Catherine DillThe Jinkouhe Gaseous Diffusion Plant is hot!

Much to my dismay, I find that I’m still catching up on reading from the summer.  In August, Hui Zhang published a new report on China’s uranium enrichment capacity, and he makes a wonderful identification of a new civilian centrifuge facility associated with Plant 814 in Sichuan province. While on the subject of Plant 814, this report’s publication gives me the perfect opportunity to say a few words about the somewhat mysterious gaseous diffusion facility also associated with Plant 814.

The Plant 814 gaseous diffusion facility—commonly referred to as the Heping GDP because of it its proximity to the Heping Yizu (和平彝族), a third-level administrative township of the Yi minority—is also known as the Jinkouhe GDP as it is located near Jinkouhe (金口河), in the Leshan City prefecture of Sichuan province. It is the second of China’s gaseous diffusion plants (the first is in Lanzhou).

Source: Google Earth

The Plant’s history remains somewhat opaque, but the facility was built as part of the third line industries in the mid-1960s and likely began producing enriched uranium in 1970. It has been difficult to determine the current status of Jinkouhe. China reportedly stopped producing HEU for weapons in 1987, but China does not openly discuss the Jinkouhe GDP. Some analysts have assumed the plant is no longer operational, but as Zhang notes, local press accounts suggest the plant is still operational. China may still enrich uranium for military purposes, including fuel for naval reactors. Indeed, the district remains closed to foreigners, lending credence to the idea that the facility still has a sensitive military purpose:



I’m inclined to agree that the plant remains operational, both for the reasons Zhang cites in his most recent publication and because of something I can see in satellite imagery.

My colleagues at CNS and I have been exploring the idea of using Landsat 8 satellite imagery from the US Geological Survey to detect changes at nuclear facilities. Landsat 8 images, in addition to being free, contain thermal infrared data from thermal infrared sensors (TIRS) on board the satellite. As NASA explains, the TIRS sensors can detect land temperature by using

Quantum Well Infrared Photodetectors (QWIPs) to detect long wavelengths of light emitted by the Earth whose intensity depends on surface temperature. These wavelengths, called thermal infrared, are well beyond the range of human vision.

Most nuclear facilities are not “hot” enough for observable temperature differences to appear, but gaseous diffusion plants are more observable as they are very hot. Unfortunately, the resolution of the thermal infrared images is quite low, only 100 meters. As it happens, the Jinkouhe GDP is enormous, large enough that it is possible to observe a difference in temperature between the gaseous diffusion hall buildings and the surrounding environment at 100m resolution.

At first blush, Landsat 8 imagery appears to confirm that plant remains operational:

For scale, the Jinkouhe GDP is located in the red rectangle in the top-left quadrant of the image.  Source: US Geological Survey
The image zoomed to the red rectangle which encloses the Jinkouhe GDP. The two long white rectangles are the gaseous diffusion halls, which are warmer than the surrounding environment.
The image with adjusted contrast for enhanced visibility.

Some Chinese accounts suggest that China has converted all uranium enrichment to centrifuge technology, but there is little evidence to confirm that the Jinkouhe facility has been converted, nor are signs of conversion like construction or increased vehicle activity visible on available satellite imagery. Moreover, centrifuge facilities use less energy and are far “cooler” than a gaseous diffusion plant.  One would not expect to see a centrifuge plant on a Landsat 8 thermal image.

How hot is the Jinkouhe GDP? This is a question I’m still trying to figure out how to answer. Just this year we were able at CNS to obtain licenses for ArcGIS and ERDAS Imagine, and we are now teaching ourselves how best to process and analyze Landsat 8 imagery. What I can observe at the moment is that the Jinkouhe GDP is hot.


Courtesy of Jeffrey, here is a short clip (from here) to help our readers visualize the “hot” part:


  1. AEL (History)

    What time of the year, or time of the day was this taken?

    Given that all the big buildings look very bright in this picture (all other bright spots seem to look something like warehouses) it may be poor insulation in the winter, or it may be a hot roof in the sun

    • catherine (History)

      A good point. This image was captured on July 6, 2015 (its ID number is LC81300402015187LGN00 if you’re interested). I’ve also considered the season, and after looking at many catalog images, the plant appears warmer than the surrounding environment throughout the year. But I haven’t yet done a proper brightness comparison between the plant and large warehouse buildings nearby.

  2. YankeeCynic (History)

    It’s worth considering that this plant could be kept running due more to internal bureaucratic realities than out of real need. Paducah Gaseous Diffusion Plant immediately jumps to mind.

    • catherine (History)

      Yes, certainly worth considering. Given the plant’s age, it is not at all out of the realm of possibility that it is slated for conversion in the near future. And, as Nick points out below, there are what are possibly new administrative and support facilities at the site, but no changes yet that I observe to the two gaseous diffusion halls.

  3. nick (History)

    expansion of administrative&support facilities during last five years is clearly visible

  4. Sparks (History)

    I agree that the plant is operational, but only partially, probably around 50%. On the right side of the site is a small vertically oriented rectangular structure that looks very much like a recirculating chiller (a type of cooling tower). I have a moderate size unit of this type at my plant and both its visual and thermal images are consistent with what is shown at JGDP. The reason I say partial operation is that only the lower half (reference to the image) of the structure is hot. Its temperature is close to the temperature of the lower main building. The top is somewhat above ambient, but mot much. This type of chiller has sections and only half of them are operating. To me this is convincing. If I knew the ambient temperature that this picture was taken at, I could give a reasonable estimate of the delta-t of the chiller relative to the lower main building. Unfortunately, the thermal image covers a larger area than the visual image. There is a small hot spot towards the bottom, generally centered in a meander of the river. This may be an water inlet. This would be consistent with the recirculating chiller. There is also a string of small hot spots to the upper right of the image, same side of the river, and along its bank in another meander. The size and distribution of these hot spots suggests support buildings, residences, labs, offices, etc. These details are not in the cropped visual image.

    • catherine (History)

      Thanks for sharing your useful observations. Sorry about the cropped image–file size restrictions, etc. For your reference, the site is located at 29°14’7.97″N 103° 3’45.40″E. I must admit I don’t know much about recirculating chillers, and I hadn’t yet considered looking at the potential warm spot on the river. I’ll have to look into recirculating chillers more. There also appear to be two coal plants located on the river, which may account for a few of the warm spots.

  5. Allen Thomson (History)

    It might be worthwhile to see if there is visible light imagery taken shortly after a light snowfall. Snow (and frost and dew and, on water, ice) can be a useful indicator of temperature and temperature differences.

    • catherine (History)

      Thanks for the suggestion, Allen.

  6. Andreas Persbo (History)

    Shameless self-promotion. One of our former interns, Jasper Pandza, wrote about this for Trust & Verify back in 2009. It’s available here It’s still one of my favourite articles.

    • catherine (History)

      Thanks for sharing the link, Andreas! It’s interesting to observe the improvements in the last six years in Landsat imaging with the launch of Landsat 8, as well as some of the analytic possibilities with GIS software (which admittedly I’m just now exploring).