These images center on the town of Mutarara, Mozambique, in southeastern Africa.

Look at the false-color satellite images from 1973, 1985, 1992, 1994 and 1995. What can you tell from them? It is hard to read the subtle changes, especially at this scale. But in these years Mozambique was embroiled in civil war, and expert interpretation of these images at full resolution reveals the war's effects.

Mozambique: three decades of war

Mozambique was a colony of Portugal until 1975, when an 11-year war of independence ended with the establishment of an independent, Marxist government. But a 17-year civil war started soon after independence, with an internal military uprising that was supported by some foreign governments.

The civil war affected Mozambicans severely, especially in rural areas. Hundreds of thousands of people were killed. Over a million people fled the country, especially to Malawi, and more than a million others were displaced within Mozambique.¹ Many rural people migrated to the cities, especially along the coast where the government maintained control. The country went into severe economic depression.² Agriculture was disrupted, so the country could not feed itself. By the late 1980s Mozambique had one of the lowest per-capita caloric intakes in the world.³

Mozambicans also suffered though an El Niño-related drought during the war's last year, 1991-1992. This regional drought was especially hard on Mozambique.⁴ People were already malnourished, and had few resources to fall back on. Even when relief food eventually flowed in from other countries, Mozambique's war-ravaged transportation system prevented efficient distribution of the food, and internal politics interfered with relief work. Mozambique was said to have the highest "misery index" in the world, with 5 million of 16 million inhabitants in need of food and medicine.⁵

Remote sensing: early warning

The U.S. Agency for International Development (USAID) runs the Famine Early Warning System (FEWS), which monitors drought-prone countries in Sub-Saharan Africa, to quickly detect droughts and crop failures which could lead to mass hunger and starvation. FEWS was established in reaction to the 1984 Ethiopian famine, which most of the world learned about only when thousands of people were already starving.

FEWS personnel monitor rainfall, crops, grazing and markets on the ground, but they also use satellite data to monitor large areas of cropland, by using maps of NDVI (Normalized-Difference Vegetation Index). Healthy plants have the special quality of absorbing visible sunlight to photosynthesize, but reflecting infrared sunlight to keep cool. Therefore satellite data over a vegetated area should have a large difference between the infrared and visible bands, represented by a high NDVI value (here on a scale of 0 to 1). If the satellite flies over croplands during growing season and "sees" a low level of NDVI, then FEWS knows there may be a problem.

This is what happened in Mozambique during 1991-1992. By January 1992, governments and organizations in southern Africa described a looming food shortage so large that many foreign governments and organizations had to send their own people to check before they could believe it. This led to delays, and full-scale relief efforts did not start until June or July. The "notable exception", according to one African report, "was the Government of the USA which was able to verify the anticipated severity of the drought through its USAID supported Famine Early Warning System... [and] which took immediate action". U.S. grain shipments were among the first to reach African ports.⁶

Remote sensing: relief support

The USGS EROS Data Center (EDC) has supported FEWS since 1987, handling its satellite images and other electronic data. These data are available through EDC's Africa Data Dissemination Service.

For the 1991-1992 crisis, EDC scientists made a map showing where crops had previously been grown, to help assess the situation and target assistance. This Cropland Use Intensity (CUI) map was created by an experienced satellite-image interpreter drawing on large printed Landsat scenes, outlining areas with similar levels of cropping. A high-CUI zone is 70-100% agricultural fields. The lowest CUI category has 0-5% cropland. The photointerpreter used clues such as color, texture, and context to estimate CUI. The prints were then converted into the electronic map you see here.

Remote sensing: historical documentation

In 1995, three years after war ended, USAID asked the USGS to try documenting the migration of rural Mozambicans during the civil war using Landsat and other satellite data. These satellites can not see people, but they can see farms; for this reason, CUI was used again. In areas of subsistence agriculture, CUI approximates population density. CUI can also be used to spatially disaggregate agricultural statistics-- for example, if Sofala Province reports a 50% rise in planted grains, CUI can tell us where in the province the grains were planted.

Using the process outlined above, a small part of Mozambique was mapped for five different times, representing various stages in the civil war. The CUI maps showed that 22% of the land was cropped before independence (1973). This dropped to 5% by the end of the civil war (1992), and rebounded to 17% by 1995. These maps also included administrative boundaries, towns, rivers and roads. The maps distinguished mined roads, unmined roads, and demined roads (from December 1994 data; all roads were assumed to be unmined in 1973 and 1985, and any roads mined or demined in 1994 were assumed to have been mined in 1992). This is an example of how satellite data and other digitized data can complement each other when juxtaposed.

