Figure 1. The Aral Sea area is now a growing desert area.
Occurring in arid and semi-arid regions around the world, dust storms are a symptom of serious land degradation caused by human activities. The storms are also a problem in their own right. In Central Asia, large-scale anthropogenic changes have led to the formation of a new dust-raising region: the exposed and dried bottom of the Aral Sea (see figure 1). The wind picks up the newly exposed salt, dust, and pollutants and distributes them over a significant part of the basin in an ongoing ecological and human-health disaster.
As part of an effort to develop the cotton industry in the former USSR, the area of irrigated land in the Aral Sea basin rose from 3.442 million x 103 km2 in the 1950s to 7.86 million x 103 km2 at the end of 1990s. The side effects have been severe: The inflow from the Amudarya and Syrdarya Rivers (the main rivers feeding the Aral Sea) has decreased from 55 to 60 km3 per year in 1960 between 4 and 5 km3 per year currently. Water salinity has increased from 10 to 30 g/l.
Figure 2. The shrinking of the Aral Sea.
The exposed bottom of the sea dries into desert that includes areas of sand, clay, and salt-pan (solonchak). The area of this new Aralkum desert is 40,300 km2 (see figure 2). Years of uncontrolled use of herbicides, pesticides, and defoliants polluted the sea water, and they are now pollutants in the dried bottom. For example, water in the Amudarya River contains manganese at 1.3 to 2 times the maximum permissible concentration (MPC), iron at 1.5 to 3.3 times MPC, lead at 5 to 10 times MPC, cadmium at 6 to 8 times MPC, oil products at 36 to 46 times the MPC, phenols at 400 to 1000 times the MPC, and other toxic compounds.1
These developments provide the basis for powerful wind-driven (aeolian) processes that pick up the salt and dust and deposit them elsewhere. In 1975, 1990, and 1998, the wind transferred an estimated 45, 90, and 180 million tons, respectively, of salt and dust from the dried bottom of the Aral Sea.2, 3 Salty crusts form in adjacent areas, reducing the productivity of arable and range lands and making fresh water more saline.
For the last 25 years, the Aralkum desert has been expanding at about 1,000 km2 a year. The natural factors that determine the potential danger of the landscapes of this area as a salt and dust source are the relief of the dried bottom, the lithology of the ground deposits, the salinity of the wind-processed layer, the depth and salinity of the ground water, the temperature and wind, the dynamics of aeolian processes, the composition and density of vegetation growing on the dried bottom, and the distance the sea has retreated from the area. studying the landscape
Our study of the salty-dust storms in the Aral Sea region based on remote-sensing data, in addition to monitoring salty-dust storms, seeks to reveal where the dust and salt are being raised. The latter could allow the use of phyto-amelioration and phyto-remediation measuresusing plants to stabilize shifting sands and to clean up contaminated soil and water, respectivelyto reduce the formation of salty-dust storms.
We first must discriminate between three main types of landscapes: solonchaks (which exist in the littoral areas where ground water level is shallow and is connected with the surface by capillary action); takyrs (which form as ground water level decreases); and sandy areas (which form as a result of aeolian processes). Takyrs are the most active source of salt-dust blowouts.
We use Landsat Thematic Mapper and Enhanced Thematic Mapper Plus satellite images to reveal the sites from which salt dust originates. They provide high spatial resolution (30 m) and sufficient temporal resolution (16 days) to allow the repetitive monitoring of the same sites. We process the images using software that allows us to visualize, manipulate, analyze, measure, and integrate any type of geographic imagery and geospatial information into 2-D and 3-D environments.
