult for the Arakanami
- 13:25to advance into the city. Sittwe also holds thousands
- 13:30of people who have taken refuge in the city fleeing
- 13:33from other parts of Rakhine. So taking over Sittwe
- 13:38would mean whether the Arakanami is able to take
- 13:42over or whether the military is able to retain
- 13:44control. It doesn't look good for the civilian
- 13:48population there because whatever side wins or
- 13:51loses, there is going to be loss of life. And
- 13:58Situwe is difficult for the Arakanami to take
- 14:01over because it's surrounded by water on three
- 14:03sides and the Arakanami does not have naval capabilities.
- 14:09Same with Chaopiu. It's an island. it's going
- 14:14to be difficult and challenging for their economy
- 14:16to advance there because of lack of naval capabilities.
- 14:20These areas, experts told us, also has jamming
- 14:23technology. So our economy, they do have some
- 14:25drones, but it's possible that the military would
- 14:29jam them. And both Situ Wei and Chaopiao are
- 14:34very important because they have foreign investments,
- 14:38both India and China have. invested in infrastructure
- 14:41projects there. And if the Rakanami is able to
- 14:45take over both those regions, it would also give
- 14:47them some diplomatic advantage, you know, because
- 14:51then this is an arm -glue which will be able
- 14:53to negotiate potentially with governments of
- 14:57neighboring countries. So lots of important issues
- 15:05happening in Rakhine and we'll see what happens
- 15:08next year. If the EA is able to take over, it
- 15:11would be the first armed group that has taken
- 15:15over a capital. When I say it weighs the capital,
- 15:18it's the capital of Rakhine. It's the capital
- 15:20of the state of the country. Yeah, so that was
- 15:25mainly the research and my colleague and co -author,
- 15:30Miguel. made a scroll for the story just to explain
- 15:33what's happening and show visually how the areas
- 15:36look. And even if a certain area has a destruction
- 15:44which is large enough, we don't need high -resolution
- 15:49satellite image. You can still tell with mid
- 15:51-resolution satellite image. Google Earth, in
- 15:54this case, was very useful because Google Earth
- 15:57had a lot of imagery. The timeline was big enough.
- 16:02It was five years. So Google Earth also had a
- 16:05lot of imagery that we could rely on. And the
- 16:09story also includes images from Google Earth.
- 16:13That was quite a long brief for the story. No,
- 16:18I appreciate it. It's really useful to hear it
- 16:21from your perspective and hear the methodology
- 16:23as well and the obstacles that you've faced.
- 16:26I've put in the chats, the links that were mentioned,
- 16:29including the link to yesterday's article, the
- 16:332024 article that was mentioned, ACLED, NASA
- 16:38firms, which you mentioned. And I've also put
- 16:40in links to spaces within this server where you
- 16:44can talk about some of the things mentioned,
- 16:47including the specific munitions and the region.
- 16:51So please do check out those spaces if you want
- 16:53to continue the conversation here. I wanted to
- 16:56start off the question and answer. portion of
- 16:59this talk. And again, please feel free to put
- 17:03in the chat if you have any questions for Pooja.
- 17:07I wanted to ask what makes reporting and verification
- 17:10particularly difficult in that region? You wrote
- 17:13in the article that evidence is often slow to
- 17:16emerge. How does that delay and affect verification
- 17:23workflows and public understanding? Yeah, so
- 17:29I think accountability becomes difficult because
- 17:33there's only so much you can do with open source
- 17:35reporting. The Human Rights Watch recently published,
- 17:39I think last week published a report on what
- 17:42happened in this village in Rakhine, which is
- 17:46in Bhutidong. I'm not going to try to pronounce
- 17:48the village, just the name of the village. But
- 17:51according to Human Rights Watch investigation,
- 17:58170 Rohingya men, women and children were killed
- 18:01by the Rakanami last year in April, May. I briefly
- 18:05spoke about what happened in Bhutanang in April,
- 18:08May last year. And this village was also part
- 18:13of at least 40 villages that were targeted then.
- 18:17And I mean, we knew what had happened. We knew
- 18:20that the first attack The township was first
- 18:24attacked by the military and Rohingya militants.
- 18:26And later on, in retaliation attacks, the Arabian
- 18:29army just burned down large parts of the township.
- 18:36But there was denial from the Arabian army that
- 18:42they did not target civilian areas. They said
- 18:44that they targeted military soldiers and they
- 18:49targeted Rohingya militants. But it took more
- 18:53than a year for this report to come out because
- 18:58people who were displaced, the Human Rights Watch
- 19:01spoke to them in Bangladesh when they reached
- 19:05the shelters. Lots of Rohingya asylum seekers
- 19:09are in Bangladesh. So they did ground reporting,
- 19:13they spoke to witnesses, and then they were able
- 19:15to... There's satellite imagery of the village
- 19:18that shows that... shows that it has been attacked.