Click here to read a report on the making and interpretation of these maps.

Question

Look at the picture of the Sangadeze (Sangadza) River. This river meets the Zambezi River just north of these Landsat images. What do you see happening here? What is the hole in the foreground? (See the answer below.)

Footnotes

Thanks to SPOT Image Corporation for use of the 20 April 1995 scene.

1. Mike Sill, 1992, A geography of war: Geographical Magazine, November 1992, p. 45-50.

2. Steven Kyle, 1991, Economic reform and armed conflict in Mozambique: World Development, v. 19, no. 6, p. 637-649.

3. Sill, 1992.

4. Southern African Development Community Food Security Technical and Administrative Unit, 1993, Assessment of the response to the 1991/92 drought in the SADC region: Harare, Zimbabwe, July 1993, p. 5.

5. Shawn H. McCormick, 1993, Mozambique's cautious steps toward lasting peace: Current History, v. 92, no. 574, May 1993, p. 224-228.

6. SADC, 1993, p. 8-9.

Other references

Tom Crawford, Ron Lietzow and Jim Verdin, 1996, 1995 Pilot Study; Cropland Use Intensity Interpreted from Satellite Imagery of the Mutarara/Sena Area, Mozambique: Hughes STX Corp. Center for Global Change newsletter, Vol. 3, December 1996. Click here to read this article.

Kevin J. Dalsted and Fred C. Westin, 1996, An overview and evaluation of cropland use intensity in Africa: Sioux Falls, S.D., USGS EROS Data Center, Pecora 13 conference proceedings, in press.

Harold D. Nelson, ed., 1984, Mozambique; a country study: Washington, D.C., Foreign Area Studies, American University.

Satellite images

LM1179072007322790 (Landsat 1 MSS, 15 August 1973)

XF8F41030007061 (Landsat 4 MSS, 11 May 1985, held by the South African station of the Landsat Ground Station Operations Working Group (LGSOWG))

LT4167072009205510 (Landsat 4 TM, 24 February 1992)

YFYF16772940528 (Landsat 5 TM, 28 May 1994, held by the South African station of the Landsat Ground Station Operations Working Group (LGSOWG))

31433859504200752001X> (SPOT 3 HRV multispectral, 20 April 1995, courtesy of SPOT Image Corporation)

Maps

Army Map Service, U.S. Corps of Engineers, 1955 [compiled 1955 from 1939-1954 sources], Salisbury: International Map of the World SE-36, edition 2, series 1301, scale 1:1,000,000.

Mutarara/Sena CUI maps: U.S. Geological Survey, 1995, Mozambique Cropland Use Intensity [1973-1995]: Sioux Falls, S.D., EROS Data Center.

Four-country CUI map: U.S. Geological Survey, 1992, Southern Africa Region Cropland Use Intensity (CUI) [1986-1988]: Sioux Falls, S.D., EROS Data Center. Available from the Africa Data Dissemination Service (ADDS).

Four-country NDVI averages: U.S. Geological Survey, 1997, Normalized Difference Vegetation Index (NDVI), Dekadal Average for Africa, ed. 1: Sioux Falls, S.D., EROS Data Center. Available from the Africa Data Dissemination Service (ADDS).

Photographs

Thanks to John Robert Egan, former UNHCR Field Officer for Sofala Province, Mozambique, for these photographs.

River crossing: dry streambed believed to be the Sangadeze (Sangadza) River, at the road running along the Zambezi.

Zambezi basin #1 and #2: believed to be about 8 km east of Chemba (slightly outside these Landsat images), 20 April 1995.

Zambezi at Sena: about 1 km east of the Mutarara/Sena bridge, which was still impassable in April 1995.

a>

The streambed of this river is dry sand in this photograph. On the left, a group of women and children is collected on the remnants of a bridge, apparently doing laundry (they have a large yellow bucket). The hole in the foreground, surrounded by footprints, is a hand-dug open well. The white vehicle with the blue flag is carrying workers from the United Nations High Commissioner for Refugees, who are assessing the reintegration of returning refugees.