This report uses nighttime lighting data to track the relocation of Rohingya refugees from the Cox's Bazaar region of Bangladesh to Bhasan Char, a 'floating island' in the Bay of Bengal built by the Government of Bangladesh to house roughly 100,000 refugees. Constructed on reclaimed land in a monsoon, cyclone, and tsunami-prone region, Bhasan Char poses not only humanitarian challenges, but physical and human rights challenges as the Rohingya become isolated from the mainland.
This report demonstrates how nighttime lighting data serves as a useful independent measure of the Government of Bangladesh's efforts to relocate Rohingya refugees onto the island over time, which could prove beneficial to those involved in humanitarian and disaster resilience and response planning over the coming months and years.
GEOINT analysis of nighttime lighting and other open data sources shows that refugee facilities in Cox's Bazaar continue to grow despite efforts to transfer more refugees to newly constructed facilities on Bhasan Char. Independent nighttime lighting analysis affirms public estimates of the refugee population on Bhasan Char at around 19,000 and reveals that semi-isolated housing plots on the island may be housing at-risk minority populations within the Rohingya based on previous safety concerns at Cox's Bazaar.
The Rohingya people are a stateless Muslim minority group that have historically resided in the northern Rakhine state in Burma (Myanmar), along the Bay of Bengal in the northeastern Indian Ocean. They have suffered frequent persecution at the hands of Burmese government forces over the last several decades, spurring periodic waves of Rohingya refugee flows across the border into the Cox's Bazaar region of neighboring Bangladesh.
Two prior waves of persecution in the late 20th century forced hundreds of thousands of Rohingya refugees into refugee camps in Cox's Bazaar district in 1978 and 1991-1992. In the years that followed, the vast majority of these refugees returned to Burma under a series of repatriation agreements between Bangladesh and Burma. A third major wave of violence in 2016 forced roughly 87,000 Rohingya refugees into the Ukiya Upazila portion of Cox's Bazaar, where many still reside to this day. By 2017, before the start of the most recent Rohingya crisis later that year, roughly 200,000 Rohingya refugees already resided in refugee camps in Bangladesh.
Figure One. Key Locations in Burma and Bangladesh
In August 2017, Burmese government forces began a new wave of extreme violence against the Rohingya throughout northern Rakhine state, committing atrocities against local villagers in what the United States has since deemed a campaign of ethnic cleansing. These events initiated the largest migration of Rohingya refugees to date, forcing approximately 725,000 people to flee Rakhine state across the border into Bangladesh. The majority of these refugees, roughly 670,000, arrived at Cox's Bazaar in September 2017, with the remainder arriving over the next three months. Many of these individuals settled into the vicinity of former refugee camps that had been built to serve previous displaced Rohingya populations in the 1990s.
Despite these concerns, construction on Bhasan Char would surge throughout 2018, with plans to begin resettling Rohingya refugees from Cox's Bazaar to the island later that year. Ultimately, it would take until January 2020 for the Government of Bangladesh to declare Bhasan Char ready for habitation. By May 2020, the first Rohingya refugees were resettled into facilities on Bhasan Char after they were first barred entry into Malaysia at sea, and rescued by Bangladesh's Coast Guard. Subsequently, the first large-scale refugee transfers to Bhasan Char occurred in December 2020, when 3,400 Rohingya were relocated from Cox's Bazaar onto the island. Initial reports suggested that Rohingya Christians were the first to be relocated, in part because these and other minority groups within the broader refugee population had faced frequent threats of violence, including kidnappings, within the crowded camps of Cox's Bazaar.
Bangladesh's continued transfer of refugees onto Bhasan Char also poses challenges for the international community as it continues to provide support to vulnerable Rohingya refugees throughout Bangladesh. Continued aid will increasingly need to be split between disparate locations in Cox's Bazaar and Bhasan Char, which are roughly 150 kilometers apart by sea. Beyond these logistical challenges, future catastrophic weather events impacting Bhasan Char could require a large-scale civilian evacuation or disaster response effort that likely exceeds the capacity of the Government of Bangladesh. As such, the rate of transfer of refugees to Bhasan Char is a critical factor for international humanitarian organizations, government agencies, and even military planners who are likely to play a role in future disaster response efforts.
