Investigating the Allegations of Indian Nuclear Test Preparations in the Rajasthan Desert

A CTB Verification Exercise Using Commercial Satellite Imagery

July 1996

Vipin Gupta and Frank Pabian

From the Original Document at: http://www.ca.sandia.gov/casite/gupta/analy.htm




Paper Sections Paper Indices

Table of
Contents		 Appendices Last
Page		 Next
Page


Image Analysis


Acquisition, Processing, and Analysis of Commercial Satellite Imagery


If the allegations against India were a CTB compliance dispute, analysts would seek to resolve the issue by completing three principal objectives:

  1. Collect new information from reliable, independent sources.
  2. Separate fact from fiction in India's official responses, India's nuclear testing history, and the media reports.
  3. Determine whether the body of evidence implicated or exonerated India.

This verification exercise tested the utility of commercial satellite imagery for completing these three objectives. Imaged portions of the Rajasthan Desert were obtained from the archives and recently acquired by commercial imaging satellites. Table 2 lists all of the satellite images that were used and explains why each specific image was selected for detailed analysis. The images were processed, enhanced, and interpreted in an attempt to fulfill three successive tasks.


Table 2: List of satellite images that were obtained, processed, and interpreted for the CTB verification exercise.

       
Satellite and Sensor Ground Sample Distance* Scene Center (Path-Row) Acquisition Date and Time (GMT) Relevance to Study
CORONA, KH-3 ~10 m 27.17° N, 72.98° E December 12, 1961 Oldest available image of Pokharan area acquired 13 years before India demonstrated a nuclear testing capability
Cosmos, KVR-1000 3-4 m 27.17° N, 71.80° E May 24, 1992 03:26 High spatial resolution image of 1974 nuclear test site and surrounding area
SPOT-3, HRV-1 Panchromatic mode 10 m 26.98° N, 71.69° E 196-296 March 25, 1995 05:57:34 Most current, cloud-free image of Pokharan area acquired before the allegations of Indian nuclear test preparations were made
SPOT-3, HRV-1 Panchromatic mode 10 m 27.06° N, 71.76° E 197-295 March 2, 1996 06:02:29 Oldest, cloud-free image of Pokharan area acquired after the allegations of Indian nuclear test preparations were made
Radarsat, Synthetic Aperture Radar (SAR) 6.25 m 27.03° N, 71.80° E March 6, 1996 13:22:35 Oldest, radar image of Pokharan area acquired after the allegations of Indian nuclear test preparations were made
SPOT-3, HRV-1 Panchromatic mode 10 m 27.11° N, 71.76° E 196-296 March 18, 1996 05:54:34 Stereo mate of March 2, 1996 SPOT image
* Ground sample distance (GSD) is the length and width of each pixel in a digital image projected onto the ground. Images captured on film are described at an equivalent GSD. The size of the GSD is a key factor in the amount of spatial detail in an image.


The first task was to find the subsidence crater formed by the May 18, 1974 test. This task was undertaken not only to resolve the conflicting reports on the test location, but also to pinpoint the locale where subsequent nuclear testing activity allegedly took place. All of the Indian media reports on nuclear testing preparations in 1981, 1982, and 1995 described the location of the activity relative to the May 18, 1974 ground zero. [38] The crater formed by this test was found in the satellite imagery, which made it possible to attempt the second task - search for the existence of a nearby site reserved for future Indian nuclear tests.

All of the media reports that described Indian nuclear test preparations claimed that the activity had taken place at a site reserved for nuclear testing. Even the two articles in the Times of India and Jansatta that strongly denied Indian nuclear testing activity stated that such a site existed. The satellite images were analyzed in an attempt to either confirm or refute these claims of an Indian nuclear test site. This was done by interpreting the imaged features within the vicinity of the May 18, 1974 crater and comparing the image-derived information with the media reports. Digital change detection was also performed, which revealed new activity in the area. The third task was to determine whether the new activity could be attributed to recent conventional military exercises, nuclear test preparations, Prithvi field testing, or something completely different.



Locating the May 18, 1974 Subsidence Crater

Two archived satellite images were selected for use in the search for the May 18, 1974 subsidence crater. One image was acquired on March 25, 1995 by the HRV-1 panchromatic sensor on board the French SPOT-3 satellite. The other image was acquired on May 24, 1992 by the KVR-1000 camera flown on one of the Russian Cosmos satellites. The SPOT-3 image covered a 60x60 km area at a ten-meter ground sample distance. The KVR-1000 image covered a 40x40 km area at approximately 4-meter ground sample distance. [39]

SPOT panchromatic imagery was chosen for the crater search because of its relatively large areal coverage at moderately fine spatial resolution. The March 25, 1995 image (Path 196, Row 296) was selected because it was the most current, cloud-free image in the SPOT archives that captured virtually all of the error circle associated with the ISC seismic location estimate of India's 1974 nuclear test (see figure 2). [40] The image also covered Pokharan town as well as the villages of Loharki, Khetolai, and Malka - all of the places that were reportedly near the May 18, 1974 explosion site. [41] In addition, the image contained the feature found in the October 26, 1988 IRS-1A image that was previously identified by one researcher as the crater formed by the nuclear test (see figure 2).

[Figure 2]

Figure 2: Geographic coverage of the archived March 25, 1995 SPOT image and May 24, 1992 KVR-1000 image. Both images cover large portions of the error circle associated with the ISC seismic location estimate of the May 18, 1974 test.

