Water Loss Analysis of a Disappearing

Introduction 1988 2009 Raw Images The is lo- cated on the border of Four Landsat 4-5 Thematic and Uz- Mapper (TM) images for each bekistan in central classification year were 1 Asia. The Aral is fed downloaded from GLOVIS . by two rivers: the Syr The images are from the same Darya in the north and season (summer) and were the in the selected to minimize cloud south. Once home to cover. Each image was gener- a thriving fishing in- ated by combining six spectral dustry, the Aral has di- bands to form one image. minished greatly in Bands 3-2-1 are displayed size due to water here for a true color image. diversions.

Since the 1960s, water has been withdrawn from Mosaicking

Syr Darya and Amu Darya in order to irrigate Composites of the four images crops, such as cotton. The size of the Aral has were generated in ENVI using steadily declined since water withdrawals began, georeferenced mosaicking. leading to the formation of two distinct water Images were then clipped to bodies (North and South Aral) instead of one only include the area of analy- continuous sea. Salinity has increased from 10 sis (South Aral Sea). A mask to 100 grams per liter, making the water unin- was used such that the same habitable for fish. clipped area was taken for each image year. Clouds and This remote sensing analysis will focus on vol- other atmospheric interfer- ume and surface area changes to the South Aral ences were obscured using Sea (also called the Large Aral) between 1988 clips from images with clear and 2009. ENVI remote sensing software and skies in order to ensure cor- ArcGIS were used to analyze satellite images. rect classification occurred.

Change Detection - Area Loss Classification

Change detection (CD) statistics were calculated Unsupervised classification based on the classified images (right) to deter- using the K-Means algorithm mine the water area loss in the South Aral re- was conducted for both analy- gion. Literature values for years closest to those sis years. Majority analysis analyzed are shown for comparison. was used to remove spurious pixels from areas dominated Water area was reduced by 89 percent from by another classification. Clas- 1988 to 2009 with most (82%) of the former sification errors are present water area converted to non-water (i.e. land). due to similarities in land cover pixel properties, but these CD Analysis Values Area (sq. km) % Change from 1988 errors do not appear signifi- 1988 Water 39,204 - cant in the analysis. 1988 Non-Water 12,254 -

2009 Water 4,517 -89% 2009 Non-water 44,187 +261%

Literature Values3 Area (sq. km) % Change from 1989 Volume Loss 1989 Water 36,930 - 2006 Water 14,325 -61% Images from the Shuttle Radar Topography Mission2 (SRTM) were used along with the ENVI Acknowledgements: Thank you to Dr. Eman Ghoneim and Simcha Levental for all of their help. classification (above) and 3D Analyst in ArcGIS. Analysis re- References: 1 USGS Global Visualization Viewer (http://glovis.usgs.gov/) sults revealed a loss of 947 2 Shuttle Radar Topography Mission (http://srtm.csi.cgiar.org/) km3 of water from the South 3 Micklin, Philip. “The Aral Sea Disaster.” Annu. Rev. Earth Planet. Sci. 2007. 35:47–72. Aral between 1988 and 2009. 4 Waltham, T. and Sholji, I. “The demise of the Aral Sea – an environmental disaster.” Geology Today, Vol. 17, No. 6, Nov.–Dec. 2001

Jeffrey Trull 3 SRTM Analysis Literature UEP 294 Year Volume (km3) Year Volume (km3) Remote Sensing 1988 1005 1989 341 May 6, 2010 2009 58 2006 81