- 19:24But there were no confirmed, you know, witness
- 19:27reports. And so it takes a lot of time for accountability,
- 19:32some sort of accountability to happen. And that
- 19:35is, like I said, so only so much you can do with
- 19:37open source reporting. The way a village or a
- 19:43town is attacked gives some ideas, idea about
- 19:46who is behind the attack. And there is some reporting
- 19:52that happens by, you know, tiny media organizations
- 19:56that still exist. But a lot of journalists from
- 20:00Myanmar are in exile and a lot of them operate
- 20:04from Thailand. So most journalists or a lot of
- 20:08journalists that are reporting on Rakhine or
- 20:10in Myanmar or sorry, on Myanmar, they're doing
- 20:14it not from the country. And so reporting just
- 20:19takes time. And it's not an area which is covered
- 20:22a lot by satellite imagery. So the reports are
- 20:27sort of delayed. What is happening right now,
- 20:30you may be able to report on it a few months
- 20:33from now, once the satellite images are more
- 20:36updated. But I think one good way to report on
- 20:42it is to, again, look at the overview. of what's
- 20:46happening by looking at data from Accolade. Like
- 20:51how I would research Myanmar, Accolade would
- 20:55do it in the same way. Like refer to news reports,
- 20:58refer to social media and try to collect the
- 21:00data and give it in a more organized way. There
- 21:04aren't links to the reports, but you can look
- 21:08them up. They have like description of what happened
- 21:11in a certain village. So it will give you an
- 21:12idea of what's happening. But yeah, it's a difficult
- 21:16region to report on just because it's under -reported
- 21:21and it's not covered as much by satellite image.
- 21:26But if you want to, say, report on what happened
- 21:31to the villages that were destroyed during the
- 21:34Rohingya genocide in 2017, Google Earth would
- 21:38be good enough for that reporting. You can go
- 21:41and look at imagery on Google Earth and try to
- 21:44see what happened to those villages. Is there
- 21:46redevelopment? Is there barren land? Like I said,
- 21:53in Situwe, the military cleared villages to build
- 21:57defense installations. Are there defense installations
- 22:01in those villages that were destroyed back in
- 22:042017? What do they look like now? Are they still
- 22:07on maps? So lots of things that you can do, going
- 22:11back in time and seeing what is the condition
- 22:14right now. So I think one way to report on it
- 22:17is just it requires a bit of, I think, patience.
- 22:20It's because you don't have as much as you would
- 22:24want in terms of satellite image. But there are
- 22:27other avenues. And sometimes a disruption is
- 22:30large enough to. look at mid resolution or low
- 22:32resolution, which is more updated satellite image
- 22:35than high resolution. I guess the difficulty
- 22:38of reporting on this region really highlights
- 22:41why it's important to report on this region because
- 22:44there's probably less eyes and less people attempting
- 22:49to do it. And that doesn't mean that these areas
- 22:53don't deserve that kind of investigation. In
- 22:57regards to the data that you've been talking
- 23:00about, we had a question earlier about removal
- 23:02of data. You mentioned that one of the townships
- 23:06you didn't have data for, was that data removed
- 23:11and what kinds of evidence tend to disappear
- 23:14fastest from your experience? What should researchers
- 23:18prioritize preserving when researching Rakhine
- 23:23State and in other areas in Myanmar? The data
- 23:27for the township that was not available, it's
- 23:31called, the approximate pronunciation is Manong.
- 23:35It's a very small island and I don't think it
- 23:38has a lot of villages, very few. So because it's
- 23:41an island, it's not covered that much by satellite
- 23:43image. There's less reporting. It's kind of,
- 23:47it's sort of isolated. And because of how small
- 23:51it is, I think that's probably why the data is
- 23:53not available. When I say data not available,
- 23:55I mean, actually didn't have data for that township.
