Aiding Afghanistan

A glance at a map of Afghanistan reveals that it is a land of extremes. The rugged peaks of the Hindu Kush run down the spine of the country, flattening out into desert in the west. Rivers form thin blue lines that seep through the mountain valleys and venture across the arid landscape like arteries bringing life to the desert. The black dots of cities and towns cling to the blue lines, and vast voids fill the pockets between rivers. The harshness of the land alone makes living difficult, but political instability and violence and an extreme climate that flip-flops between prolonged drought and heavy snow, rain, and floods also pose serious challenges.

  Page 2
  Terrain map of Afghanistan

From his office in Sioux Falls, South Dakota, Michael Budde is about as far away from Afghanistan as a person can get. Yet for him, it is more than a distant country on a map: the remote-sensing specialist is part of the Famine Early Warning System Network (FEWS NET), which works to avert famine, particularly in Africa, Central America, and Afghanistan. Set up by the U.S. Agency for International Development (USAID), FEWS NET issues warnings to governments and international aid organizations when drought or floods threaten food supplies, giving these organizations time to take preventative measures before suffering sets in.


Afghanistan consists of rugged mountains with plains in the north and southwest. The country is mostly arid, with extreme temperatures in both winter and summer. (Natural Earth image courtesy Tom Patterson, U.S. National Park Service)


“These problems are so enormous that there are a lot of times that you wonder if your work really makes a difference,” says Budde. In Afghanistan, for example, as much as 59 percent of the population suffers from chronic malnutrition. “It’s in very small, specific situations where the things you do have an impact,” Budde concludes. His involvement in Afghanistan began in 2002. “USAID wanted to know where to distribute spring wheat seed. They wanted a map that would tell them which provinces were best suited for the seed,” he recalls. He and his colleagues at FEWS NET responded by gathering satellite data and information from weather stations to track growing conditions in the mountainous country.

Among the resources he turned to were satellite vegetation maps produced at NASA Goddard Space Flight Center. As FEWS NET began to monitor Afghanistan regularly, Budde also started to track water resources using snow maps created from data gathered by the Moderate Resolution Imaging Spectroradiometer on NASA’s Terra satellite. FEWS NET staff combine the NASA satellite data, weather and satellite information from the National Oceanic and Atmospheric Administration (NOAA) and the Air Force Weather Agency, and on-the-ground information in regular reports to the Afghan government. The reports, says Karim Rahimi, the former FEWS NET representative in Afghanistan, have become an important, reliable source of information for members of the Afghan Cabinet who need to know about food security in their country.

  Child in hospital bed

In a 2002 visit to Afghanistan’s best children’s hospital, Congressman Joe Pitts discovered high mortality rates related to malnutrition. (Photo courtesy Congressman Joe Pitts)


Vegetation Maps


It was only natural for Budde to turn to NASA for vegetation maps: NASA remote-sensing ecologist Jim Tucker has been producing the maps for FEWS NET for well over two decades, almost since FEWS NET was established in the early 1980s. His approach to monitoring crops from space relies on a wide, low-resolution (low-detail) view of the region.

“When we first started processing coarse-resolution data to look at vegetation, people said we were stupid,” states Jim Tucker in a matter-of-fact tone. The idea of using satellite imagery to track plant growth over a broad area was nothing new, but scientists, crop analysts, and land managers had always used the most detailed images they could get their hands on. After all, how can you tell how drought is affecting a field of corn if you can’t see the field? In the early 1980s, when Tucker started his work, the most detailed imagery available showed 80 meters in a single image pixel. So when he and his colleagues decided to monitor vegetation with a satellite that could “see” no better than 8 kilometers per pixel, it is easy to see why people thought they were crazy.

  Page 1Page 3
  Vegetation Index map of Afghanistan

The problem with detailed images, Tucker points out, is that the satellites that acquire the images only see a small part of the Earth every day. Why not just go outside to see how things are growing? Some problems are bigger than a single person on the ground or even a single high-resolution satellite image can assess. To see just how widespread famine is, Tucker realized they needed a bigger picture, a picture they could get only with coarse-resolution data.