KVR-1000 imagery was chosen for the crater search because it was the highest resolution satellite imagery that could be bought on the open market. The May 24, 1992 image was selected because it was the only cloud-free image made available for purchase that captured a large portion of the error circle associated with the ISC seismic location estimate (see figure 2). [42] Like the SPOT image, the KVR-1000 scene covered the villages of Loharki, Khetolai, and Malka. It also contained five of the eight villages around Loharki that were reportedly planned for evacuation in 1981 (see figure 2).

The full SPOT image showed flat terrain with agricultural fields bordering desert areas. Twenty-one towns and villages were located and identified in the image. The main rail line and the primary roads were also visible. However, no circular features were found that could be identified as the 18 May 1974 crater. The feature in the IRS-1A subscene was found in the SPOT image, but it still appeared too coarse to establish a conclusive match with the helicopter photo of the crater (see plates 1 and 2).

The KVR-1000 image consisted primarily of flat, desert terrain intermixed with sand dunes. Synoptic examination of this image revealed a large, circular depression located in a remote area about six kilometers south of Loharki. [43] Plate 3 shows the salient circular feature surrounded by the sandy desert terrain. The overhead view of the depression appeared similar to the May 18, 1974 crater shown in the helicopter photo. To determine whether the depression was indeed the subsidence crater, the KVR-1000 image was rendered as a perspective view in order to replicate the oblique viewing angle of the aerial photo. By simulating the camera geometry that was used to acquire the aerial photo, the KVR-1000 perspective view of the depression could be compared directly with the helicopter photo of the crater. However, before the perspective view could be rendered, the bearing of the helicopter-borne camera had to be determined.

 
[Plate 3] Plate 3: May 24, 1992 KVR-1000 image of a large circular depression located in relatively uniform desert terrain. The inset is an enlargement of the circular depression. The two, light-toned lines slanting across the image are scratch marks on the film positive.

The helicopter photo was reportedly taken one hour after the explosion. According to the International Seismological Center (ISC), the explosion occurred at 02:34:55.40 ±.17 GMT so the photo must have been taken at approximately 03:35 hrs GMT on May 18, 1974. [44] At this instant in time, the sun was located east of the estimated test location and shadows pointed towards the west (see Appendix A). In the foreground of the helicopter photo, the bush shadows point toward the camera providing a west compass bearing (see plate 4). Given this directional indicator, the helicopter-borne camera must have been positioned west of the crater and pointing east when the picture was taken.

 
[Plate 4] Plate 4: Annotated helicopter photo of the crater. The four physical features labeled on the photo were found in the KVR-1000 image of the crater (see plate 5). Note how the bushes in the foreground cast shadows toward the camera.

Duplicating this camera orientation, the KVR-1000 image of the depression was rendered as a perspective view. The camera's elevation angle was approximated visually by distorting the circular depression into the oval crater shape shown in the helicopter photo. Fortuitously, the sun angle was also duplicated because the KVR-1000 image was acquired on almost the same date as the helicopter photo (May 24 May 18) at almost the exact same local time (03:26 hrs GMT 03:35 hrs GMT). As a result, the shading in the two images was virtually identical.

Comparison of the KVR-1000 perspective view with the helicopter photo revealed similar topographical features. These features proved to be the key for matching the two images. Four geomorphological structures from the circular depression were found in the helicopter photo, establishing conclusively that it was the subsidence crater formed 18 years earlier. [45] Plates 4 and 5 show the sloping lobe, scarp, fissure, and concentric cracking boundary that proved both images depicted the same crater.

 
[Plate 5] Plate 5: KVR-1000 perspective view of the large circular depression. The simulated aerial camera is located west of the crater and points to the east. The four physical features labeled in the image proved that the depression was the May 18, 1974 subsidence crater. The distance from the prominent scarp through the crater center to the concentric cracking boundary is 165 ± 2 meters.

After the circular depression was positively identified, its coarse outline was located in the March 25, 1995 SPOT image. The SPOT image was digitally merged with the two 1:250,000 scale maps of the area, and the geographic location of the crater was measured: 27.095 ± .001 degrees N, 71.752 ± .001 degrees E. [46] That location placed the crater 1.5 km southwest of Malka village, 5.8 km south-southwest of Loharki village, 9.0 km north-northeast of Khetolai village, 24.8 km northwest of Pokharan town, 23.4 km west-northwest of the IRS-1A feature, and 12.5 km north-northwest (azimuth angle: 339 degrees) of the ISC seismic location estimate for the May 18, 1974 test. [47] Thus, of the seven articles that reported the May 18, 1974 test location, only one - the May 9-15, 1982 Sunday Magazine article - accurately located it.



Searching for an Indian Nuclear Test Site

With the May 18, 1974 crater found, it became possible to determine whether the claims of an existing Indian nuclear test site were true. Since all of the media reports of nuclear test preparations in 1981, 1982, and 1995 locate the alleged activity within the vicinity of India's first nuclear test, the region surrounding the May 18, 1974 ground zero was analyzed to determine what kind of ordinary activity took place in the area and whether there were any abnormal features that could indicate the existence of a nuclear test site. To facilitate the analysis, a declassified KH-3 image was obtained (see plate 6). [48] Acquired on December 12, 1961, the image was the oldest one available of the area around the 1974 nuclear test site. Since it was acquired 13 years before India demonstrated its nuclear testing capability, it proved to be the ideal reference image. Comparison of recent satellite imagery with the historical KH-3 image revealed numerous changes throughout the landscape.