- 23:58It's possible, multiple reasons, maybe nothing
- 24:01actually happened there. That's why the data
- 24:02is not there. Or, you know, like, it's just not
- 24:07there because it's kind of isolated. I don't
- 24:11think there was any possible removal of data
- 24:15in that township. It's just... It's just not
- 24:18there. It's very small. Your second question
- 24:22about what should be preserved? I think one thing
- 24:28that I noticed when I was doing this research
- 24:30is I was looking at the township maps that are
- 24:35released by the UN and I was looking at the 2024
- 24:38maps. And I did want to do a comparison between
- 24:45how the map maps looked whenever they released
- 24:50the previous map. Probably, I think they released
- 24:53it in 2021. But I couldn't find those maps. So
- 24:58right now, the 2024 maps are there. It would
- 25:00be a good idea to archive them, save them locally,
- 25:04so that when the new maps are released, you can
- 25:07then refer to the new map and see what is the
- 25:10change in those two maps. And then that might
- 25:14be given an idea. of what has been, the villages
- 25:19that have been removed. And then you can see
- 25:21that, okay, what happened here? Why are these
- 25:24villages removed? And it would be a fairly simple
- 25:28investigation, right? Like if you have names
- 25:30of all the villages from 2024, and then you have
- 25:34names of all the villages from 2021, you can
- 25:35just compare and see what names do not appear.
- 25:38And then you can go and check what happened to
- 25:41those villages. So I think... That is something,
- 25:44there are some previous maps that are available,
- 25:47you know, like research gate or like some other
- 25:49websites like that, but it's not, you can't search
- 25:52them on what is the name of the website. It's
- 25:56called the MIMO, the Myanmar information management
- 26:02unit. I mean, I can send the link to you. This
- 26:05is where you can look at all the maps, township
- 26:09maps, but they don't have. it's like a PDF of
- 26:13a map. So you still have to like see, like open
- 26:20up Google Earth or something, or like plan it
- 26:22and try to see where you're looking at. And if
- 26:26you have the tech skill of converting, which
- 26:29Miguel did for my story, is if you can convert
- 26:34the PDF into some sort of GeoJSON or something,
- 26:38and then try to like open it up on QGIS or some
- 26:43other software that lets you open GeoJSON files,
- 26:49then you could potentially use the PDF maps to
- 26:53just get the coordinates. How I was doing it
- 26:58was very, very time consuming. I was like, okay,
- 27:01so the main town is here and then there is a
- 27:03village here. And I was like tracking where the
- 27:05river is flowing and then seeing, okay, it's
- 27:07near this... you know, this village name appears
- 27:11next to this river. So I was like going on Google
- 27:13Earth and trying to find the river. And it was
- 27:15very time consuming, but you could make it easier
- 27:18for yourself if you have some tech skills. Um,
- 27:22and, uh, yeah, but I guess do archive the maps
- 27:25and save them locally if you're interested in
- 27:28Myanmar. Get archiving folks. Uh, thank you for
- 27:32that tip. Really important to highlight. you
- 27:36know, where the gaps in data are and what could
- 27:38possibly be important that you might not think
- 27:41about immediately. Your piece can combine satellite
- 27:45imagery, firms, fire data, accurate data, social
- 27:48media, local reporting, as you said. How do you
- 27:51decide when there's enough cooperation to confidently
- 27:54identify a village as completely destroyed? Like,
- 27:59how do you make that editorial call that the
- 28:04villages that you're reporting on are completely
- 28:07wiped off the map, other than their erasure from
- 28:10these government maps. How do you visually tell
- 28:13when something is, how do you visually and confidently
- 28:18say that these villages have been impacted that
- 28:21badly? Yeah, there's no confident way to say
- 28:26unless something has been officially removed
- 28:28because that it's possible that a village was
- 28:34destroyed and then rebuilt later. Unless you
- 28:39have the latest available satellite image, you
- 28:42don't exactly know what happened there. For a
- 28:46lot of the villages, we did have latest satellite
- 28:49image, so we were able to tell that currently
- 28:52these villages are destroyed. But I think I've
- 28:55written in the story as well that At least in
- 28:57five villages, we saw that there was signs of
- 29:01reconstruction. It's difficult to tell whether
- 29:04the reconstruction was for civilian purposes
- 29:07or otherwise. They didn't look like defense structures.
- 29:10They did look like buildings. But again, it's
- 29:16tough. It's tough to exactly say what happened
- 29:19somewhere, unless you have latest satellite image.
- 29:24And how do you detect something? that looks destroyed.
- 29:29I can share my screen. Let me share my screen.