We preprocess, enhance, and classify the data in multiple steps4: Data preprocessing
First, we perform atmospheric correction by adjusting the digital values for the effect of haze in the atmosphere. In most cases we don't know the number of parameters required to describe the atmospheric condition for the period of image taking. Instead, we apply the histogram minimum method, a simple procedure of adjusting brightness values for atmospheric effects. Next, we perform radiometric correction to convert image data from a discrete integer value into physical units such as spectral radiance and/or reflectance, permitting normalized comparison of the data. The last part of data preprocessing is performing geometric correction by registering the image with a coordinate system using collected-in-the-field Global Positioning System information or large-scale topographic maps. Data enhancement
The next step is data enhancement, which starts with calculation of spectral indices and albedo. For our purposes, two spectral indices are useful: the normalized difference vegetation index (NDVI) and crust index (CI).5 NDVI is usually used for monitoring specific crop conditions, but it can also distinguish vegetated from non-vegetated areas, or densely vegetated from sparsely vegetated areas, which aids in image classification. It is defined as:
where NIR and RED are the reflectivity over the bands from 760 to 900 nm and 630 to 690 nm, respectively.
In our study NDVI identifies the higher concentrations of green vegetation along irrigation canals, drainage collectors, and Amudarya and Syrdarya delta channels. Values of NDVI range from 0.1 to 0.4 in oases to 0.1 to 1 on the solonchaks (which are devoid of any higher vegetation). NDVI values on the sands and takyrs are around zero and slightly below. After removing densely vegetated areas from the image via a masking procedure, we detect the solonchaks using CI, which is defined as:
where BLUE is the reflectivity over the band from 450 to 520 nm. Solonchaks have relatively high CI values of about one compared to takyrs and sands with CI values between 0.7 and 0.8. After identifying the solonchaks and masking them, we distinguish between takyrs and sand by computing an albedo image from the reflectance image. We do this by summing up the reflectance of each band:
where xi is pixel value in each spectral band.
The takyrs, as barren lands with dense crust on top, reach higher albedo values (around 0.65 to 0.7) than the sands surrounding them (0.45 to 0.55). Data classification
Finally, we use a computer algorithm that provides unsupervised classification/clustering to identify groups of pixels within the image that have similar spectral characteristics, by means of distance measurements. First, the algorithm sorts the NDVI image into 24 classes in order to detect the densely vegetated areas. Next, it sorts the CI image into 24 classes to discriminate the solonchaks. Lastly, it classifies the remaining area into takyrs and sands by the same procedure.
Analysis of the time-series of Landsat images allows us to monitor active changes in the landscapes of the dried bottom of the Aral Sea. We believe temporal steps of two to three years are useful for study of landscapes dynamics. sensing dust storms
For monitoring of salty-dust events, we need images with high temporal resolution, such as those provided by NOAA's Advanced Very High Resolution Radiometer (AVHRR) and NASA's Sea-Viewing Wide Field-of-View Sensor (SeaWiFS) satellites. These images provide global coverage at high temporal resolution (two images per day), which lets us monitor the dust source, distribution and movement of the dust plume, dust transport routes, and areas of dust fall in a single scene. In our research we've used NOAA-14 and NOAA-16 satellite images from the Tiris receiving station established at the Department of Solar Energy and Environmental Physics, Jacob Blaustein Institute for Desert Research, Ben-Gurion University (Negev, Israel).
After obtaining data, we preview the image to estimate the cloudiness and noise and identify scenes indicating dust-storm activity. In most cases the salty-dust plume is transported above the Aral Sea surface, which allows us to observe it even before processing. We then build a multispectral image consisting of five spectral bands (visible, near-IR, and three thermal IR channels) and use the ERDAS Imagine program for further processing and interpretation.
The most common problem in monitoring dust storms is distinguishing them from cloud cover; thus, it is important to apply a cloud screening procedure. We use the fact that clouds have higher reflectance and lower temperature than the underlying earth surface and dust plume: therefore, simple visible (channel 1) and IR (channel 2) window thresholds help us detect the difference between clouds and dust. We classify a pixel as cloudy if the reflectance ratio of channel 2 to channel 1 is between 0.7 and 1.1. We can even distinguish a dust plume from cloud cover by eye simply by looking at combinations of two IR and one visible band, or two visible and one IR band.
Figure 3. A dust storm above the Aral Sea on 4 May 2000. The false color composite (RGB = bands 4,2,1) is formed from images obtained by the NOAA-14 satellite.