 
[Plate 6] Plate 6: December 12, 1961 KH-3 image covering the 1974 nuclear test site and surrounding region. The small, light gray patches scattered throughout the image are farming plots. The black splotches on the image are cloud shadows. The inset is an enlargement of section I. showing Malka village and the exact area that was later used for the May 18, 1974 nuclear test.

Plate 6 shows a small section of the KH-3 image strip that covered the 1974 nuclear test site and surrounding region. The image was acquired at approximately the same spatial resolution (10 meters) as SPOT imagery. Plate 7 shows a section of the March 25, 1995 SPOT image covering the same area. Both images show identical sand dune distributions, but appreciably different land use patterns. During the intervening 34 years, the Loharki-Bardana road and the Khetolai-Dholia road were paved and an east-west rail line through Lathi was constructed (see plate 7). In addition, Malka village ceased to exist and virtually all of the land between the two paved roads was taken out of farm production (see plates 3, 6 and 7). [49] Analysis of the March 25, 1995 SPOT image and May 24, 1992 KVR-1000 image revealed that much of this region had been taken over by the military. Features that appeared to relate to military activity were found in five distinct areas and subjected to closer examination (see plate 7).

 
[Plate 7] Plate 7: March 25, 1995 SPOT image covering approximately the same area shown in plate 6. Note the absence of polygonal farm plots over most of the land between the Khetolai-Dholia road and Loharki- Bardana road. Areas 1-5 contain features that appeared to relate to military activity. Area 5 contains all of the landmarks connected to alleged nuclear test activity in 1981, 1982, and 1995. Rectangle QRST represents the 8 km x 3 km fenced perimeter located three kilometers south of the 18 May 1974 crater, as described in the May 5, 1981 Indian Express issue. The perimeter was reportedly constructed in early 1981 shortly before Ajasar and Loharki village leaders were informed of a meeting on a four-day evacuation plan. Since the azimuthal orientation of the perimeter was not specified, the arc represents all of the possible perimeter orientations with the fence line at Q fixed three kilometers south of the 1974 crater.

The line connecting the May 18, 1974 crater and Khetolai represents the location of the area that was reportedly cordoned off by the Indian army and prepared for a nuclear test, as described in the May 9-15, 1982 issue of Sunday Magazine. The point on the May 18, 1974 crater represents the area where barbed wire fencing was reportedly erected in 1995, as described in the December 30, 1995 Hindu article, the January 1, 1996 Business Standard article, and the January 3, 1996 Hindustan Times article.

 
[Plate 8] Plate 8: March 25, 1995 SPOT panchromatic image showing area 1 (see plate 7). Although the spatial resolution of the image is relatively low, it was possible to determine the functionality of several structures in the image by identifying distinct features and linking them together through deductive reasoning.

The image shows the coarse outline of a secured, military road-rail transfer station near a civilian rail station. Just north of the Khetolai-Dholia road and rail line, two rows of black dots off a single side road can be seen. These structures were inferred to be part of a military base.

At area 1, a feature shaped as a right triangle was located near Lathi (see plate 8). In contrast with the nearby agrarian lands, the triangular feature appeared dark in tone due the growth of natural, healthy vegetation in its interior. The presence of such vegetation was noteworthy because media reports described extensive livestock grazing in the area. [50] This feature, however, showed no signs of such agrarian activity which indicated that it was secured by perimeter fencing that kept wandering animals (and farmers) out. The sharp tonal contrast along the triangle's boundaries can be attributed to livestock grazing outside and up to the perimeter fence.

Inside the fenced area, the coarse outline of a facility can be seen. It is located next to the main road and rail line that lie along the hypotenuse of the right triangular perimeter (see plate 7). It is also located near a rail station south of Lathi village, which is indicated by the coarse outline of side tracks and buildings along the rail line connected to Lathi by a light-toned, well-trodden dirt road (see plate 8). Given its link to two primary transportation routes, the secured facility was inferred to be a military road-rail transfer point. The military role of the facility was deduced not only by the presence of the perimeter fencing, but also by the presence of the rail station south of Lathi. The existence of an unsecured and secured rail facility next to each other strongly suggested that the former was used primarily for civilian traffic and the latter was used principally for military transport purposes.

Just north of the military transport point, there is a site consisting of two rows of black dots with barren ground in between (see plate 8). These black dots can be inferred to be large buildings similar in coarse appearance to the structures shown in Lathi. However, in contrast with the structures in the village, the buildings in the remote site are widely spaced and distributed in an orderly fashion. In addition, the site can only be reached by taking a single, paved side road that ends at two large structures. Based on these site characteristics and its close proximity to the military transport point, the feature was inferred to be a military base.

At area 2, a multistory, enclosed facility covering approximately 195,000 square meters was found near Sri Bhadria (see plate 9). It is located 0.5 km southeast of the village and 1.5 km west of the largest farming area between the Khetolai-Dholia road and Loharki-Bardana road (see plates 7 and 9). A single access road links the Sri Bhadria installation to the main road. To reach the facility, one has to travel on the Khetolai-Dholia road, turn onto the access road, and drive 8.3 km north through a fenced reservation. [51] Sri Bhadria can be reached by taking the single access road and driving through the facility (see plate 9).

 
[Plate 9] Plate 9: March 25, 1995 SPOT panchromatic image of area 2 (see plate 7). The image shows an unidentified, enclosed facility near Sri Bhadria. The inset of Section I. is the May 24, 1992 KVR-1000 image of the multistory facility showing several animal pens near the northwest perimeter.