- 29:35While you're getting that up, perhaps also while
- 29:37you're sharing the screen, it would be really
- 29:39interesting to understand, maybe for people who
- 29:41aren't used to highlighting when military fortifications
- 29:47have appeared in Saas -Wa -Imji, what indicators
- 29:51in Saas -Wa -Imji tell you that a civilian settlement
- 29:53has become a military position as well? that
- 29:55might be something interesting to look at while
- 29:57you're, while you're sharing your screen, if
- 29:59that's possible. Yeah. Let me, let me talk about
- 30:01the military. That's a good question. Like if
- 30:03you Google like say trenches from trenches satellite,
- 30:09because Myanmar is not unique in this. A lot
- 30:13of areas have military installations and structures
- 30:17built, right? So if you just like familiarize
- 30:21yourself with like how military installations
- 30:25would look like then you when you look at a satellite
- 30:27image you're like okay you know that that is
- 30:29probably what I'm looking at like for instance
- 30:32like this those are like sort of like trenches
- 30:36and this is what it would look like on satellite
- 30:39image like zig zigzag lines or you know like
- 30:47something like this has it too like something
- 30:52which look has like some sort of a pattern and
- 30:57When you're looking at like I think all of us
- 30:59have looked at it looked at you know how a place
- 31:04looks which has like some sort of a Human settlement
- 31:08right there would be houses there would be buildings
- 31:10When that is replete by what you know odd patterns
- 31:14you know that something happened there were clearance
- 31:16operations or like something has happened there,
- 31:18but the easiest like way to familiarize yourself
- 31:22is just like Google and see how other people
- 31:25have reported on it. I didn't know what trenches
- 31:27looked like. And I wasn't sure what I was looking
- 31:31at. So this is what I did. And I saw that, OK,
- 31:35if there are structures like that, these are
- 31:37probably like military installations. And then
- 31:40you will also be able to see bunkers and things
- 31:42like that. So in this story, I'm just going to...
- 31:50So this is Sitwe. This entire region is Sitwe,
- 31:57the township. And this is Sitwe city, which is
- 32:00the capital of Rakhine. And lots of villages
- 32:04around the city, we saw that there were some
- 32:08sort of clearance operations and they were replaced
- 32:10by some of the villages were replaced by defense
- 32:16installations, like this one. This village, this
- 32:20is what it appeared intact in March 2024 Google
- 32:24Earth image. And then later on, we saw that it
- 32:29was destroyed. Now, another thing, when you're
- 32:34looking at satellite image and you can see buildings
- 32:38have been destroyed, you can sort of tell the
- 32:41method used to destroy. This does not look like
- 32:45some sort of aerial 500 pound bomb dropped because
- 32:49then the vegetation would also be destroyed.
- 32:51It would look more burnt. Here, you can see that
- 32:55specifically the structures itself were destroyed.
- 33:00These buildings have been destroyed. It almost
- 33:03looks like they were destroyed one by one. You
- 33:07can see that the nearby vegetation does not look
- 33:09impacted. So you can tell that, okay, maybe this
- 33:13was like targeted some sort of clearance operation.
- 33:16So this is from 2025 Google Earth image. And
- 33:19if you notice here on the side here, there is
- 33:24nothing right now in 2024. And when I'm scrolling
- 33:27to 2025, there is some sort of those exact lines
- 33:31have appeared. So next to the village, there
- 33:34has been some military installation. And this
- 33:38is again from Google Earth, we saw trenches and
- 33:43outposts being built. So something like this,
- 33:46if you look at satellite image, this does not
- 33:50look like a civilian development, development
- 33:54for civil populations. So anything that looks
- 33:57odd or stands out, looks kind of barren, where
- 34:00there's some structures there, that could be
- 34:02a hint that there has been some defense installations
- 34:05there. And some of this you can also tell by
- 34:09looking at low resolution satellite image. Again,
- 34:11if it's large enough. This is low resolution
- 34:14satellite image from planet. And you can see
- 34:18that this village, you can just see vegetation.
- 34:22If you were able to see buildings, you would
- 34:25see like tiny white spots or something like that.
- 34:28But nothing is visible in this village. So it's
- 34:32a hint that it's destroyed. And next to them,
- 34:36you can see some sort of lines built around those
- 34:40villages. And that is also a hint that, okay,
- 34:44there's some structures that has been built there.
- 34:47So it's possible even with low resolution satellite
- 34:50image to be able to tell sometimes that there
- 34:53has been some construction that the military
- 34:58has done. This is another village. And here the
- 35:01village is almost replaced by defense installations.
- 35:03You can see those zigzag lines on the side of
- 35:08the villages. Instead of next to the village,
- 35:12you can almost see that they are on top of the
- 35:14village. So this is low resolution image, satellite
- 35:20image, which compares two villages. This one,
- 35:23which had the zigzag lines, and this one, which
- 35:27does not. So you can also compare, even with
- 35:30low resolution image, which villages have had
- 35:34military development versus which hasn't had.
