Hydrologic Applications: Monitoring Drought and Snow Cover - Remote Sensing Application - facegis.com
Hydrologic Applications: Monitoring Drought and Snow Cover

These views of frozen water carry us naturally into the last topic - hydrology - in this Section. In this context, hydrology refers to water distribution and management on land surfaces. We discuss here three of the four principal uses of satellites for hydrologic applications: drought assessment, runoff prediction, and flood monitoring/damage assessment. The fourth topic - drainage basin characterization - we do not consider.

First, several statistics: 1) There are approximately 2 million cubic miles of freshwater throughout the world, mostly in the top 173 m (570 ft) of the crust (both in and on); 2) most freshwater (70%) resides in Antarctic and Greenland ice; 3) there is ~30 times more freshwater as groundwater than in rivers and lakes; and 4) Lake Baikal (see below) has 20% of all lake freshwater. One added fact: water consumption (humans and animals) is about 400 billion gallons/day - about 65 gallons for each higher order creature on Earth.

A primary use of meteorological satellites is simply to measure the regional and continental distribution of normal, excessive, or sparse rainfall. TRMM is especially suited to that task. As an example, here is a six weeks summation of rainfall over most of Africa in August- September 2007. The rainy season centers around late northern hemisphere summer, at which time the amount of rainfall will determine whether drought or good harvest conditions will prevail for the year. 2007 has proved to be a period of better than average precipitation:

Rainfall in Africa, 2007.

We touched upon the use of space imagery to monitor vegetation stress and other effects of drought on page page 3-4. These tend to be seasonal or short-termed over years, with eventual reversal of the climatic conditions that bring about water shortages which threaten crops and habitats.

One significant long-term change is the gradual reduction or disappearance of lakes. Lakes tend to be ephemeral when considered at geologic time scales. However, notable shifts in climate - such as is now much talked about in terms of global warming - can sometimes destroy major bodies of water in a century or less. In the U.S., the Great Salt Lake (see Overview) has experienced some fluctuations of size and water level in the 20th Century. Drought often accompanies increased aridity and desertification which pose a threat to lakes that have survived intact since their host areas were first settled. Two drastic examples will illustrate this.

The first is at the southern end of the Sahara Desert which now appears to be enlarging at the expense of once vegetated central African lands. Lake Chad, which lies at the triple junction borders of the nations of Chad, Niger, and Nigeria, was once the 6th largest freshwater lake in the world. This AVHRR image locates it within the yellow brackets (lower left of Lake Chad Basin label):

AVHRR image of North Africa, showing major drainage basins and the location of Lake Chad (yellow brackets).

This body of water has been essential to nomadic peoples living around its shores. But its area and volume have been rapidly diminishing over the last half century, as the vegetation line pushes south. This image - actually a photograph taken by astronauts - shows its state in 1963. Below that are two Landsat images - the top acquired in 1973 and the bottom in 1997.

Astronaut photo of Lake Chad in 1963.
Landsat image of Lake Chad in 1973.
Landsat image of Lake Chad in 1997.

These views almost speak for themselves. In 1973, the southern part of the lake began converting to marshlands. By 1997, almost all of the open waters of Lake Chad had disappeared as both marsh and landfill replaced them. A time-sequential plot of water levels over a 120 year span, made from data processed by the U.S. Geological Survey, shows (unfortunately, with information in red too small to be legible here) the pronounced changes since 1960. Unless there is a miraculous reversal in climate and rainfall, the future for this lake is obvious: Chad has been Had.

Plot of Lake Chad water levels (variations relative to a benchmark elevation on the vertical axis) shown in yearly cycles with decade dates on the abscissa.

A comparable situation faces the both the Caspian and Aral Seas in southern central Asia, as seen in this AVHRR image. The somewhat saltier Caspian Sea (the larger body of water near the left edge of the view below) lies within Turkmenistan and abuts parts of southern Russia. The Caspian Sea holds the distinction of being the largest lake containing near-freshwater in the world. Lake Baikal in southeastern Siberia is the deepest (1.7 km or 5712 ft).

AVHRR image showing the Caspian Sea (left) and the Aral Sea (near center) in Uzbekistan.