The layout and architecture of the buildings provided little information on the functional purpose of the facility. Twelve animal pens located in the northwest corner of the perimeter were the only identifiable features (see plate 9 inset). These pens - along with the facility's close proximity to a large, isolated farming area - suggested that the facility may possibly be involved in agricultural activity. However, this did not explain definitively what kind of activities took place at the facility. As a result, the complex was left unidentified.

At area 3, two large, fanned sets of tracks were found just north of the Khetolai-Dholia road (see plate 10). The scale, tone, and texture of these converging tracks is identical in overhead appearance to published SPOT images of known tank firing ranges in former East Germany. [52] The appearance also matches a veteran photointerpreter's description of tank firing ranges worldwide. [53] Based on these references, the converging tracks shown in plate 10 were identified as tank firing ranges. [54] The firing ranges are connected by several roads to a large complex located just off the Khetolai-Dholia road (see plate 10 inset). In contrast with the village layouts, the complex consists of numerous columns of buildings and a parade ground. Given its configuration and proximity to the firing ranges, it was inferred to be a military base consisting of Indian mechanized units.

 
[Plate 10] Plate 10: March 25, 1995 SPOT panchromatic image of area 3 (see plate 7). The image shows vehicle track patterns that are characteristic of tank firing ranges. The inset is the May 24, 1992 KVR-1000 image of the nearby complex linked by several roads to the firing ranges. This complex was inferred to be a military base consisting of Indian mechanized units.

At area 4, two sandy zones littered with numerous, small pockmarks were located in the May 24, 1992 KVR-1000 image (see plate 11). Both zones are located at the end of a paved road from the tank firing ranges and military base in area 3. One crater zone is located 0.8 km east of Nautala and the other is located 2.25 km northeast of the village. Virtually all of the craters are approximately 15 meters in diameter. [55] The village itself appeared to be in a dilapidated state. In contrast with nearby Loharki village, the rooftops at Nautala did not form enclosed polygons, suggesting that the roof structures were incomplete (see plate 12). Also, no animal pens were present at Nautala. The March 6, 1996 Radarsat image also showed differences between the two villages. When Radarsat imaged Loharki village, it produced a strong radar return. In contrast, Nautala produced a weak radar return, suggesting the absence of standing structures capable of acting as corner reflectors (see plate 12). All of this evidence led to the conclusion that Nautala village had been abandoned. [56] Thus, the two cratering zones near the uninhabited village must be conventional target areas for artillery shelling, air bombardment, or both. [57]

 
[Plate 11] Plate 11: May 24, 1992 KVR-1000 panchromatic image of area 4 (see plate 7). The image shows two distinct zones littered with small craters. Both are located near the end of a paved road from the tank firing ranges and military base in area 3. These zones were identified as conventional target areas for artillery firing, aerial bombing, or both.

 
[Plate 12] Plate 12: Panchromatic and radar images of Nautala and Loharki. The top two panels are sections from the May 24, 1992 KVR-1000 image. The bottom two panels are sections from the March 6, 1996 Radarsat image. The two KVR-1000 panels of each village were contrast enhanced using the same linear stretch algorithm. A different linear stretch algorithm was used for the two Radarsat panels.

At area 5, the terrain was searched for features and activity that could relate to nuclear testing. The boundaries of area 5 surround the geographic region where nuclear test preparations reportedly took place in 1981, 1982, and 1995 (see plate 7). Visual analysis of this area in the March 25, 1995 SPOT image and May 24, 1992 KVR-1000 image revealed a large, triangular-shaped perimeter with unpaved roads, buildings, and berms located inside (see plate 13). The base and height of the isosceles triangle are 5.7 km and 6.1 km respectively. Two sides of the triangular perimeter are clearly delineated by the dark-toned vegetation on the inside and barren, sandy terrain on the outside. Like the perimeter around the military road-rail transport station in area 1, the sharp tonal contrast can be attributed to the livestock that reportedly graze in and around the May 18, 1974 crater. [58] The western side of the triangular perimeter was not easily visible from the differential vegetation, but it was discernible in the high-resolution KVR-1000 image.

 
[Plate 13] Plate 13: March 25, 1995 SPOT image of area 5 (see plate 7). The image shows a large, triangular perimeter with a network of roads inside. The insets are May 24, 1992 KVR-1000 sections showing large buildings and structures inside the perimeter located at areas that have been cleared of vegetation.

As shown in plate 13, the center of the triangular perimeter is located 3.3 km southwest of the May 18, 1974 crater. The location of the secured area matched the May 9-15, 1982 Sunday Magazine report, which described a cordoned area between the May 18, 1974 crater and Khetolai that was allegedly prepared for a nuclear test (see plates 7 and 13). The secured area also overlapped the 8 km x 3 km perimeter described in the May 5, 1981 Indian Express article that was constructed shortly before two village leaders were reportedly informed of an official meeting on a four-day evacuation plan (see plates 7 and 13). [59]

Roads intersect the perimeter at four different points indicating several portals for entry and exit. The portals are located along the north, northeast, east, and south perimeter. The south portal appears to be the principal transportation point to and from the secured area as it is closest to the main road. To reach it from the Khetolai-Dholia road, one has to turn off the main road, drive north through the military base at area 3, and then proceed 4.0 km northeast to the south entrance. [60] Once there, one can drive around the entire site along the inner perimeter road passing by a south perimeter base and an east perimeter base (see plate 13 insets).