Methodology and Data
In the past, RAND researchers have used nighttime lighting data to track changes in levels of activity at key infrastructure in remote or denied parts of the world, often where other sources of data are unavailable to provide reliable insights. In Iraq and Syria, RAND used nighttime lighting data to understand changes in economic activity in cities taken over by the Islamic State. In western China, RAND has used nighttime lighting data to track China's efforts to reeducate and imprison its Uyghur and other ethnic Muslim minority populations in rural Xinjiang Province.
In this report, we use nighttime lighting data to track the relocation of Rohingya refugees from Cox's Bazaar to Bhasan Char. This approach allows us to measure changes in the intensity of activity at refugee camps that may not be visible in overhead imagery. As in past research, we calculate nighttime lighting estimates over specific locations using monthly cloud-free composite raster images from the National Oceanic and Atmospheric Administration's (NOAA) Visible Infrared Imaging Radiometer Suite (VIIRS) Day/Night Band, made available via the Earth Observation Group at the Colorado School of Mines.
To assess nighttime lighting trends over specific refugee camps in both Cox's Bazaar and Bhasan Char, we first reproject each composite nighttime lighting image from its original resolution (15 arc-seconds, or approximately 350 meters at 40-degrees latitude) to a finer-scale 30-meter resolution using simple linear interpolation. This improves our ability to accurately measure nighttime lighting for smaller refugee camps and helps to reduce potential bias from adjacent infrastructure. Then, we produce monthly trendlines over each refugee camp that represent the raw nighttime lighting value from the NOAA VIIRS sensor averaged over each facility. We then normalize these measurements over each camp to their nighttime lighting value in February 2018, improving our ability to make meaningful comparisons of the relative growth of specific locations over time and across similar locations with differing baseline levels of electrification.
In prior research, we have adjusted these normalized raw values using moving averages to smooth out month-to-month variation in the relatively noisy VIIRS data. In this case, however, given the relatively small size of each refugee camp and their close proximity to adjacent camps, we choose not to artificially smooth these values to avoid potential bias in our interpretation.
The resulting dataset produces a noisy yet informative assessment of the timeliness of periods of major construction and electrification at key refugee camps in both Cox's Bazaar and Bhasan Char. The next section reviews these findings, examining trends in the growth of facilities in each location and eventually estimating the size of the refugee population on Bhasan Char as of May 2021, the last date for which we have composite nighttime lighting data.1
The two largest camps in Cox's Bazaar are in northern Ukhia Upazila (number one in the map above), as well as southern Teknaf Upazila (number eight in the map above). These camps largely consist of sprawling, temporary structures that are loosely organized yet compactly built into the surrounding landscape. For context, Figure Two above also shows an image from November 2018, taken inside the Balukhali portion of the larger Kutapalong Refugee Camp in Ukhia Upazila. This image also shows the types of nighttime lighting likely to be captured by NOAA's satellite sensor, with solar-powered lamps inside the refugee camp lit up at night.
Figure Three below plots the average monthly nighttime lighting estimates over these two largest camps in Cox's Bazaar, the Kutapalong Refugee Camp in Ukhia Upazila (in blue), and the southernmost camp in Teknaf Upazila (in red). Given the temporary nature of most buildings in these camps, and the relative poverty of those refugees living in them, increased nighttime lighting over these facilities is likely to be a product of efforts by the Government of Bangladesh and the international community to build more permanent or resilient structures within these facilities, accompanied by solar or other sources of power.