But could less detailed images provide enough information to be valuable? “Temperatures and precipitation are similar on a small scale,” Tucker points out. All of the plants in the region covered by detailed images are affected roughly the same way. To see large-scale patterns of drought and rainfall, you need coarse resolution imagery.

“The idea is to have a lot of tools,” says Tucker. The lower-resolution, daily images reveal broad patterns, show how conditions change from day to day, and provide a long-term record, while the detailed images allow scientists to zoom in to get a close look at the effects of drought.

The images that Tucker produced are not like photos; they are a vegetation “index,” or scale, based on the kind of light reflected from the ground or vegetation surface in a region. Plant leaves absorb visible light for photosynthesis and reflect near-infrared light. If a satellite detects significantly more near-infrared light than visible light, the region is likely to be densely vegetated. By comparing the difference in intensity between visible and near-infrared light measured over crop areas in current imagery with the difference measured at the same time in past years, Tucker can tell how leafy a crop region is compared to normal. Plotting out the anomalies reveals where vegetation is thicker than normal because of good conditions or thinner than normal as a result of drought. [To learn more see Measuring Vegetation on the Earth Observatory.]


This map shows Afghanistan’s vegetation index for August 13-28, 2005. Deep green indicates dense vegetation, and earth tones indicate sparse vegetation. Most of the country is sparsely vegetated. (Map courtesy FEWS NET)

  Rain-fed Agricultural Areas map of Afghanistan

“It works well because you have a strong relationship between precipitation and photosynthesis,” says Tucker. In non-irrigated grass and shrublands, the more rain a plant gets, the more leaves it can produce, and more leaves mean more photosynthesis. According to Tucker, “Satellite [vegetation] data are a better representation of rainfall than a scattered network of ground stations. Satellite data provide a consistent, unbiased way to see what really happened on the ground.”

But perhaps the most valuable aspect of using coarse-resolution imagery is the long-term record that scientists can build from it. Low- and medium-resolution sensors—the Advanced Very High Resolution Radiometer (AVHRR), Spot Vegetation, and the Moderate Resolution Imaging Spectroradiometer (MODIS)—see the entire Earth every day. Tucker combines the daily images into ten-day composites to minimize cloud cover. By contrast, a high-resolution sensor may only acquire a single image during the same period. If it is cloudy, then there is no record of the crop area for that ten-day period. Since the low-resolution composites are made for every ten-day period of every year, the method provides a way to contrast current conditions with past conditions during the same period. For example, knowing that the first ten days of June 2005 were drier than the first ten days of June 2004 and much drier than the average of similar periods over a 20-year record can help decision makers understand the seriousness of the drought. The record also shows vegetation change over time, which could be valuable when tracking climate trends, monitoring natural resources or agricultural production, or reducing human suffering.

“FEWS NET saw NASA work and asked if we would be willing to work with them,” Tucker recalls. More than two decades later, he and his team are still supplying FEWS NET with regular vegetation maps. “It’s an excellent example of the value of satellite data,” says Tucker. “I’m proud to be involved in FEWS NET.”


Along Afghanistan’s northern border, most of the crops are rainfed. This map shows the vegetation index of the rainfed agricultural regions for August 13-28, 2005, with all other types of vegetation removed. Deep green indicates dense vegetation, and earth tones indicate sparse vegetation. Maps such as these isolate rainfed crops, making it easier for decision makers to gauge growing conditions quickly. (Map courtesy FEWS NET)


Floods or Famine: Water in Afghanistan


The vegetation images reveal how plants are growing across a wide region, but they aren’t the only things that FEWS NET uses to detect famine. “FEWS NET uses a convergence of evidence,” says Budde. “One thing may tip us off, but we look at many things to determine what is going on.” In addition to the vegetation index maps, FEWS NET tracks rainfall and temperatures over crop areas. They match the weather factors with the growing cycle of the crops being produced to see if rain and sun are coming at the right times in the crops’ development. They project how much water the crops will need and compare that with the amount of water available. They chart the locations of agriculture and rangeland so they know what kind of vegetation is growing in each area. All of this information is combined with information about the people living in the region to predict how they will be affected by a potential crop failure.