Deeper inside the perimeter, there are two discernible areas with unpaved roads leading to sites cleared of vegetation. The May 24, 1992 KVR-1000 image shows the easternmost site located next to a berm (see plate 13 inset). No distinct features can be resolved here. Near the center of the triangular area, there are four large sites enclosed by berms or fencing as well as several small site encircled by berms. Several buildings are located within each site and there is an enclosing berm opposite a single access road (see plate 14).

 
[Plate 14] Plate 14: May 24, 1992 KVR-1000 image of section IV. from plate 13 merged with the 25 March 1995 SPOT image. The merge was done by laying the two images over each other (geometric registration) and adding the two together (digital summation). The cleared areas shown in the March 25, 1995 SPOT image overlap four large sites with buildings that can be reached from a single access road. There is an enclosing berm opposite the single access road, and there are berms or fences surrounding all of the sites. The light-toned, horizontal lines that cross the entire image are scratch marks on the KVR-1000 film positive.

Area 5 has all of the appearances of a military test range. It is located in a remote area with open space inside and outside the fenced perimeter. It contains infrastructure, a road network, several berms, and a limited number of portals into the secured area. In addition, it is linked by road to a military armor base with two conventional testing areas. All of these features make area 5 suitable for a variety of military exercises and weapons tests. And since this area was used for the May 18, 1974 nuclear test and has been the geographic focal point of subsequent allegations of nuclear test preparations, future nuclear testing could be one of the purposes for area 5. With that possibility in mind, area 5 was designated the Khetolai military range and tentatively considered a potential Indian nuclear test site until more recent imagery could be analyzed to establish its purpose in a more definitive way.



Detecting Changes at the Military Areas near Khetolai Village

On March 2, 1996, the French SPOT-3 satellite acquired a cloud-free panchromatic image of the Pokharan area. The image was acquired almost three months after the initial allegations of Indian nuclear test preparations were made. Four days later during its engineering test phase, the Canadian Radarsat acquired a fine beam radar image of the same area, which was obtained as a single look image at a 6.25 meter ground sample distance. On March 18, 1996, the French SPOT-3 satellite acquired a cloud-free stereo mate of the 2 March 1996 panchromatic image. All three images were analyzed for recent activity at the military areas near Khetolai.

The 1996 SPOT images were interpreted first in stereo following contrast and edge enhancement. [61] Besides the stabilized sand dunes scattered throughout the image, no other mounds of sand and earth were found. This suggested that either the Hindu and Hindustan Times reports of a new hill in the area were incorrect, or the hill was removed before the 1996 SPOT images were acquired. It is also possible that the pile of earth was imaged by SPOT-3, but the height differential was too small to be detected in the stereo pair. [62]

Following the stereo analysis, the 1996 SPOT images were registered to the 1995 SPOT image using a second-order polynomial and nearest neighbor resampling. The old and new images were then rendered as a time-lapse, repeating sequence showing landscape changes as small as one 10 m x 10 m pixel. In all three military areas near Khetolai, the time-lapse sequence revealed substantial change between March 25, 1995 and March 2, 1996 and little change between March 2, 1996 and March 18, 1996. [63]

Plate 15 is a section of the March 18, 1996 SPOT image showing the large-scale changes that occurred over the one year period. The most conspicuous change is a large area denuded of vegetation inside the security perimeter at the Khetolai military range (See plate 15 inset). The light-toned area covers approximately 1.6 sq km and is visible in both 1996 SPOT images. Shaped as a plume with two point sources, it extends past a berm ending at an unpaved road. The denuded area exhibits the characteristic effects of a brush fire. [64] At an unknown date and time after the SPOT image acquisition on 25 March 1995, two fires must have started at the point sources, spread downwind over the berm, and stopped at the unpaved road which acted as a fire break. The light-toned appearance of the burnt area can be attributed to the sunlight reflectance off the sandy soil that was previously covered with vegetation.

 
[Plate 15] Plate 15: March 18, 1996 SPOT image showing the tank firing ranges, the Nautala target areas, and the Khetolai military range. The vehicle tracks identified in this image were not present in the March 25, 1995 SPOT image, which means the surface markings were less than one year old. Section I. shows the left and right turn tracks at the north portal of the Khetolai military range. These turns indicate vehicle traffic between the Nautala target area and sites inside the secured perimeter as well as traffic between the May 18, 1974 crater and the inner area. Section II. shows the damage caused by two brush fires that occurred inside the fenced perimeter. The cause and exact date of each fire could not be determined from the imagery.

Besides the fire damage, new vehicle tracks were detected in the tank firing ranges and the Nautala target area. Several new vehicle tracks were observed connecting the Nautala target area to the Khetolai military range (see plate 15). No road traverses the sandy terrain between these two areas so the tread marks must have been produced by four-wheel drive or tracked vehicles. At the northern perimeter of the Khetolai military range, the tracks divide into three sets. One set connects to the north portal, indicating vehicle traffic between the Nautala target area and the secured areas inside the Khetolai military range (see plate 15 inset). Another set indicates direct vehicle traffic between the Nautala target area and the area around the May 18, 1974 crater. The third set of tracks indicate vehicle traffic between the secured areas inside the Khetolai military range and the area around the 1974 subsidence crater.

 
[Plate 16] Plate 16: March 2, 1996 SPOT image of the Khetolai military range. The image shows new vehicle tracks around the 1974 subsidence crater proceeding towards the southeast. The 1974 control point location was mapped onto the image (see Appendix B).