- 35:37So even if you don't have access to high resolution
- 35:39image, you can get some hint even with low resolution
- 35:43image. Like this looks really different from
- 35:46this village. This almost looks like barren land
- 35:49with some greenery in it. But this one, you can
- 35:53tell there are lines there. So it's possible
- 35:55that there has been some development there. So
- 35:58that is one way to detect what has happened after
- 36:05a clearance operation happens, after human settlements
- 36:07or civilian areas have been cleared or removed
- 36:10or destroyed. This is another village where trenches
- 36:13you can see are like all over the circumference
- 36:18of the village. Okay. Hope that was helpful.
- 36:25Really clear. And I just wanted to shout out
- 36:29Miguel again, because even as a presentation
- 36:32tool, that scrolly telly is super useful. And
- 36:35if you're listening back on the podcast, everything
- 36:40that Pooja just showed in this talk can be seen
- 36:45and scrolled through by yourself on the article
- 36:48that will be linked in the description. So do
- 36:52check out there for some of those examples of
- 36:54fortifications that she was just talking about.
- 36:57We've been very careful in the article not to
- 37:01attribute blame, but are there methods and open
- 37:07source ways of identifying which potential armed
- 37:13group could be responsible for some of these
- 37:16attacks? Obviously, in terms of, we're not suggesting
- 37:21you name them right now, but As there's multiple
- 37:26armed actors operating in the same area, how
- 37:28could you possibly identify who's at play in
- 37:33these particular regions or states? Right. So
- 37:38which armed group is going to be difficult? I
- 37:43haven't looked at the whole country. If you're
- 37:47looking at Myanmar, it's It's tough to report
- 37:54on the entire region because there are so many
- 37:58ethnic groups fighting the military. So I have
- 38:03been focused on Rakhine. I would not be able
- 38:06to talk about the rest of the country. If there
- 38:09are large patterns of villages destroyed or the
- 38:15military also destroys religious establishments
- 38:18like... There is a state called Chin State, which
- 38:20has a lot of churches, and the military has been
- 38:23destroying churches as well. So if it's a broad
- 38:27overview story like that, sure, there are some
- 38:33hints that which group might have attacked a
- 38:36certain area. You can check if the ethnic armed
- 38:40group that you're looking at, whether or not
- 38:42it has air power. If an area looks like it has
- 38:48been destroyed by a bomb or something, if you
- 38:52can see like a crater or something, you can tell
- 38:55that, okay, this armed group does not have air
- 38:57power, so potentially the military did this attack.
- 39:01So that's one way of going about it. And armed
- 39:06groups generally, like this is again... This
- 39:10is not always true. We generally don't attack
- 39:13civilian areas, at least in Rakhine. Their fight
- 39:17is with the military. So if a military outpost
- 39:22has been attacked, they attack police stations
- 39:24as well. You know that the attack was by the
- 39:27armed group. Now there are exceptions to this,
- 39:30of course, because like I said, the Rohingya
- 39:32population in Rakhine has been attacked both
- 39:36by the military and by armed groups. The type
- 39:39of attack is a good way to tell what has happened.
- 39:45You can check if a particular township currently
- 39:48is controlled by the military or is controlled
- 39:52by the armed group, then you can tell who did
- 39:56the attack. Because an expert also told us that
- 40:00the military attacks villages or towns that it
- 40:03has lost. leave an ungovernable region for the
- 40:09armed group. You know, it's sort of like you
- 40:12want to take this land from us, go ahead and
- 40:14do that. You are not going to be able to govern
- 40:16it because there's nothing left. It's bad enough
- 40:18and there is no people left. So what are you
- 40:20going to govern? So those are some hints into,
- 40:25you know, telling what happened in a region or
- 40:29in a specific area. But due to lack of reporting,
- 40:33You cannot always conclusively tell what happened,
- 40:36but that doesn't mean there's no reporting coming
- 40:38out of Myanmar. There are smaller organizations
- 40:40that are doing ground reporting. Witnesses, when
- 40:44they, you know, human rights groups, when they
- 40:46talk to them, you can tell what exactly happened.
- 40:51Large scale destruction most likely would be
- 40:54military of, you know, civilian areas. So those
- 40:58are some things that give a hint about who may
- 41:00have carried an attack. So it's really about
- 41:04scale and also weaponry used, which is really,
- 41:08really useful to note. I've also linked in the
- 41:12chat your other report from last year on this
- 41:16area following the earthquake ceasefire. So you
- 41:20can have a read of that as well. We've got about
- 41:2415 minutes left for questions, so please pop
- 41:27them in the chat if you have any. I wanted to
- 41:30quickly ask... Because this is a longitudinal
- 41:34analysis, you've done it over a five year period.
- 41:38Did you spot any significant patterns or perhaps
- 41:42escalation within the data sets? Was there more
- 41:49reported violence or more reported destruction
- 41:54over a certain year period in Rakhine State?