We observed five dust storms in the period of 19992002, all in the spring. In every case the dust was transported over the water surface, which made detecting the dust plume easier. According to our findings, the dust storms originated from the southeastern part of the dried bottom. The dust transportation route was to the southwest into irrigated areas of the Amudarya delta in all cases, and once to the west, into the pastures of the Ustyurt Plateau. We observed a dust plume on 4 May 2000 that extended 243 km over an area of 4362 km2 (see figure 3). For the first time, we could observe an additional source of the dust, which spread over an area of more than 2000 km2: the former island Vozrozhdenie, now a peninsula connected to the southern shore of the Aral Sea. The dust was transported to the west and reached the Ustyurt Plateau.
Human activity in the Aral Sea basin has resulted in changes that spread salt and dust from the dried-up bottom of the Aral Sea over a significant part of the basin. The intense transfer of salts by air is one of the most negative manifestations of the Aral disaster.6 It affects the living conditions of the local population, the biodiversity and biological productivity of the environment, and agricultural and pastoral productivity. Remote-sensing monitoring of the dried bottom of the Aral Sea allows us to detect the most active sources of dust transportation. With this knowledge, phyto-remediation can be applied in these areas. oe
1. "Sanitary rules and norms of protection of surface water from pollution,"1988, Ministry of Health Protection of the USSR, No. 15-21-76, p. 105.
2. A. Grigor'ev Al and M. Djogova, Proceedings of the Russian Academy of Sciences, 325, pp. 672-675 (1992).
3. G. A. Tolkacheva, Scientific and Methodological Basis Of Monitoring Of Atmospheric Precipitation In The Central Asian Region, Tashkent, Sanigmi, p. 204 (2000).
4. S. Herrmann, A. Karnieli, et al., Seasonal change detection for discriminating solonchaks from takyrs in the Karakum Desert (submitted to Int. J. Remote Sens.).
5. A. Karnieli, Int. J. Remote Sens., 18: 1207-1220 (1997).
6. "State of Environment" at www.grida.no/aral/aralsea. Global Alarm: Dust and Sandstorms from the World's Dryland. Yang Youlin, V. Squires and Lu Qi (eds.), UNCCD, pp. 172-204, Bangkok, 2002.
through the eyes of a satellite
Leah Orlovsky always wanted to be a Soviet cosmonaut. Unfortunately, it was not to be, she discovered after talking with staff at the Cosmonaut's Preparation Center. "My chances to become a cosmonaut weren't even equal to zero, but a negative value," Orlovsky says wryly.
Instead, she became a geographer, finishing her PhD at the Institute for Desert Research of the Turkmen Academy of Sciences, where she used satellite imaging to look at the Central Asian deserts from space. Her group was among the first who studied degradation processes in the desert at a large scale. Instead of digital images, she used hard prints, then created maps using transparent films.
"Ecological application of satellite imagery is a very fortunate and realistic substitute to my dream about space travel," she says. "It is still very exciting to come to the Gobi or Karakum Deserts and recognize the places you once saw and interpreted on the satellite image."
In 1995, Orlovsky emigrated to Israel with her family and got a job at the Jacob Blaustein Institute for Desert Research of Ben-Gurion University of the Negev. Orlovsky's goal is to attract attention of the Western scientific community to the environmental problems in Central Asia, such as water scarcity, soil salinization, and degradation of pastures. Currently, she spearheads several research projects in Central Asia, all connected with the study of different desertification processes and factors. She is particularly proud of a small Haloxylon aphyllum (saxaul) grove growing in the middle of the Karakum Desert. Here, researchers divert runoff water from indigenous water harvesting systems to create a little oasis from the wasteland.
She also wishes to act as a connecting link between her Central Asian colleagues and the Western scientific world to help her colleagues survive as scientists by means of international joint research. -Laurie Ann Toupin
Leah Orlovsky, Nikolai Orlovsky
Leah Orlovsky and Nikolai Orlovsky are researchers at the J. Blaustein Institute for Desert Research, Ben-Gurion University of the Negev, Sede Boker Campus, Israel.