Turpan Depression
Xin Jiang, China

ES 775 Advanced Image Processing
Jue Jiang
Fall 2011

Introduction Image Processing
Conclusion References

a Landsat 4-5 TM bands 147 false-color composite image of the Turpan Depression


Turpan Depression is located about 150 km southeast of Urumqi, the capital of Xinjiang Autonomous Region. The entire depression is about 50,000 km2. Some geographers name the entire depression Turpan-Hami Basin, and divide it into Turpan Depression to the west, which is about 28,600km2, and Hami Depression to the east, which is about km2. The Turpan Depression is an interior drainage basin formed in a shear zone during the Late Permian by movement between the East European and the Angaran craton, which is located in the heart of Siberia. It was then deformed by the collision of Indian plate with the Eurasian Plate during the Cenozoic (Dahlkamp, 2009). Most of the depression is covered by subangular to subrounded dark quartzite gravels. It is surrounded by the Tian Shan in the west and the Bogda Shan (mountains) in the northwest. A rain-shadow desert is formed in the depression due to the mountain topography.

With a lowest elevation of an approximately 154 m below sea level, where the Aydingkol Lake sits, Turpan Depression is considered to be the lowest point in China, and the third lowest land surface on the Earth’s surface, after the Dead Sea Depression, which is about 413 m below sea level, and Lake Assal in the Afar Depression, which is about 155 m below sea level. It has an extremely arid and hot climate in which the evaporation of the area exceeds the annual precipitation. The highest record of the temperature in the depression is ~49 °C (NASA, 2009).

Wind flows to the northeast and northwest of the depression. The wind is strong enough to overturn trucks and trains. A strong wind storm with speeds more than 36 m/s carried a large amount of sands that buried the houses in the area about 50 years ago, and another storm destroyed the whole wheat crop in 1961. To avoid such disasters, shield belts have been built to reduce the wind speed over the farm lands (NASA, 2009; Watts, 2008). The Dabancheng Wind Farm, the biggest Wind Farm in Asia with a total capacity of 500 MW from more than 300 turbines, was built between the provincial capital Urumqi and the Turpan Depression because of the good wind condition (Wu, 2010).

Some vegetation including reeds, shrubs and weeds exist around the Aydingkol Lake, and the vegetation is absent at the areas away from the sources of water. Water is supplied by Karez Irrigation System, which collects water from the base of the mountains and channels the water to desired destination by wells, dams and underground canals. There are about 1000 Karez Systems in the depression (Zhao, 2006). Spring-fed oases and small lakes exist in the center of the depression.

Image Processing and Analysis

Landsat 4-5 TM datasets of the wind farm vicinity (path 142, row 30) recorded on 13 Sep. 1998 and 13 Aug. 2010 along with the Turpan Depression (path 141, row 30) recorded on 22 Sep. 1998 and 7 Sep. 2010 were downloaded from USGS Glovis, imported to IDRISI 16.0 as GeoTIFF, and merged together with Mosaic tool. The datasets were then resized by using Window function. The spatial resolution for the TM datasets is 30 m X 30 m, the images have 9146 columns and 4102 rows, therefore the entire study area is about 33,765 km2 (= 30 m x 30 m x 9146 x 4102).

NDVI (Normalized Difference Vegetation Index)

Active vegetation is the only common material on the Earth’s surface that reflects near-infrared radiation (TM band 4) and absorbs red light (TM band 3). NDVI = (band 4 - band 3)/(band 4 + band 3). The values fall within the range -1 to 1, in which positive values indicate active vegetation, and negative values represents lack of vegetation. Steps: 1) Haze Correction with Scalar; 2) Create NDVI images with VEGINDEX.

The 2010 NDVI image has a larger range of value than the 1998 NDVI image. The bigger positive value in 2010 NDVI image reveals that the vegetation in 2010 is more active than the vegetation in 1998. The vegetation covering around the depression decreased from 1998 to 2010, but the vegetation inside of the depression, where the farm lands and irrigated vegetation are, increased from 1998 to 2010. The water bodies apparent in 1998 were almost disappeared in 2010. This is probably caused by the irrigation system, by which water collected in the lakes and from the base of the mountains was transported to irrigate the vegetation inside of the depression.

NDVI Classification and Area Calculation

NDVI images are classified into two categories, vegetation and other land with Reclass tool. NDVI values from 0.1 to 1 are assigned to be 1, which indicates vegetation, and the rest values are assigned to be 0, which indicates other land. The area covered with vegetation is calculated with Area tool.

In 1998, the vegetation area is about 4905 km2, which accounts for 14.5% of the whole study area; whereas in 2010, the vegetation area is about 4862 km2, which accounts for 14.4% of the whole study area. So the vegetation covering has decreased about 0.1% from 1998 to 2010.

NDVI Composite

To make NDVI composite images, the real-number NDVI values need to be converted to byte-binary format. Steps: 1) Add 1 to NDVI and multiply NDVI by 125 with Scalar; 2) Convert real number to byte-binary scale with Convert tool; 3) make NDVI composite images by combining band 2, NDVI and band 5 as blue, green and red.

NDVI Composites have strong color contrast. Water bodies and glacial ice appear bright blue, vegetation shows bright green, and the rest lands, including barren ground and sand, are bright pink and purple. Aydingkol Lake, which sits in the southern part of the depression, is an intermittent salt lake. The water volume decreased from 1998 to 2010.

TM Bands 345 Composite

In the TM 345 composite image, vegetation appears green, water bodies and glacial ice appear blue and cyan, and barren lands and sand are pinkish.

The north side of Bogda Shan is glaciated. A segment of Tian Shan is seen at the southwest of the depression. The north windward side of the Bogda Shan has ice and more vegetation than the south leeward side. The glaciers on the top of the north side of the mountain range melt and partially flow to the depression, leaving drainage channels and alluvial fans on the north slope of the depression, and most of the glacial melt water flows into Junggar Basin along the north side of the Bogda Shan.

Alluvial fans that have high water content show shades of dark blue. A 2500-km2 area of sand dunes, named as Sand Mountain, is located in the eastern part of the depression. It is formed by the accumulation of the sand in the depression transported by wind and streams. The NE wind seems to be the dominant wind over the Sand Mountain based on the dune pattern. The linear feature along the north side of the Sand Mountain is Flaming Hills, which is an anticlinal structure.


Turpan Depression is the third lowest point on the Earth's surface. A 33765 km2 area of the Turpan Depression has been examined by using several image processing techniques with Idrisi Taiga program in this project. The geographical location made the depression extremely dry and hot. The excellent wind condition of the area enables the setting of Dabancheng Wind Farm. NDVI is practical to identify vegetation covering and water bodies. The decrease in water bodies and increase in vegetation inside of the depression from 1998 to 2010 may have resulted from the irrigation system or other factors in the area. By clasification and area calculation tools in Idrisi, the exact area of vegetation covering can be calculated, and 0.1% decrease in vegetation covering from 1998 to 2010 was concluded. The composite image by combining bands 2, NDVI and 5 corresponded with blue, green and red show dramastic color contrast. Water bodies are clearly shown in the NDVI composite images, and the water in Aydingkol Lake decreased from 1998 to 2010. TM bands 345 and 147 composites are efficient to detect sand dunes and alluvial fans.