The Caspian Sea is slowly losing water. The best indication of this is the Kara-Bogas Gol (Gulf) located on its eastern shore near the center. It is nearly filled in the AVHRR image. But this is now a rapid process. Two Landsat subscenes, in 1972 and 1987, show its progression in recent years, as salt is filling its interior.:

Kara=Bogas Gol, 1972 (Landsat-1).
Kara-Bogas Gol, 1987 (Landsat-4).

The Aral Sea in Uzbekistan has lost 50% of its area and 75% of its water volume since 1960. Some of this loss may be due to decrease in annual rainfall in that part of Asia, but the diminution is largely caused by diversion of water from rivers feeding this inland lake for use in agriculture elsewhere. This has impacted the means of living by residents around the sea: loss of suitable conditions for cotton growing and great dropoff in fish supply. The changes this has wrought are evident in the next two views made by Russian satellites. The first shows the Aral Sea in the mid-1970s. The second consists of images taken on three separate years:

The Aral Sea in 1976
The Aral Sea imaged in the years shown.

The reduction in water-covered area is obviously matched by a gain in land surface. The next three images, taken in 1973, 1987, and 2000 by Landsats, clearly reveals the growth in land area along a part of the Aral Sea.

Landsat views of changes around part of the Aral Sea betwen 1973 and 2000.

The trend towards shrinking of the Aral Sea has caused alarm in the Soviet spinoff countries of Kazakhstan and Uzbekistan, within which this body of water lies. It is an important resource whose loss would hurt fishing and other user groups. So, a plan underwritten by the World Bank to replenish water in the Aral Sea by damming its main outlet on the south has been executed. However, by August 2009 little has been accomplished; a continuing drought in the region has led to almost a complete drying up except for part of the northern Aral Sea.

 The Aral Sea in 2009.

Further east, along the border between Afghanistan and Iran, the Hamound Wetlands were still in good shape in the early 1970s, with natural vegetation and some cropland benefiting from the water carried in by the Helmand River. The extended drought in that part of the world converted this wetland to a wasteland of mud and salt. These Landsat images show this abrupt change:

Left: 1976 Landsat TM image of the Hamound Wetlands; Right: 2001 Landsat 7 ETM+ image of the same area, now dried up.

Space imagery is also of significant value in monitoring changing conditions within lakes, as we saw elsewhere in the Tutorial. Here are two ASTER images of Lake Ichkeul in Tunisia. The top one (November 14, 2001) displays the lake filled with sediments during a winter rainy season. The bottom image (July 29, 2005) examines a now clear lake that is being encroached by algae (red)(Utah Lake in the Overview-1 has a similar dual set of conditions):

Two ASTER images of Lake Ichkeul in Tunisia.

As discussed on page 3-4, the North American continent has been experiencing droughts of varied severities in the 1990s into the 21st Century. That occurring in the Summer of 2002 has been especially severe and widepread. Much of the East and a large percentage of the West were most seriously affected. One consequence: a large number of wildfires burning over 4 million acres. This map shows satellite microwave measurements converted into surface wetness across the U.S. for a 6 day span in August. This does not disclose the full extent of the drought (see page 3-4) because it reflects rainfall over just one week but it does help to assess the patterns of soil moisture that will influence longer term conditions that contribute to the water deficit underlying the on-going drought. For this week, the rain shortfall has been greatest in the eastern half of the country but in prior weeks the West has had similar deficiencies.

Microwave-determined soil moisture across the U.S. for the period August 6-12, 2002.

Thus, parts of the (48 contiguous) United States has been experiencing significant and worrisome drought conditions since the 1990s, largely caused by reduced rainfall and warmer temperatures. This pair of satellite images shows vegetation calculated as NDVI for parts of the western U.S. under normal rainfall conditions (top) and in a moderate drought state (bottom):

Normal and drought conditions in the western U.S.

NOAA and the U.S. Dept. of Agriculture for decades now have been using satellite imagery routinely to publish maps showing the degrees of drought in the 48 contiguous states. Drought conditions vary from year to year and even seasonally. Here are three maps for 2004, 2007, and 2009:

Drought in 2004.
Drought in 2007.
Drought in 2009.