Section I. shows how the area around the May 18, 1974 crater looked on March 2, 1996, and Section II. shows the same area almost one year earlier on March 25, 1995. Comparison of the two insets revealed a new, light-toned ring around the crater and a nearby concentration of vehicle tracks.

Plate 16 shows the series of new vehicle tracks converging at an area next to the 1974 crater and proceeding towards the southeast. It also shows a ring around the subsidence crater that was not present in the March 25, 1995 SPOT image (see plate 16 insets). These changes around the old nuclear test site prompted a reexamination of the published ground and aerial photos that were acquired at the time the first Indian nuclear test was conducted. [65] The historical pictures were analyzed for topographic clues that could help explain why there was renewed activity around the old nuclear test site. The analysis of the archived information did not provide a conclusive answer. However, it did reveal the location of the control point for the May 18, 1974 test (see Appendix B and plate 16).

Inside the fenced perimeter of the Khetolai military range, other new activity was observed. Plates 17 and 18 show how the secured area changed between the March 25, 1995 and March 2, 1996 SPOT-3 image acquisitions. Comparison of the two images revealed two, new concentric sets of curved traces that enclose all sites located in the center of the military range. A new linear trace was also seen intersecting both sets of curved features. The trace lies perpendicular to the prevailing southwest-northeast wind direction, which is indicated by the orientation of the nearby sand dunes (see plate 17). [66]

From point E to point G, it is 2.1 km in length (see plate 18). At one end, it goes through an existing road connecting to point G. At the other end, it branches out in four separate directions passing through at least one established berm or fence and ending at sites A-D in the center of the military range (see plates 18 and 19). Relative to the adjacent land, sites A-D appear slightly larger and brighter in the time-lapse sequence. This ground scarring can be attributed to increased activity at the sites over a one year period. [67]

 
[Plate 17] Plate 17: March 25, 1995 SPOT image of the secured area inside the Khetolai military range. The sand dunes lie along the southwest-northeast axis of the prevailing wind flow. The inset is a contrast enhanced enlargement of sites A-D covering the same area depicted in the plate 18 inset.

 
[Plate 18] Plate 18: March 2, 1996 SPOT image of the secured area inside the Khetolai military range. The image covers the same area shown in plate 17. It shows new linear and curved traces that were constructed over a one year period. Line EG lies perpendicular to the sand dunes, which point in the direction of the prevailing southwest-northeast winds. Points A-G are the endpoints of the new linear features. The inset is a contrast enhanced enlargement of sites A-D pointing out four linear traces that branch off line EG. These traces were not present in the SPOT image acquired one year earlier (see plate 17 inset).

 
[Plate 19] Plate 19: March 18, 1996 SPOT image of sites A-E merged with the May 24, 1992 KVR-1000 image. The merge was done by laying the two images over each other (geometric registration) and adding the two together (digital summation). The linear traces shown in the March 2, 1996 SPOT inset were mapped onto the image. At sites A-C, the traces end at the center of the secured perimeters. At sites A and C, the linear traces intersect the same edge of identical rectangular structures that are visible in the KVR-1000 image. The coarse outline of these two structures are visible in the March 2, 1996 SPOT inset and plate 20.

All of the new features that were found in the 1996 SPOT images were then checked in the March 6, 1996 Radarsat image. The radar image was analyzed to gather information on the surface structure and metallic composition of the new manmade features. In arid regions, these two traits are typically the dominant factors that determine the magnitude of the radar return. [68] If the illuminated objects act as corner reflectors or electrical conductors, the microwave reflectance will be higher than the surrounding background. Conversely, features with flat surface profiles or low conductivity either appear darker than the background or are indistinguishable from the surrounding environment.

Radarsat imaged the three military ranges around Khetolai at a 43- to 46-degree incidence angle relative to nadir (i.e., straight down) pointing towards the east-northeast. The synthetic aperture radar (SAR) operated in the C-Band (5.6 cm). In contrast with the panchromatic visible images, the SAR image appears speckly due to the random interference of the coherent waveforms (see plate 20). [69] Despite its grainy appearance, the SAR image shows a relatively strong radar return from buildings, berms, and fence lines identified in the KVR-1000 image as well as a barely discernible radar return from fence lines delineated in the SPOT images by the differential vegetation. Segments of these identified features positioned perpendicular to the incident microwave beam produced the strongest relative radar return. Similarly oriented segments of the curved traces in the 1996 SPOT images were also seen in the SAR image, including a faint circular arc inside the new light-toned ring around the 1974 subsidence crater (see plate 20). [70] None of the new linear traces or vehicle tracks were detected in the Radarsat image, suggesting the absence of structures or conductive materials.

 
[Plate 20] Plate 20: March 6, 1996 Radarsat image of the secured area inside the Khetolai military range. The image covers the same area shown in plates 17 and 18. Although the SAR image appears speckly, it shows the coarse outline of known buildings, berms, and fence lines. It also shows segments of the curved traces around sites A-D and the 1974 subsidence crater that were oriented perpendicular to the incident radar beam. These curved traces were identified as new perimeter barriers. The inset is an enlargement of the central sites showing strong radar returns from the buildings and structures in sites A-C and weak microwave reflections from the fence lines around sites A and C.