Figure Three. Nighttime Lighting over Cox's Bazaar since 2014
These nighttime lighting estimates reveal several initial insights. First, despite the settlement of roughly 87,000 refugees into Ukhia Upazila throughout 2016, there were only marginal increases in nighttime lighting over the course of that same year. Similarly, nighttime lighting over both camps increased at a relatively modest pace in the months following the onset of violence against the Rohingya in late 2017. Major increases in nighttime lighting following these refugee flows, particularly in Ukhia Upazila, did not come until two periods in mid-to-late 2018 as well as throughout 2020 and early 2021. As such, we assess that these increases are likely the product of increased efforts to electrify certain portions of these camps, rather than major inflows of refugees.
Finally, and perhaps of most relevance to our efforts to track the relocation of Rohingya to Bhasan Char, this data reveals that nighttime lighting over Cox's Bazaar's two largest camps has actually continued to expand in recent years despite the increasing relocation of refugees out of these camps. This suggests either that refugee relocations have been marginal in scale so far, or just as likely, that efforts to electrify portions of Cox's Bazaar have continued (and are likely to continue) despite these relocations. This is logical to the extent that Bhasan Char's capacity is estimated to be around 100,000 refugees, just a fraction of the nearly one million Rohingya refugees currently living in Cox's Bazaar.
Nighttime Lighting over Bhasan Char
Unlike the ad-hoc camps of Cox's Bazaar, facilities on Bhasan Char were purpose-built by the Government of Bangladesh to house refugees on reclaimed, largely flat land. As a result, Bhasan Char's housing facilities were constructed across roughly 120 nearly identical housing plots arrayed across the island in a grid-like fashion, each containing twelve separate buildings and a storm shelter. Figure Four below shows the distribution of these plots across the island, as well as an image taken from the island showing the types of buildings constructed.
Figure Four. Refugee Facilities on Bhasan Char.
In previous research focused on Uyghur detention efforts in western China, RAND assessed through overhead satellite imagery that major spikes in nighttime lighting over key infrastructure were often correlated with periods of active construction of new buildings, particularly in areas without major preexisting infrastructure. Figure Five below plots the average nighttime over Bhasan Char since 2014, revealing similar trends.
Figure Five. Nighttime Lighting over Bhasan Char
These results suggest that the most elevated periods of nighttime lighting seen on the island occurred during a series of major spikes during the summer of 2018. This closely aligns with official statements from the Government of Bangladesh during that time period, which prematurely suggested that construction on the island would soon be finished. Later spikes in nighttime lighting would similarly occur in early 2020 and early 2021. These two time periods correlate well with other points on our timeline - including efforts by the Government of Bangladesh to finalize the island for habitation in 2020, and the early relocation of refugees from Cox's Bazaar beginning that same year and through 2021.
At least anecdotally, these results suggest that nighttime lighting may serve as a useful proxy for tracking the relocation of Rohingya refugees onto Bhasan Char over time. And yet, given the unique structure and layout of the facilities constructed on Bhasan Char, we can extend our methodology to attempt a more precise estimate of the number of Rohingya relocated to the island throughout 2021. Figure Six below provides a more detailed view of these facilities.
Figure Six. Bhasan Char Refugee Housing in Detail
Across the island, 120 separate plots of land are arrayed in a grid-like fashion. Open source estimates suggest that each plot contains 12 single-story buildings, each meant to house 16 families of four individuals, along with one larger storm shelter that houses 23 families. Each building's roof also contains solar panels to provide electricity. Altogether, this suggests that each plot can house up to 860 refugees, for a total of 103,200 individuals at maximum capacity across Bhasan Char's 120 separate plots.
With these numbers in mind, we can then use plot-specific changes in nighttime lighting within Bhasan Char to assess whether certain locations on the island have experienced more growth than others, and use these locations to estimate a more precise total population of refugees currently living on Bhasan Char.
