Chapter 15. Use of satellite remote-sensing techniques in West Africa*

* Contributed by the ILO.

INFORMATION ON RESOURCES such as forests, grasslands, soils, water, minerals and materials is necessary in order to enable developing countries to exploit their full potential. Inadequacy of information often leads to a failure to make crucial decisions or to the incorrect decisions being made. However, with the financial constraints facing these countries it is sometimes impossible to embark on detailed information collection by conventional means. In addition, traditional methods of resource information collection might not be capable of detecting rapid changes.

One useful tool which could aid decision-makers in improving resource base information is remote sensing which means “the collection of information without direct contact, by use of such instruments as cameras, radar systems, acoustic sensors, seismographs, magnetometers and sonar.”¹ It involves data collection in ultraviolet to radio regions of the electromagnetic spectrum thus neglecting acoustic, seismic and some other sensors. Remote sensing has a long history in photointerpretation but in recent times it is characterised by the use of new sensors such as thermal infrared and radar, the use of satellites such as the American LANDSAT series² to collect data and the use of computers to analyse them. LANDSAT orbits the earth 14 times a day at an altitude of 920 kilometres and returns to the same orbit each 18 days recording the same set of images. It has two sensor systems, a television sensor, the Return Beam Vidicon (RBV) system and a multispectral scanner (MSS) which record differences in sun reflectance on the earth’s surface.

The main characteristics of the LANDSAT data are their worldwide and repetitive coverage; synoptic view; uniformity over time and over large areas; multispectoral nature; availability in digital form for computer analysis; planimetric (near orthographic) images; the ready availability, ease of use and inexpensiveness of data (as of October 1982, a 18.5 cm × 18.5 cm colour print covering 34,000 square kilometres cost US$45).

Remotely-sensed data are most useful when combined with other sources of information such as topographic soils or geological maps.

Applications of remote-sensing techniques are varied and include the following:

- land resources and soil surveys;
- water resources;
- forestry and wildlife;
- fisheries (coastal and inland fisheries, marine fisheries);
- pasture and rangeland development;
- natural disasters: monitoring and anticipation;
- agricultural statistics;
- thematic mapping.

One limitation of remote-sensing satellites is that when they are beyond the range of a ground-data receiving station the data have to be recorded and transmitted as the satellites pass receiving stations. These recorded data are usually unsatisfactory. However, increased demand would make it possible to install more earth-receiving stations. For LANDSAT the common reception radius for each station is 2,780 kilometres.

I. THE UNITED NATIONS AND REMOTE SENSING

At its twenty-first session (1978),the United Nations Committee on the Peaceful Uses of Outer Space (COUOS) recommended that two international remote-sensing centres be set up within the United Nations system. Upon endorsement of the General Assembly in that same year, the two centres were opened in the United Nation’s Division of Natural Resources and Energy, Department of Technical Cooperation for Development (DTCD) in New York and in the Food and Agricultural Organisation (FAO) in Rome, respectively. The former was to be concerned with the area of non-renewable resources while the latter was to deal with renewable resources.

The FAO centre was established in January 1980 to undertake the following functions:³

- provision of advisory services and technical assistance to member States;
- provision of training in remote sensing;
- support to FAO’s worldwide field programmes;
- coordination of space activities at Headquarters and in the field;
- liaison between FAO and other major organisations concerned with space applications.

This centre holds microfiches and microfilms of all LANDSAT imagery and maintains a browse system for imagery and imagery analysis. It assists in the following areas: training, setting up national remote-sensing centres, conducting pilot studies and advising no less than 80 member states on request.

The second centre in DTCD participates in seminars and runs training programmes.

In this chapter, two case studies of LANDSAT applications in West Africa are presented. These are in agricultural forecasting and water resources survey respectively.

II. RICE PRODUCTION FORECASTING IN MALI AND GUINEA⁴

The aim of the project was to apply remote-sensing methodologies for rice culture inventory in West Africa. This would assist decision-makers in the field of rice culture management and crop production forecasting for optimisation of food supply. The project is financed by the European Development Fund which is involved in large projects for rice culture in the deltas of the Niger and Bani rivers both in Mali and Guinea.