The segments of the curved traces that were observed in the SAR image had radar signatures that were similar to the known berms and fence lines in the area. And since these features completely surround specific sites, it was reasonable to deduce that the curved traces were new perimeter barriers. Between March 25, 1995 and March 2, 1996, these barriers were constructed around the centrally located sites within the Khetolai military range. An additional barrier was built around the 1974 subsidence crater, which must be the new fence described in the Business Standard, Hindu, and Hindustan Times articles.



Interpreting the Activity at the Khetolai Military Range

The image-derived information conclusively established three facts. First, between March 1995 and March 1996, substantial activity did take place at the Khetolai military range in the immediate vicinity of the 1974 nuclear explosion site. Compared with the changes at the nearby tank firing ranges and the conventional target areas at Nautala, the activity at the Khetolai military range left the most and the largest "footprints" on the ground. Second, this activity involved the military. This was evident because the detected changes happened within established military ranges and consisted of perimeter reinforcement, all terrain vehicle traffic, and extensive ground clearing. Third, the observed activity was different in scale and appearance from activities that had taken place in the recent past. The May 24, 1992 KVR-1000 image and the March 25, 1995 SPOT image indicated that no large-scale changes took place at the Khetolai military range over the three year period. Furthermore, neither image shows vehicle track patterns from the 1991 and 1994 winter military exercises that resemble the track patterns in the 1996 SPOT images, which were acquired three months after the 1995 winter exercises.

With these three facts in mind, the image-derived information was integrated into the body of evidence. The collection of claims and facts were sifted and organized into groups in order to determine whether the observed large-scale, unusual activity could be attributed to conventional exercises, nuclear test preparations, Prithvi field testing, or something completely different. This was done in three steps.

The first step was a credibility assessment of each media report through comparison with the image-derived information. If a specific factual claim in a news article was corroborated or consistent with the image-derived information, it was assigned a '+' indicator. If the claim was inconsistent with the imagery, it was assigned a '-' indicator. If a factual statement was checked with overhead imagery and the result was inconclusive, it was given a '*' indicator. Other articles that could be linked to these factual claims were also assigned the corresponding indicators. The total number of the three indicators for each article provided a semi-quantitative way of assessing the believability of other factual claims in the article that could not - for purely technical reasons - be checked with the available commercial satellite imagery. If an article contained several verified facts, the other reported facts were deemed as believable. If an article contained numerous false or inconsistent statements, the remaining claims were not considered credible.

Appendix C illustrates how this credibility assessment was done and tabulates the results for each news report that described activity (or inactivity) in the Pokharan area. Media reports with at least one factual claim that was consistent with the image-derived information were classified as minimally credible and were used as evidence for identifying the activity at the Khetolai military range. Articles that could not be linked to any substantiated facts were excluded from the analysis. Articles with factual claims that could not be geographically linked to the Khetolai military range were also set aside.

The second step was an examination of each remaining item of evidence to determine whether it could be attributed to conventional, missile, nuclear, or innocuous activity at the Khetolai military range. This was done by researching the hallmark and peripheral signatures of these four activities and comparing the evidence with these signatures. A table of suppositions was compiled to determine whether each set of evidence could be logically attributed to at least one of the four activities (see Appendix D). [71] This table outlines the various plausible explanations for the evidence.

The third step involved the placement of the evidence onto a Venn diagram. A Venn diagram is a logic tool consisting of overlapping circles that represent different sets and the relationships between them. [72] In this case, each circle represented a different explanation for the activity at the Khetolai military range (see figure 3). If an item of evidence could be best explained by just one of the four activities, it was placed inside the corresponding circle and outside the other circles. If it could plausibly be attributed to more than one activity, it was placed into the appropriate overlapping area demarcated by the representative circles.

[Figure 3]

Figure 3: Venn diagram of the Khetolai military range. The circles represent the four separate explanations for the observed activity. Each item of credible evidence was placed onto the diagram in accordance with the table of suppositions (see Appendices C and D). Spatially linked items of evidence were connected together with a solid line, and evidence that could be spatially related were connected with dashed lines. Evidence that predated the 1995 allegations was shaded in order to make it distinct from the new evidence.



Analyzing the Venn Diagram of the Khetolai Military Range

Figure 3 shows the Venn diagram that was produced from the evidence and suppositions listed in Appendices C and D. Spatially linked items of evidence were connected together with a solid line, and evidence that could be spatially related were connected with dashed lines. In addition, since the diagram was drawn to address specifically the 1995 allegations against India, evidence that predated the allegations was shaded in order to make it distinct from the new evidence.

The analysis of the Venn diagram led to three main conclusions. First, the diagram shows that the Khetolai military range has a history of nuclear test activity. It was used for the May 18, 1974 nuclear test, and there is credible evidence from four different sources that indicates shafts were constructed in the early 1980s for two underground nuclear tests there. As shown in the diagram, this evidence predates the 1995 allegations. Second, the diagram shows that there is believable evidence that supports the claim of planned Prithvi field testing at the Khetolai military range. The December 19, 1995 Asian Age article that claimed such activity is believable because it made four other factual claims that were verified with the satellite imagery; it was the only news report that accurately described the conventional test ranges in the area as well as the sudden increase in activity at the Khetolai military range (see figure 3 and Appendix C). Moreover, there are five military operation areas nearby that could be well-suited as impact points for Prithvi field tests (see figure 5 inset). Third, the diagram shows that none of the image-derived evidence could be attributed to just one of the four possible activities; one cluster of the image information fit within all four circles, and another cluster fit in both the Prithvi and nuclear test circles. The latter cluster proved to be the most intriguing because it could be spatially linked to other items of evidence that fit in the nuclear test and Prithvi circles. This indicated that the recent activity was consistent with nuclear test preparations, planned Prithvi field testing, or a combination of both. Figures 4 and 5 show how the activity could be interpreted in this way.