As discussed earlier in this report, the major periods of construction on Bhasan Char occurred in 2018 and 2020, with initial refugee relocations onto the island reportedly beginning in 2020 throughout 2021. As a result, we focus on isolating those locations on the island which experienced outsized growth in nighttime lighting in the most recent months of our dataset (throughout 2021) relative to their average nighttime lighting in 2019, thereby avoiding potential bias from periods of construction in preceding and intervening years.
Figure Seven below shows all 120 housing plots of Bhasan Char, categorized by the level of growth in nighttime lighting over each housing plot as measured in 2021 compared to 2019. We break down this growth into four categories. First, we assess that 22 housing plots experienced major growth in nighttime lighting in 2021 (through May 2021, our data cutoff), defined as having greater than 65 percent growth between the maximum monthly nighttime lighting in 2021 relative to average monthly nighttime lighting across all of 2019. Second, we assess that 37 housing plots experienced moderate growth, defined as having between 45 and 65 percent growth between these years. Third, we assess that 32 housing plots experienced modest growth of between 25 and 45 percent. And finally, we assess that 29 housing plots experienced only marginal growth or even declines in nighttime lighting, defined as less than 25 percent growth between 2019 and 2021. Cut-points between each category of nighttime lighting growth were determined by identifying substantive breaks in the distribution of growth across all 120 plots.
Figure Seven. Nighttime Lighting Growth in 2021 on Bhasan Char, by Housing Plot
Several insights emerge from these results. First, major nighttime lighting growth in 2021 appears to have occurred in housing plots primarily along the southwest corner of the island, closest to the maritime port of entry for refugees arriving from the mainland, as well as the primary set of administrative and logistical buildings on Bhasan Char located on the western side of the island. These results are logical to the extent that the first tranche of relocated refugees would be housed closest to the infrastructure required to support their presence, rather than the far end of the island. Alternatively, elevated nighttime lighting over these housing plots could also be the product of adjacency effects resulting from nighttime lighting over this infrastructure spilling over into the measurements for those nearby housing plots.
And yet, a second key insight would suggest that these effects are indeed the result of relocated refugees moving into these housing plots and not just adjacency effects. Specifically, we see major nighttime lighting growth in early 2021 in a more isolated set of eight housing plots in the southernmost portion of Bhasan Char, as seen near the bottom of Figure Seven. These plots are isolated from the main body of refugee housing plots by a set of trees and connected only by one road. Recall from our earlier discussion that initial reports suggested that initial priority was given to transferring vulnerable minorities within the Rohingya population onto the island in 2020 and 2021, in part to minimize the risks of violence they faced within the crowded camps on the mainland. The growth in nighttime lighting seen in this isolated set of housing plots is consistent with a potentially deliberate effort to house these vulnerable populations in separate, more isolated facilities on the island.
Using these four categories of nighttime lighting in housing plots across Bhasan Char (major, moderate, modest, marginal), we can now attempt to estimate the population of Rohingya refugees likely already resettled on Bhasan Char as of May 2021, the most recent month in which our nighttime lighting data are available. Based upon the estimated 860-person capacity of each individual housing plot, and our finding that only 22 housing plots experienced major nighttime lighting growth through early 2021, we assess the likely minimum population of Rohingya refugees on the island as of May 2021 was roughly 18,920. This correlates almost precisely with the public estimate of 19,000 refugees offered by the Government of Bangladesh later that same year. Our estimates are summarized below in Table One.
Table One. Estimating the Refugee Population Resettled on Bhasan Char, 2021
Of note, we cannot assess with precision using nighttime lighting whether each of the 22 housing plots experiencing major nighttime lighting growth through May 2021 were housed to full capacity within each individual plot at that point in time. And yet, these findings suggest that nighttime lighting growth on Bhasan Char may be a closely lagging or even leading indicator of eventual inhabitation of specific housing plots by relocated refugees. Given that the island is primarily run on solar power and diesel generation, it is logical that such power would be prioritized to those areas with refugees present or soon to arrive, and not to those housing plots which remain empty - particularly at night.