Rice culture in the Niger Delta depends entirely on the hydrology of the Niger river for water availability. Submersion of the crop by flood waters is controlled by extended dykes built along the river and by floodgates. The success of the culture depends on synchronism between plant development and the river flooding which is itself dependent on the amount and distribution of precipitation over the upper basin in the mountain ranges in Guinea. There are several weeks between precipitation and river flooding due to the long transit time of water through the marshy areas along the river shores. The final rice production depends on the water availability during the rice development cycle (July to November/December).

Three periods can be distinguished in the rice development cycle:

- July to mid-September: under rain-fed conditions, sowing, emergence and tilling;
- September to November: submersion by the flood waters: booting, heading and flowering;
- November to December after the water withdrawal: maturation phase.

Remote sensing can be used to tackle two sets of problems, namely hydrological and agronomical.

Hydrological Problems

Prediction can be made of the extension of the flooded areas, first, before sowing for optimising the area to be sown (thus limiting seed losses), as well as for determining the rice varieties to be planted (floating rice in deep waters and erected rice in shallow waters); and second, at the end of the rain-fed period (July to mid-September), in order to estimate the flooded areas where rice harvest is expected.

Agronomical Problems

Prediction can also be made at the end of the rain-fed period, of the size of the areas sown with the floating and erected rice varieties and the extent to which each has survived through the excessive water stresses. At the end of the period of submersion by flood waters, prediction can be made of the area which has been sufficiently covered to ensure maturation, and after the third period, it is possible to estimate areas of rice which have reached maturity.

The report is concerned only with the agronomical problems. In order to determine the harvestable rice, assessment was performed at two stages: three months and one month before harvest.

Methodology and Results

The Tamani area was chosen for the study. The methodology consisted of visual interpretation and digital analysis of two pictures taken by the LANDSAT satellite. The first was a LANDSAT-MS image of November 28, 1975 at the beginning of the maturation phase and the second, a LANDSAT-RBV image of October 5, 1980 at the beginning of booting, three months before harvest.

In the first case, a landcover map was drawn from the visual interpretation of the LANDSAT image under its colour-composite configuration at a scale of 1:100,000. Harvestable areas were clearly observed at this stage. In addition, a digital analysis distinguishing water bodies, rice, dry crop areas and savannah, was made. The estimated areas of harvestable rice in both cases were compared with ground-truth information.⁵ These are shown in Table 15.1 below.

Table 15.1: Inventory of harvestable rice in Tamani Region, Mali - Interpretation of LANDSAT-MSS scene of November 28, 1975

Method of Calculation	Rice Areas (hectares)	Error (percentages) of ground truth
Visual interpretation	13,220	3
Digital analysis	10,554	-18
Ground-truth	12,828	-

Source: A. Berg and J.M. Gregoire, “Rice Production in West Africa (Mali and Guinea: Niger-Bani Project)” in Remote sensing for developing countries. Proceedings of EARSel-ESA Symposium, Austria. April, 1982.

In the second case, mapping of the flooded areas during booting was done. The rice fields were submerged under flooding waters and a black and white positive film at a scale of 1:125,000 was used to determine the extension and localisation of the flooded rice fields. These were then compared with those of harvestable rice plots recognised in December three weeks before harvest on infra-red photographs at 1:15,000 scale. Table 2 shows the comparison between the cultivated “flooded” and “harvestable” areas on the Tamani rice perimeter in 1980.

Table 15.2: Cultivated, flooded and harvestable rice areas on the Tamani rice perimeter (1980)

Sub-perimeter	Cultivated (hectares)	Flooded (hectares) in October (from LANDSAT RBV image)	Harvestable (hectares) in December from IR photographs
I+II	1396	1511	1317
III	683	658	642

Source: A. Berg and J.M. Gregoire. “Rice Production in West Africa (Mali and Guinea: Niger-Bani Project)”, op.cit.