- 41:58Has it been kind of a steady process in all of
- 42:04these particular township? That's a good question.
- 42:11And there has been pattern in the violence because
- 42:17in late 2023 is when armed groups sort of restarted
- 42:22their offensive against the military. So you
- 42:24will see a spike. I can even share my, because
- 42:29I've been talking about and giving a shout out
- 42:30to Akhlet so much. And I think I should talk
- 42:37about how I used Akhlet. So there is also a chart
- 42:41in my story here, which shows how this is just
- 42:48the military's air and drone strikes in Rakhine.
- 42:52I've only done two years, but I looked at I think
- 42:55from 2022 to 2025 or maybe like April 2026. But
- 43:04you can see a clear pattern between 2024 and
- 43:072025. So late 2023 is when the offensive by armed
- 43:12groups against the military increased. So when
- 43:15any offensive increases against the military,
- 43:17the military also increases their attacks. So
- 43:20it will do more airstrikes and drone strikes.
- 43:22It will target more civilian areas. One of the
- 43:26experts also told us that it is the way of the
- 43:29military to tell the local population that if
- 43:32you support these armed groups, you are going
- 43:35to get attacked. Like armed groups sometimes
- 43:38can take shelters, you know, villages can provide
- 43:41you resources and things like that. So the idea
- 43:45is to destroy civilian areas so that the armed
- 43:48groups don't get shelter, don't get resources.
- 43:51So that's why the military attacked civilian
- 43:53areas. Now, if you see in 2024, in all of these
- 43:58areas, compared to 2025, there were a lot more.
- 44:06And the attacks then fell. These are military
- 44:08attacks. In all of these townships in 2025, they
- 44:12decreased. In almost all the townships, they
- 44:17decrease except one, this one where it's remained
- 44:19the same, barring two townships, which is Chaopiu
- 44:23and Sittwe, where the attacks increased in 2025.
- 44:28Everywhere else, they decreased in 2025. Now,
- 44:31why that has happened is because all of these
- 44:35other townships were already captured by the
- 44:37Rakanami. 2025. So the military has now shifted
- 44:42its focus to Chaopu and Sitwe. So now it's mainly
- 44:46attacking those two regions and so many attacks
- 44:49in Chaopu. Again, like if you read about it,
- 44:52Like I said, Chaopu has a lot of infrastructural
- 44:58projects. China has invested heavily in Chaopu
- 45:01and it has an oil and gas pipeline, which provides
- 45:04an alternate route to China to directly transport
- 45:08oil from Myanmar to China through the pipeline.
- 45:11So it's super important for China. China has
- 45:13also sent some of its soldiers to protect its
- 45:18infrastructure there. There has been a lot of
- 45:22attacks in Xiaopu, just because of its strategic
- 45:26importance. Sittwe as well, not a heavy increase,
- 45:30but at least according to ACLIN data, there has
- 45:33been an increase because, again, it's the capital.
- 45:38A lot of the military is trying to protect the
- 45:40capital. An army would get a lot of power if
- 45:44it controls the state capital. Charlie, can you
- 45:50see this? Can you see I'm sharing? Yeah, so this
- 45:56is like an actually data. If you have even basic
- 46:00Python, it's very easy to analyze because the
- 46:05data is very formatted. You don't have to do
- 46:11a lot of cleaning with the data. Or you can also,
- 46:15you know. wipe code and like look at chat GPT
- 46:18and it might help you with the coding. And this
- 46:21is the complete data for all those years. I only
- 46:28did 2024 and 2025, but if you see from 2022 to
- 46:322026, but 2026 of course is not complete data.
- 46:36You can see the pattern. You can see that it
- 46:39really increased in 2024 and then it dropped
- 46:42in 2025. And right now in 2026, it's not that
- 46:47much, because again, you only had three months
- 46:49of data. But even in 2026, you can then go and
- 46:52see where the attacks are concentrated. So I
- 46:55also did like, this is year -wise townships,
- 47:02what's happening in the townships, like how many
- 47:04attacks in particular townships. So data from
- 47:09places like Accolade can give you a good idea,
- 47:12even though of course new data is going to be
- 47:14exhaustive. It can give you a good idea about
- 47:18where fighting is concentrated, what areas, you
- 47:21know, I should be looking at. Why is there an
- 47:24uptick in violence in a particular year? And
- 47:28this is date -wise. So if you want, you can go
- 47:31even more granular. You can do like month -wise.
- 47:33Why in a certain month, the military attacked
- 47:36a certain region more? Was there uptick in daily
- 47:40attacks? You know, you can do... do a lot with
- 47:43it. And then you can look at, like I said, it
- 47:47has coordinates, it has village coordinates.