[Figure 4]

Figure 4: Line map of the Khetolai military range. The figure shows how the various features can be interpreted as nuclear test preparations. The letters correspond with the features labeled in plate 18. The numbers denote the sequence of the deductions that were made to identify the spatially connected features.

In figure 4, the linear traces can be interpreted as the cable lines mentioned in the January 3, 1996 Hindustan Times article, oriented perpendicular to the prevailing winds. Such a cable layout would resemble the 1974 cable configuration, which was oriented similarly to keep any accidental release of radioactive debris away from the control point. If India followed the precedent it established in 1974, point G would be the control point situated southeast of the explosion sites (see figure 4). It would then logically follow that the shafts were located at the other end of the linear traces. At the end of the four branching traces, sites A and C - located one kilometer apart from each other - fit the profile of two shaft locations (see figure 4); both are surrounded by multiple layers of perimeter barriers, and both have identical rectangular structures where the same side intersects the inferred cable lines at the exact center of the fenced perimeter (see figure 4). This configuration closely resembles the layout that was used for vertical shaft tests at the US Nevada Test Site. [73]

Site B is equidistant from both of the inferred shaft locations. It is located 500 meters from both sites, which makes it suitable for serving as a diagnostics station for two nuclear tests. Such a station would receive and record the nuclear test data from the ground zero via cable just before the explosion destroyed the hardwired connection. From such a close vantage point, the time response degradation in the cable would be mitigated by the shortened cable length, thereby accommodating the fast rise time requirements for measuring nuclear phenomena from near instantaneous, explosive events. Diagnostic and recording instruments are typically placed close to the explosion point for this reason. [74] The control point could monitor the equipment at site B via a separate cable link along line BG (see figure 4). Site D is located near the entrance to the central area. It can be interpreted as a support base for nuclear test preparations with an unpaved road link to the inferred control point.

[Figure 5]

Figure 5: Line map of the Khetolai military range. The figure shows how the various features can be interpreted as planned Prithvi missile testing. The letters correspond with the features labeled in plate 18. The numbers denote the sequence of the deductions that were made to identify the spatially connected features.

The inset is a map of the Indo-Pakistan border in the Rajasthan Desert. The Khetolai military range is located at the X and the polygons are nearby military operation areas. The perimeter of the military operation areas were derived from the March 1991 1:500,000 scale Tactical Pilotage Chart (TPC) H-8C produced by the Defense Mapping Agency Aerospace Center.

In figure 5, point G can be inferred as a missile launch site. Although it lacks the wide radius turn loops and it has not been cleared of vegetation, the site is located at the end of a linear trace in a remote area. [75] The linear traces can be interpreted as unpaved road links that connect the launch point with Prithvi and fuel storage sites. The identical rectangular structures at sites A and C can be interpreted as the facilities for temporarily storing the Prithvi missiles, the transporter-erector-launchers (TELs), and volatile fuel (see figure 5). These buildings could be used for such a function given that both are heavily secured, isolated, and connected to the inferred launch point. The road link is approximately three kilometers long so test launches could be done at a safe distance from the basing areas, while still within reach of fire fighting units which could be situated at the support base (site D). It is oriented perpendicular to the prevailing winds, which would keep the toxic exhaust plumes away from the basing area. Site B is equidistant from the two inferred Prithvi facilities, and it is connected to the inferred launch point (see figure 5). It can be interpreted as the missile control point which could be used to coordinate the deployment of Prithvi TELs to the launch point, track the missile after launch, and communicate with the impact site. The cabling described in the January 3, 1996 Hindustan Times article could be ignition cables for firing the Prithvi at targets located within five different military operation areas (see figure 5 inset). [76] With respect to the inferred launch point, these operation areas could accommodate the Prithvi-1 minimum range trajectory of 40 km as well as its maximum range trajectory of 150 km. [77]

Furthermore, the field tests could be performed within the Prithvi theater of operations under realistic battlefield conditions while mitigating the chances of unduly alarming Pakistan. As shown in the figure 5 inset, the trajectories from the inferred launch point to the military operation areas would not put the missile on a flight path towards the border with Pakistan. The trajectories would either run parallel with the border or away from the border deeper into Indian territory. Although the Prithvi-1 could reach a small portion of Pakistani territory from the inferred launch point, it would be out of reach of all major civilian or military targets. [78]

Since there are items of evidence that point to nuclear test preparations as well as planned Prithvi field testing, it is within the realm of possibility that the Khetolai military range serves a dual purpose. That is, in the absence of an active nuclear testing program, the facilities and infrastructure at the range could be used to conduct a variety of Prithvi exercises for training missile crews. Such activity in the missile's planned theater of operation would be a logical follow-on to the completed series of Prithvi-1 missile development tests along the eastern Indian coastline. [79] If the need for nuclear testing arose later in the future, the facilities and infrastructure at the range could revert back to its original function as a nuclear test site. By assigning a dual role for the Khetolai military range in this way, the personnel, equipment, and infrastructure could be utilized in a routine manner while maintaining India's reserved nuclear testing capability.


Paper Sections Paper Indices

Table of
Contents		 Appendices Last
Page		 Next
Page


Paper by: Vipin Gupta and Frank Pabian
Webmaster


SAND 96-8252W