Research Dossier:
Mapping the Environmental Impacts of the Siachen Glacier Military Occupation
Hamza Jilani
March 2024
Cartographic Workflows + Media
As mentioned in my research statement, my vision for a project that answers my research question would be a TimeMap-like app using a stable Siachen Glacier outline developed from the Randolph Glacier Inventory (RGI) as the base map. Clicking on each event in a timeline below the map allows the user to zoom into an associated location pin on or around the glacier to view more information about the event, including relevant political context as well as a timelapse that shows how the military-associated event has changed or interacted with the environment around the particular site of military operation or infrastructural project. The timeline below the map would include a slider that turns the base satellite imagery layer into an interactive timelapse showing a broader view of glacial change in the region over time using false color visualizations of glacial area/extent and velocity as well as true color and highlight-optimized true color.
Above is a mockup of the user experience that first shows the base map with the timeline, then illustrates how a user might interact with the timelapse function, and finally shows three examples of ‘event’ cards associated with specific times and locations.
I developed the base map on Earth Engine using Randolph Glacier Inventory. My code filters the glacier feature collection to an area of interest as marked by a polygon import, makes a raster image out of the land area attribute, and creates a binary mask to generate a static polygon that represents the extent of Siachen Glacier. I wanted to use RGI to draw this base map because it is a static snapshot within a critical database that measures glaciers at a specific point in time (2000), and I wanted my base map to incorporate a static, consistent measure. Moreover, the RGI polygon’s inclusion in my base map would contribute to my project if it is considered to be a historical document of sensing capabilities and technologies as functions of political and material realities. I also used Earth Engine to plot points of interest on the map by coordinates with a great deal of accuracy, though, as mentioned in my Data/Evidence section, I used Google Earth to search for these locations due to the enhanced detail in the high resolution optimized true color imaging available.
I developed the larger timelapse that visualizes glacial changes using the timelapse layer in Google Earth in order to provide broader context for changes in the glacier over time as the military occupation has occurred alongside the hyper-local context I provide in the event cards. In doing so, I provide a holistic view of the occupation that privileges both broad and local knowledge.
However, my timelapse feature would ideally include information regarding glacial extent and velocity as well as true color imagery to get a more detailed picture of these broader changes. I would use Cartostat to perform digital elevation modeling of the glacier – radiometric and spatial resolutions as well as the time difference between acquisitions of the members of stereo image pairs make this dataset especially useful for modeling glacial changes – and use this data together with Landsat Data with <30% cloud cover in minimal snow months, as the “Area and mass changes of Siachen Glacier” study did to create visualizations for use in timelapses that can show changes in glacial area and velocity. This would illustrate how the glacier has melted over the course of the occupation and affected the environment around it. However, my current representations of glacial extent are limited to what is available to me via Earth Engine and GLIMS; thus, I developed code that overlays GLIMS’ more detailed glacier polygons onto a base map to model glacial extent. This allows for more dynamic representations than RGI, since Earth Engine has 6 different years of GLIMS data.
Moreover, using Sentinel-2 L2A data through the Sentinel Hub EO Browser, I have created timelapses that use satellite bands 4, 3, and 2 to illustrate changes in the glacier over time through true color and highlight optimized natural color visualizations. I hope that, in allowing users to have a variety of ways to view change in the glacier, I will be providing a map that is holistic and contextualized, despite my sources’ contributions to the ‘gigantification’ of data.
I also used timelapses that use satellite bands 4, 3, and 2 of Sentinel-2 L2A for true color and highlight optimized natural color visualizations to illustrate changes in the environment and military-environment interactions at a more local level as they are associated with the events that I mapped out for this project. Here, I ran into some difficulty, as this imagery sometimes did not show me very meaningful change due to political and material challenges with remote sensing at strategic and extreme-weather sites on or around Siachen glacier. I had to manipulate the maximum cloud coverage and minimum tile coverage to approximately 30% and 60% based on my precedent investigations and the visual quality of the images in the timelapse to generate visualizations that could be useful. However, even then, the imagery requires acute attention to particular sections in order to get a picture of changes in the environment or changes in interactions with the environment over time. Some of my timelapses barely show a clear pattern of change, but I decided to keep them in my project, because, ultimately, the representations that this project generates are evidence of the struggle of ‘seeing’ this information. By documenting the struggle of 'seeing' this information, my project becomes not just a tool for analysis but a historical record of the constraints and capabilities of remote sensing technologies within current political and material realities. In addition to using true color imagery, in order to highlight fluctuations in water conditions and the associated extreme conditions facing Indian military operations and infrastructural projects, I created a timeline of Normalized Difference Water Index (NDWI) visualizations to illustrate river fluctuations due to glacial melting. Below are examples of my NDWI, true color, and highlight optimized true color imagery timelapses.
The ‘event’ cards that hold this timelapse information as well as relevant political and historical context about each event are organized in an instance of Timemap, a software developed by Forensic Architecture to dynamically visualize time and space events from a spreadsheet database in a browser by proxying source data in a spreadsheet and making it available as JSON endpoints. In order to build my project out as a Timemap, I would have to clone and install timemap, configure it using the provided examples, and run it locally, then modify the data sheet to include my events of military-environment interactions at Siachen Glacier. My own Timemap would highlight and distinguish between military operations, military infrastructural projects, and civilian/tourist-related events such as the opening of the Siachen Base Camp to Indian tourists, or the construction of tourist infrastructures or religious sites. Mapping these events within a timeline and a map would provide a clear visual representation of the imperial linkages between military, infrastructural, and civilian-oriented projects across space (on the map) and time (on the timeline) at Siachen. Below is a mockup of what the Timemap would look like.
I ran into some difficulty while looking for information regarding particular events to map onto my Timemap. Although military operations and infrastructures are sometimes made hypervisible through press releases, there is limited publicly available information regarding more confidential strategic operations; moreover, remote sensing capabilities provide a limited view of the changes associated with these events. However, my mapping project attempts to laminate the concurrent hypervisibility and opacity of military occupation at the glacier, self-reflexively recognizing the multifaceted nature of visibility and illustrating the reality of this visibility while attempting to generate visibility by creating a clear visualization of military presence on the base.
In order to extend my project’s ability to do this through the incorporation of even more events, I want to use machine learning in a similar manner as the Battle of Ilovaisk project to datamine YouTube and Twitter for pictures and videos of evidence of military operations, infrastructures, and civilian/tourist areas. Given ML’s image-searching capabilities, I would train the ML model to look for things like military equipment, infrastructural equipment, and tourist signage to gather more detailed information about events at and around Siachen: these are examples of evidence of the kinds of events that I want to learn more about and pinpoint locations and timestamps for, and the images and metadata gathered from searches on these social media sites would provide a great deal of this context. This would allow me to base my project on data made hypervisible by political and material forces to more effectively uncover a landscape of information made opaque by these very same forces while still illustrating the reality of this complex visibility.