Experienced in monitoring rainfall and its effects on crops in Africa and Central America, Budde and other FEWS NET scientists immediately faced a new challenge when they began to monitor Afghanistan in 2002. Afghan farmers get their water from melting snow, not rainfall. “When everything came together, we realized that snow cover is of vital importance to water availability for an area,” says Budde. Snow cover was an issue they hadn’t had to consider when monitoring Central America or Africa.

  Page 2Page 4
  Filling water canisters

Like the U.S. West, water in Afghanistan most often falls in the form of mountain snow. The snow pack melts gradually through the spring and summer, supplying irrigated farmland with water through the growing season. Unlike the U.S. West, where an intricate system of reservoirs preserves water so it can be used when crops need it, Afghanistan has limited means to store water.

“The infrastructure for irrigation water has been destroyed,” says Budde. Decades of conflict has left canals and reservoirs in rubble. Much of the remaining infrastructure is old and in poor condition. The loss of infrastructure means that snow that melts too early or too quickly can’t be used to water crops later in the season. “More than seventy percent of agriculture is irrigated in Afghanistan,” Budde explains, so if the timing of the snow melt is wrong or snowfall is scarce, a poor harvest and hunger are likely to follow.

But how do you monitor snowfall in rugged mountains where roads are hazardous and violence is commonplace? “Ground truth is hard to get in Afghanistan,” Budde agrees. “We are not able to get out in the field as easily as in other countries because of security.” In the absence of ground measurements, remote sensing is crucial. Among those sensors used to monitor snow is the Moderate Resolution Imaging Spectroradiometer (MODIS) on NASA’s Terra satellite.


Afghanistan residents line up to collect water. Many parts of the country lack adequate water supplies and water storage facilities. (Photo courtesy U.S. Geological Survey)


MODIS Snow Maps


Looking down on the mountains from a high airplane, bright white snow is easy to spot against the darker background of rocks, earth, and plants. An airline passenger might have more difficulty telling the difference between white clouds and white snow, unless she were wearing shortwave-infrared goggles. “If you could see in shortwave infrared (where light has a wavelength of 1.6 micrometers), snow would look dark, and most clouds would look bright,” says Dorothy Hall, Senior Scientist at NASA Goddard Space Flight Center.

Objects take on their color based on the light they reflect—grass is green, for example, because it reflects green light, but absorbs other wavelengths. Snow absorbs shortwave-infrared energy, so little of that energy is reflected back to a goggle-wearing, high-flyer, to whom the snow would appear dark. Clouds, on the other hand, reflect shortwave infrared, so they would appear very bright.

  Page 3Page 5
  Afghanistan snow cover on February 25, 2005

The MODIS sensor can detect energy in a shortwave-infrared part of the electromagnetic spectrum, so it has the ability to separate snow and ice from clouds, says Hall. The problem, Hall points out, is that there are so many clouds, it’s often hard to tell what the extent of snow cover is in any given image. To get around the problem, Hall and her team gather the imagery into eight-day and monthly snow maps that effectively screen out most of the clouds.

The first step in generating a snow map is to divide the world into a grid where each cell covers 500 square meters. This area corresponds to the highest resolution that MODIS can see in the shortwave infrared part of the spectrum. Each cell is then sorted into one of four categories: snow, no snow, cloud, or no data. If MODIS sees snow in a 500-meter cell at any time during the eight-day period, the cell is marked as snow-covered on the map. If clouds covered the cell during the entire period, a “cloud flag” is placed in the cell, and if MODIS doesn’t detect snow during the eight days, the cell is classified as snow-free. When the grid is put together, the result is a snow map showing the maximum extent of snow during that eight-day period.


