Ports, pipelines, straits in the Middle East, and the maritime transport routes connecting to Japan. Understanding the relative positions of each is useful for grasping news about crude oil transport.
We have released the map generation code for the “Middle East Crude Oil Transport Routes” and “Crude Oil Transport Routes to Japan” that we have been creating and updating on GitHub. The mapping is done by the Hidenori Watanabe Laboratory at the University of Tokyo, and Codex is used for creating and modifying the code.
The repository contains the drawing scripts, geographic data, background maps, sources, and update procedures. In addition to using the completed images, you can also add information you have researched yourself and update the maps.
In the Middle East map, we have shown the connection between crude oil pipelines and maritime transport. We have also included details such as loading via SBM off the coast of Sidi Kerir, and loading/transshipment off the coast of Oman, so that the differences in methods can be understood.
In the global map, we have shown the main routes connecting the Middle East and North America to Japan, and noted the costs and risks associated with procurement for Japan. Ship-to-ship transshipment off the coast of Malacca is also included. This is a diagram for reading long-distance detours, transshipments, and ship type or canal conditions along with their relative positions.
Maps are generated from coordinates and code.
We use OpenStreetMap-based geographic data for the background and render it in a dark theme. On top of that, we use Python to overlay routes, representative points, place names, and notes based on public documents to generate an SVG. We then render it in a browser and convert it to a PNG for posting and distribution.
Generative AI assists in organizing materials and creating or modifying code. Since the placement of routes and text remains as code and data, modifications such as “move this place name slightly” or “shorten the note” can be reflected in the next drawing. Once the initial environment setup is complete, redrawing using saved backgrounds can be done offline.
The repository can be passed to an AI and edited via chat.
If the generative AI you are using supports it, you can start by opening the repository folder as the work target, passing the GitHub URL, or attaching a ZIP file.
For example, you can start by making the following request.
このリポジトリを読み込み、対話しながら地図を編集できるようにしてください。
README.md、AGENTS.md、docs/AI_WORKFLOW.mdを読み、
地図の構成と情報基準日を確認してください。
実際に読めたファイルと、この環境で編集・実行・画像確認ができるかを教えてください。Once you have confirmed the structure, you can communicate specific changes.
全球図に、この資料で確認できる積替え地点を追加してください。
出典と観測日を残し、既存の地名に重ならない位置に注記を置いて、完成画像を見せてください。Continuing while looking at the image, you can make adjustments such as “summarize the text into two lines,” or “don’t move the point, just move the label to the right.”
In the README, we have organized the order in which the AI reads, the files to be edited, and how to execute them. We have also prepared request templates that can be used for adding, updating, or deleting proprietary data, as well as procedures for handing over work to a different conversation.
However, even if the AI can read the files, it does not necessarily mean it can edit, execute, and confirm the images. Processes that are impossible in your usage environment must be executed manually.
Ports, routes, and notes added independently are saved in GeoJSON format, which records geographic information. Since sources, observation dates, and confirmation status can also be recorded, you can trace the basis for updates later. The update date of the additional information and the information reference date of the original map are handled separately.
Usage: Check the creation date of the map and the validity of the data.
The situation and cost assessment in the published maps are based onSeptember 29, 2026, andsome routes and loading methods were added on October 12. Even if you regenerate the map, the information does not automatically become the latest.
The maritime routes are schematic lines. They do not represent real-time AIS tracks, daily operations or facility status, or safe passage. The bidirectional arrows indicate the round trips of ships; they do not confirm that crude oil is transported in both directions or that individual ships have completed their round trips.
The notes on costs also organize factors such as fuel, charter days, cargo handling, and waiting time. They cannot be read as a comparison of freight rates under the same conditions or as a ranking of procurement costs. The scope is limited to crude oil and does not include transport networks for LNG, LPG, or petroleum products.
While the code includes mechanisms to check input formats and text overlap, the accuracy of the content must be verified by comparing it with source materials and the completed map.
The independently created code, explanatory text, and map components are provided under CC BY 4.0. You may reuse them by displaying the source, creator, and license, and by clearly stating if any changes were made. Background maps and externally sourced data are subject to their respective attribution requirements and terms of use.
I hope you will use this for classes, research, or news commentary while adding your own collected materials. Instructions on how to use it and the sources can be found in the GitHub README.