To that extent, it is worth noting that the additional 37 housing plots that experienced moderate nighttime lighting growth through May 2021 have a combined capacity to house an additional 31,820 refugees, primarily in the center of the island. Similarly, 32 housing plots that experienced more modest nighttime lighting growth in early 2021 have a combined capacity to house an additional 27,520 refugees. It is certainly possible that some percentage of these housing plots are either currently inhabited, or likely soon to be inhabited by relocated Rohingya refugees.
Either way, this analysis suggests that nighttime lighting growth will serve as a useful signal of the rate with which the Government of Bangladesh continues to relocate Rohingya refugees onto the island throughout 2022. Given the need to provide humanitarian support to these populations, human rights concerns over their ability to transit back and forth between Bhasan Char and mainland Bangladesh, and the potential for catastrophic weather events over Bhasan Char to require a complex evacuation or disaster response from the international community, close tracking of the population on the island may prove essential in the coming months and years.
Key findings and implications from this report include:
Nighttime lighting serves as a useful proxy measure for the rate of relocation of Rohingya refugees from Cox's Bazaar to Bhasan Char in Bangladesh.
This data offers evidence consistent with official public estimates that roughly 19,000 Rohingya refugees in total have been relocated to Bhasan Char in 2021, based upon the housing capacity of specific plots on the island experiencing major nighttime lighting growth in 2021.
Minority populations within the Rohingya may have been the first groups resettled onto Bhasan Char, as supported by evidence of major nighttime lighting growth in a semi-isolated set of housing plots on the island - perhaps due to the risk these populations previously faced in crowded camps in Cox's Bazaar.
Nighttime lighting over existing refugee camps in Cox's Bazaar reveals continued major growth in electrification through early 2021, suggesting that the Government of Bangladesh and the international community have continued to prioritize support to these camps despite the relocation of some refugees to Bhasan Char.
1. Monthly VIIRS composites from the Colorado School of Mines have been processed through September 2021, but their curated repository of VIIRS data in Google Earth Engine, which is used to produce this analysis, cuts off in May 2021. Hence, this was the most recent data available to conduct this analysis post-September 2021.
This paper presents preliminary results of RAND research. It has not been formally reviewed or edited within RAND but has gone through the Tearline review process.
Oct. 1, 2021
GoB and UN Announce Agreement for Humanitarian Support to Bhasan Char
The government of Bangladesh and the United Nations announce a new agreement to work together to provide support to Rohingya on Bhasan Char, with plans to move an additional 81,000 refugees onto the island in 2022
Nighttime Lighting Growth in 2021 on Bhasan Char, by Housing Plot
Major growth is defined as a plot experiencing 65 percent or more growth in monthly nighttime lighting at its peak in 2021 compared to its average monthly nighttime lighting in 2019. Moderate growth is defined as experiencing between 45 and 65 percent. Modest growth is defined as experiencing between 25 and 45 percent. Marginal growth is defined as experiencing less than 25 percent growth.
RAND Nighttime Lighting Analysis, NOAA VIIRS, and Google Earth Pro, 2018.
This report used nighttime lighting data to track the relocation of Rohingya refugees from Cox's Bazaar to Bhasan Char in Bangladesh. It revealed unique patterns of nighttime lighting growth on Bhasan Char across 120 separate housing plots, independently affirming public estimates of the current population of 19,000 refugees on the island. Future analysis of nighttime lighting over the island will likely prove beneficial to those involved in humanitarian planning to support refugees relocated to the island, as well as those involved in promoting disaster resilience and preparing for potential disaster response in support of Rohingya refugees in a monsoon-prone region.
Things To Watch
Will nighttime lighting growth continue to serve as a close indicator of the relocation of refugees onto Bhasan Char in 2022?
Will nighttime lighting continue to expand at a rapid pace in Cox's Bazaar, despite the relocation of refugees out of specific camps onto Bhasan Char?
Will progress be made to ensure freedom of mobility for Rohingya refugees between Bhasan Char and the mainland?