This false-color Terra MODIS image shows Afghanistan on February 25, 2005. Snow in the Hindu Kush Mountains is red; the barren desert landscape is blue-green; plant-covered land is green; and clouds appear orange or white. At the time MODIS acquired this image, heavy snows and intense cold had caused over 500 deaths in the region from avalanches, exposure, and illnesses like measles and whooping cough. (Image courtesy MODIS Rapid Response Team)

  Afghanistan snowpack

How does this help Afghanistan? The amount of snow in the mountains is directly related to the amount of water that will be available later in the year. The maps reveal how much of a basin is covered with snow, when the snow fell, and when it melted. But, says Budde, “The map doesn’t show the volume of water available.” To get that, FEWS NET scientist Guleid Artan developed a snow-water equivalent map. The map is based on a model that estimates how much snow fell in a region based on climate parameters such as relative humidity, atmospheric pressure, temperature at the surface, wind speed, and so on. Along with the MODIS snow extent, the model allows FEWS NET to estimate how much water is available in the snow. That information can point to impending drought and possible crop failure or to potential floods.

“This spring [2005] we used the output from the snow-water equivalent model to actually determine values of snow-water volume above certain reservoirs. We graphed that every day and compared this year to last year,” says Budde’s colleague, Jim Verdin, the International Program Manager at the U. S. Geological Survey EROS Data Center. “Last year is a fresh reference in people’s minds,” explains Budde, and the comparison helps people interpret the information in the context of recent experience. What FEWS NET saw caused them to warn the Afghan government that severe flooding was likely as temperatures warmed. In response, the government set up a committee to deal with the floods well before they happened, says Rahimi. In fact floods did sweep through the country in the spring of 2005, and it turned out that the FEWS NET precipitation forecasts and model predictions were one of the few completely reliable sources of information for the government committee, Rahimi adds.


Heavy snow cover in Afghanistan’s mountains on March 13, 2005, contributed to flooding as the spring melt began. This map shows snow water equivalent (the amount of liquid water in the snow). (Map by Robert Simmon, Earth Observatory, based on data from FEWS NET)


Sounding the Warning: FEWS NET


The springtime floods provide a good example of how FEWS NET works. The vegetation index maps, weather information, and water availability are digested into flood or drought warnings, which are folded into a FEWS NET bulletin and distributed to the Afghan government and aid agencies. “We identify food security issues in the bulletin,” says Rahimi, outlining the primary purpose of the bi-monthly report. The bulletin is sent to the president and cabinet members for discussion in cabinet meetings, as well as officials in each of the 33 ministries in the Afghan government and up to 400 individuals in donor agencies and international organizations. Between reports, FEWS NET sends out a weekly climate forecast, which predicts weather conditions for the next seven days. Distributed to the same people who get the food-security bulletin, the climate forecast can warn of possible weather-related disasters like severe storms, floods, or extreme temperatures.

Another example of FEWS NET in action occurred when drought threatened food supplies in 2004. “Last year [2004], FEWS NET forecast a serious shortage of water. The Cabinet formed a commission to respond,” reports Rahimi. The commission visited those provinces that FEWS NET warned were at high risk. “In these two assessments [FEWS NET’s satellite assessment and the government’s ground assessment], they found that 26 provinces were in trouble. They sent an appeal to the international food aid community.” Because the appeal went out before a serious famine developed, aid organizations had extra time to gather the resources needed to prevent disaster.

The FEWS NET reports are helpful because, as Andrew Pinney explains, the number of people who need food aid in Afghanistan is highly variable. Pinney is an advisor to the Vulnerability Analysis Unit of the Ministry of Rural Rehabilitation and Development in the Afghan government. With as much as 85 percent of the population economically dependent on agriculture, swings in the weather can lead to a rapid change in fortune. The possibility for rapid swings in the weather and food security underscores the value of including satellite observations in the FEWS NET reports. Satellites regularly provide a view of crop and water conditions in the entire country at a single glance—something that would be impossible to do otherwise. Afghanistan’s decision makers are anxious to get the FEWS NET reports as they try to assess the needs of the people, says Rahimi.

  Page 4

Sowing wheat
Harvested wheat

USAID provided 7,000 metric tons of wheat seed to Afghanistan in the spring of 2002. (Photos courtesy USAID)