- 47:49So once you know that, okay, this certain village,
- 47:52there's reporting that it was attacked, maybe
- 47:56there's a lot of, maybe it was attacked, which
- 47:57villages were attacked multiple times, you know,
- 47:59through the years, maybe the same villages being
- 48:03attacked. So you can... then say that, okay,
- 48:06this village was attacked like five times since
- 48:092022. So there's lots that you can do with it.
- 48:12And I also relied a lot on it just to get the
- 48:17research started because I didn't know, okay,
- 48:19where do I start? It's a huge region. How do
- 48:22I get all of these villages that have been partially
- 48:25or completely destroyed? So yeah, I would say
- 48:30for anyone trying to start, research in a new
- 48:33territory, I think Accolade is a good place to
- 48:36start because it gives you a good overview of
- 48:39what is happening. We've got a few questions
- 48:43in and we're running out of time, so I wanted
- 48:45to quickly go through them. You mentioned just
- 48:47then that China has investments in some of the
- 48:51townships. We had a question from Alt Grade Tab
- 48:54asking, Is there any way of knowing if Chinese
- 48:57resources have been used in these attacks? I
- 49:00know you haven't particularly covered the munitions
- 49:03found, for example, at the scenes, munition remnants
- 49:07found at the scene of these attacks, but is there
- 49:09any indication that there's foreign influence
- 49:13in regards to this particular conflict? Yeah,
- 49:18for sure. I mean, munitions is not my expertise.
- 49:22I personally get super overwhelmed when I'm looking
- 49:24at munitions. I'm like, I don't know where this
- 49:25came from. But other organizations like CIR,
- 49:29they do a really fantastic job on reporting on
- 49:33Myanmar. They have this initiative called Myanmar
- 49:36Witness, and the team reports a lot on different
- 49:40kinds of issues. And I think they did a report
- 49:43on, I think, Russian planes in Myanmar, if I'm
- 49:49not wrong. So you could report on that because
- 49:52the military has air power. How do they get fighter
- 49:57jets? Who's supplying them fighter jets? There
- 50:00are Facebook news organizations, like I said,
- 50:04that exist on Facebook. While you won't have
- 50:08visuals of the attack itself, you might have
- 50:11visuals of the aftermath and there would be remnants.
- 50:15And that could tell you where a specific munitions
- 50:19came from. But again, I haven't personally looked
- 50:23at the munitions and this specific topic, but
- 50:30there are other organizations that are reporting
- 50:33on it. And there is Chinese influence. And like
- 50:36I said, in Chaopu, because of the Chinese infrastructure
- 50:40projects there, China also sent their mercenaries
- 50:43there to just protect the area. Yeah, so foreign
- 50:48influence happens in most conflicts and in this
- 50:54one as well, there is some munitions and air
- 50:59power that the military has gotten from external
- 51:03parties. Thank you. If you are interested in
- 51:10digging into munition types and want to learn
- 51:13a little bit more, we actually did have a stage
- 51:16talk a couple of months ago now with Bellingcat
- 51:19researcher Trevor Ball, who was talking specifically
- 51:22about identifying munitions in the war with Iran,
- 51:25however gave really interesting and thorough
- 51:29talk on ways to identify different munition types
- 51:33and who might be using them, including a full
- 51:37list of resources for that kind of work. So again,
- 51:40we also have the space within the server. bombs,
- 51:45arms, drones, and other killing machines. And
- 51:48within there, there's loads of experts on munition
- 51:51types and you can learn a lot by living in that
- 51:55particular space. So please do check both of
- 51:57those resources out, both the stage talk by Trevor
- 52:00and also the space within the Discord. MatthewsNoise
- 52:05asked, there's a lot of GPS jamming in the region.
- 52:08Do you use flight tracking in your reporting
- 52:11and has the jamming affected that? Have you ever
- 52:15thought about using fight tracking in this particular
- 52:17state? So there is a lot of scope to be able
- 52:22to report on what kind of arms are being used
- 52:25in Myanmar. And it doesn't specifically have
- 52:27to be the arms that the military is using. You
- 52:30could pick a region and then see what is happening
- 52:34in that region and what kind of arms a specific
- 52:38ethnic group is using against the military. So
- 52:40that could also be a way to go about it. about
- 52:44flight tracking, no, I will not be able to speak
- 52:47about that because I haven't done it, no. Somebody
- 52:51has asked, what about funding using drug trafficking?