As one might expect, these same agencies make predictions of conditions at least three months in advance. This map shows anticipated drought conditions from August to November 2009, as a forecast:

Three months of drought as predicted in August of 2009.

In much of the northeastern region, this drought has become troublesome and serious to severe right after the turn of the millenium. New York City and the surrounding metropolitan areas were forced to apply use restrictions as their major sources - reservoirs - shrank to as low as 10% capacity by volume. The next pair of images, acquired by the ASTER sensor on NASA's Terra, shows a reservoir in the Catskills to the northwest on September 18, 2000 and again on February 3, 2002, when the volume has been reduced to 58% of normal.

ASTER image showing Catskill reservoir on Sept. 18, 2000.
ASTER image showing Catskill reservoir on February 3, 2002

Obviously, at the other extreme Metsats can provide near real-time indications of weather conditions that portend severe storms and heavy rainfall. Over a longer period, satellite imagery can show potential flooding from spring thaws and image interpreters can estimate expected quantities of water runoff by monitoring snow cover over large regions. Satellite observations of surfaces blanketed by snow (in the U.S. principally in mountains and high prairies) suffice to measure the areal extent of the masses likely to melt. Meteorological satellites are particularly adept at recognizing snow - it appears bright white and usually continuous over large areas (clouds are either discontinuous or a dull gray). This is a typical Metsat view:

Snow in North America.

These four MODIS images show the extent of snow cover in different years:

MODIS images of snow cover in the years indicated.

Seasonal variation in snow cover from year to year in a smaller region is illustrated by this pair of Landsat-1 images of the central Sierra Nevada highlands.

Snowcover differences between a near normal runoff season (1975) and a drought year (1977) in the Sierra Nevada Mountains near Lake Tahoe, California as observed by Landsat.

However, we must determine thickness variations and packing densities from onsite ground measurements, in order to estimate the volume of runoff. This information is important not only for flood warnings and control but also to estimate water supply from reservoir fillings, river channeling, and aqueduct retrieval.

Similar estimates are made routinely to forecast runoff from the Southern and Central Rocky Mountains. This MODIS image shows the distribution of a record snowfall on March 22, 2002 in the Colorado Rockies. Most of the white is snow (a few clouds are in the scene).

MODIS view of snow cover in the Rocky Mountains.

Routinely, images from the NOAA and the GOES satellites are applied to determining the extent of snow cover. We can separate snow from clouds by differences in spectral absorption in longer wavelength bands. Both are highly reflective in the visible and photographic IR. The Thematic Mapper band 5 also can distinguish snow/clouds, and thermal responses vary, as well. Here is a color composite made from three NOAA-12 AVHRR bands (1,3,4 = RGB) showing snow in April 1995 in the northwestern U.S. as red:

Color composite NOAA-12 AVHHR image of snow (bluish-white) in the northwestern U.S., April 1995.

Major snow storms can cause heavy cover that leads to extensive flooding, if a rapid thaw occurs. Look at this GOES-8 coverage (top, below) of a massive storm that hit the northeastern U.S. on January 7, 1996. The bottom image taken after storm passage and return of clear skies shows the extent of snow cover.

Colorized GOES-8 image of a massive snowstorm in the northeastern U.S., January 7 1996.
Colorized GOES-8 image of same area as previous image after the snowstorm has passed and showing the snow cover on the ground.

The MODIS sensor on Terra caught a distinctive snow pattern in the central U.S. following a "clipper" storm on December 2, 2006. The snow is deposited in a narrow band that developed rapidly across Missouri into northern Illinois. This resulted when cold polar air met warm Gulf air (moisture-laden), making snow, with the front being guided by the jet stream moving eastward.

MODIS image of a snow band running across the Midwest following a December 2006 storm.

We are accustomed to think of snow as always white in appearance. But snow can take on other colors, notably blackish, to brown or tan. This ASTER image of the snow capping the San Juan Mountains of southwestern Colorado shows the snow to have a brownish color. This is real and not a lighting artifact. It is caused by a strong coating of dust blown into these mountains from the desert to the west. The reflectance of the snow is reduced, hence it tends to absorb solar heat and thus melts sooner than normal.

Brown snow in the San Juan mountains.

Source: http://rst.gsfc.nasa.gov