Claim: China operated black jail in Dubai — Open Source Analysis

In August 2021 the Associated Press (AP) reported a claim that a Chinese woman was held at a “Chinese-run secret detention facility in Dubai”. The woman, Wu Huan, 26, alleged that she was staying at a hotel in Dubai when she was apprehended by Chinese officials and taken to a converted villa that also held two Uyghur prisoners.

This article is an ongoing project and will be regularly revised/updated as the investigation continues. Reach out to me on Twitter (@tomtomjarvis) if you want to join the project Discord.

The AP were unable to verify the claims which resulted in the story receiving brief coverage in other outlets but little more. Being able to verify/debunk whether China is running a “black jail” outside of its borders could have important implications in understanding the extent of Chinese exertion of power internationally.

It would be of particular value due to Dubai’s importance as a transit hub for travellers across Asia, with many travellers passing through. Some of these travellers may be of interest to (or wanted by) the Chinese state.

Dubai. (Source: Wikimedia Commons)

This article follows my personal attempts to use open source data, predominantly satellite data, to analyse the possible options for where this facility may be located.

I have as much interest in countering misinformation, as I do with finding the location. The ultimate goal is to either prove or disprove the claim. A secondary, and presumably more likely alternative, is to refine the analysis enough that people who can access additional sources may find it easier to achieve what I set out to do. Please feel free to collaborate.

PHASE 1 — PLANNING

What are the “knowns”?

This section outlines the “knowns” based on the claims made by Wu in the AP article. These will lay the foundation of the investigation, however, they are reliant on the honesty of the source.

While AP was unable to verify the claims of the “black site”, they were able to corroborate “evidence including stamps in her passport, a phone recording of a Chinese official asking her questions and text messages that she sent from jail to a pastor helping the couple”.

• May 27 — Wu questioned by Chinese authorities at her hotel (Element al-Jaddaf)[25.21973629, 55.33254908]
• She was then taken to Bur Dubai police station [25.2316734, 55.28972216] by the Dubai police, and held for three days.
• On day 3 of her detention at the police station, a man introduced as Li Xuhang from the Chinese consulate visited and questioned her.
• Li took her out of the police station with another Chinese man, handcuffed her and put her in a black Toyota.
• They drove past International City.
• They drove for around half an hour.
• They stopped on a deserted street with “rows of identical compounds” where she was escorted into “a white-coloured villa with three stories, where a series of rooms had been converted into individual cells”.
• She remained for eight days.

What may we presume?

• The site is likely to be secure with a high wall, gate, and potentially CCTV cameras.
• Operational needs will likely require multiple staff, therefore multiple cars may be parked.
• A quiet area is more probable than a busy area.
• Therefore, side streets are more probable than main streets.

Locations of “accidental” evidence

There are several locations where accidental evidence and traces may have been left. Most notably, areas of transition between locations at known time points. This could include tourist photographs taken near the key locations which may contain clues, as well as social media comments, and other online information.

The key places of investigation are:

• The hotel, particularly at the time when Wu was escorted out.
• The police station, particularly when Wu arrives and departs.
• The (unidentified) suspect location of the black site.

Establishing an Area of Interest

This amount of information is not much to go by, but it is enough to define a reasonable area of interest (AI). From the presumption that the car drove to destination from Bur Dubai Police Station via International City, we can pick the most probable routes of the first part of the journey.

Two main routes (red and purple lines) to drive from the Bur Dubai police station (top left, blue pin) to International City (bottom right, green pin).

Establishing this allows us to determine the direction of the car journey, as well as begin to understand an area of interest.

In this case, we can determine the likelihood that the car was travelling between NW and SW of International City, otherwise, a different route would have been taken.

Based on the route known, we can predict the area of interest (blue) to search for. Based on the (rough) half-hour drive time, the radius is roughly 50km. The actual AI shape is larger than what would be expected for the most direct route to account for potential diversions and illusive route-setting.

Refining the search

This AI is a fairly large area to search with thousands of buildings — most of which are irrelevant.

In this instance, the first approach is to rule out large areas that are not needing to be searched. By “blacking out” areas of desert, open land, and farmland, the searchable area decreases significantly.

Alongside that process, any locations that contained possible matches were highlighted green and pinned. At this stage a wide tolerance of the criteria was enforced, meaning some locations may be ruled out down the line.

As seen in the previous image, the AI was too big to scrutinise effectively. In order to resolve this, ruling out large areas that were not relevant — e.g. desert and open land — meant that focus could be directed to the important parts.

The next stage is zoning. Establishing the priority zone is a case of figuring the possible locations which would be on a direct route from the police station, via International City.

Zone 1 (white) shows areas that are approached by a direct route past International City. This makes an assumption that the car drove directly to the villa. All other locations outside this zone are still possible, but require inefficient route-setting on the driver’s end.

With a priority zone started, naming of areas to investigate began. The naming followed Z(X)-XXX format, with the “Z” number indicating the zone and the latter numbers indicating the unique area. The first location labelled was Z1–001. There have been nine locations of possible matches identified in Zone 1.