- 52:55Is there any reporting on that? I've not come
- 52:58across anything, but have you seen anything on
- 53:00that particularly in this region? Yeah, no. I've
- 53:07read a lot of reporting on the scam centers in
- 53:09Myanmar that exist in the border regions, specifically
- 53:13with Thailand, but no, I haven't looked at drug
- 53:18trafficking, unfortunately. In regards to this
- 53:23region, and maybe this might be our last question
- 53:27because of the time, but we'll see. In regards
- 53:30to this region, what developments in Myanmar
- 53:34are you watching most closely right now? Now
- 53:38your investigation is out, where would you have
- 53:40your eyes going forwards? I would still focus
- 53:46on Rakhine. Like I said, it's a big country.
- 53:52It's tough to report on the different states
- 53:54just because they have such different issues
- 53:58and things going on. Their economy is made up
- 54:02of majority Rakhine people. And the fight in
- 54:06Rakhine is unique to the geography of the state
- 54:09and to the politics of the state, to the diversity
- 54:14of the state. A different state would have different
- 54:17issues. So it's tough. to hop from one state
- 54:20to the other, because it's a very ethnically
- 54:23diverse country. So because I have focused on
- 54:27Rakhine, I think I will continue to focus on
- 54:29Rakhine, also because the AA has vowed to take
- 54:33over the state next year. It's going to be challenging,
- 54:38but would be interesting to look at how the fighting
- 54:42will progress. And if the military does any counteroffensives,
- 54:48The story that I did is still a very small overview
- 54:52of what has happened. There are thousands and
- 54:55thousands of villages. And while I found only
- 54:57115, so there is so much scope to still report
- 55:01on. And it's not just that the villages are bombed
- 55:05and destroyed, but what happens after that or
- 55:09what used to exist before that. And there's a
- 55:13lot that can be reported on. suggestions about
- 55:17looking at munitions or flight tracking and all
- 55:20of that, which I haven't looked at myself yet,
- 55:26but it could be something to look at as well.
- 55:30So yeah, my focus is still going to be Rakhine.
- 55:32I did get a follow -up idea from this story.
- 55:36It would be premature to discuss it, but yeah,
- 55:41definitely going to focus on Rakhine. Keeping
- 55:44your cards close to your chest. I appreciate
- 55:47that. Yeah, in terms of digging into the, also
- 55:54into the erasure of those villages on the maps
- 55:58as well is really interesting. And making sure
- 56:01that what you found already has been archived
- 56:05and saved somewhere is also incredibly important.
- 56:08To close, if anybody in this talk or listening
- 56:12back on the podcast is interested in reaching
- 56:16out to you, perhaps they know a little bit more
- 56:19about Rakhine State and want to give you some
- 56:23information. Perhaps they are interested in looking
- 56:27into the region themselves and want to learn
- 56:30a little bit more from you. How can people reach
- 56:32out and connect with you? You can DM me on Discord.
- 56:39You can email me. I tend to be more active on
- 56:42email. If you want to email me, it's puja at
- 56:45bellingcat .com. I would be very happy to help
- 56:52anyone with their research in Myanmar. While
- 56:55I have focused on Rakhine, the methodology can
- 57:02be used in other areas as well. We can chat about
- 57:06the different avenues there are to look at the
- 57:11region. Even if you don't have access to Planet
- 57:16or something, you can still report on Myanmar
- 57:20because again, it's an under covered area. So
- 57:25it's not like Planet has a large amount of high
- 57:29resolution image. You can do a lot with key satellite
- 57:35tools which are available for everyone. And yeah,
- 57:39I would be very, very happy to help anyone who
- 57:43wants to report on Myanmar. Thank you for that.
- 57:47And thank you for the talk today. As I said earlier,
- 57:50we do have a space in the Discord that needs
- 57:53a little bit of love actually. It used to be
- 57:56pretty popular and it hasn't been commented in
- 57:59for a little bit. So if anybody attending this
- 58:02talk wants to go into the space. Myanmar, Burma
- 58:07in brackets and to talk about some of the reporting
- 58:10just mentioned. Please do. Please jump in there
- 58:14and continue to populate it because we really
- 58:17want to keep focus on. these under -reported
- 58:20regions. And thank you again, Pooja, for doing
- 58:24this talk. We'll be back in two weeks' time with
- 58:27a stage talk with CNN. Keep your eyes peeled
- 58:31on announcements and the events tab for more
- 58:34information about that. But for now, thank you
- 58:37very much, everybody. Thanks, everyone. Thank
- 58:42you for listening to the stage talk. If you'd
- 58:45like to catch a stage tour live where you can
- 58:48ask the guest questions, join the Bellingcat
- 58:51Discord server by visiting www .discord .gg slash
- 58:57Bellingcat. The music you've heard is titled
- 59:00Dawn by Newer Self and is courtesy of Artlist.