Although seemingly satisfactory (3 to 18 per cent error in 13,000 hectares) the results of the 1975 analysis required some improvement. The second attempt increased the time lapse between forecasting and harvesting to three months. The forecasting on the basis of flooded rice fields three months before harvest provided fairly accurate results. This indicated that in 1980 the extent of flooding was the main limiting factor affecting rice production. However, this might not be the case in other years as other factors, such as the rain-fed period and crop vigour at the end of that period, affect rice production. In addition, not only the extent but the quality (occurrence at the right time) of flooding affects final production. It is felt that other satellites such as LANDSAT-D and SPOT could assess crop vigour at a phenomenal stage.

The result of this study has demonstrated the feasibility of utilising remote-sensing techniques for forecasting of crop production in developing countries. However, at this stage, this technique should be considered as a complement rather than a substitute for the conventional methods.

The importance of accurate crop production forecasting for the formulation of policy as well as national food targets cannot be overstressed. Developing countries can benefit from remote-sensing techniques in this regard provided they can gain access to and process data from the RS satellites. The availability of images at appropriate times remains a big problem as the present number of receiving stations in developing countries is quite small.

III. HYDROGEOLOGICAL INVESTIGATION IN DAMAGARAM - MOUNIO, NIGER⁶

Damagaram Mounio is situated in the south of Niger, 1,000 kilometres east of the capital, Niamey. It lies in the dry Sahel zone and covers an area of 20,000 square kilometres. Its population is 500,000 inhabitants living in 1,000 villages.

As in other areas of the zone, water supply has posed serious problems since 1973. The average annual precipitation of 500 mm has dropped to below 150 mm since that year. The rainy season is from June to September and precipitation reaches a maximum in August. The most dry months are November to April when precipitation is minimal. Since 1973, drought conditions have prevailed during this period. Due to the heavy drought, many sources of water have dried up, growth of trees reduced and cultivated areas eroded.

Geologically, the area is made up mainly of intensive granite and metamorphic rocks such as gneisses quartzites and schists of pre-Cambrian age. Sandstones of various ages and granites of Jurassic age are found in places.

A hydrogeological survey of the area, carried out in 1976-77, consisted of a study of all available data and reports and classical hydrogeological reconnaissance and interpretation of aerial photographs of the area around 200 villages afflicted by water supply problems. A geophysical investigation consisted of 79 drillings out of which 50 per cent of the boreholes were positive for exploitation of hand pumps.

An extension of this project is financed by the Danish Government. Three hundred boreholes will be drilled and 100 wells dug in this second phase of the project. In addition to the classical field activities an investigation of the application of satellite data will be carried out to:

- elaborate a satellite data map to be used by hydrogeologists in their work during exploitation;

- interpret information contained in the satellite data and determine the information extraction techniques which can be applied to hydrogeological studies.

The negative drillings of the geophysical survey were correlated with characteristic area information from the satellite data by digital-processing techniques. It was found that areas where the crystalline bedrock was slightly overlaid by sandy formation were negative zones for drilling.⁷ All drilled boreholes in this area were negative even though the sites were located after taking into account detailed aerial photo interpretations and geophysical considerations. Variations of the nature and depth of bedrocks have to be obtained in order to accurately interpret the hydrogeology of the area. Details of the relief of these areas were obtained by using linear transformation of the original four MSS bands.

A re-examination of the results of 1976-77 drilling showed that 80 per cent of the boreholes located outside the negative zones (identified from the satellite imagery) for drilling were positive. These results were confirmed by a drilling campaign performed in the autumn of 1981 in which about 50 boreholes were drilled.

Seasonal variations of the LANDSAT images showed maximum changes while areas in which the bedrock was overlaid by sandy formations showed minimal changes. This analysis of seasonal variations is of interest in areas where a dense vegetation cover would otherwise limit the capability of interpreting the geological background.

A ground-water model, which would aid hydrogeological studies of the area, is to be established on the basis of data obtained from this initial study.

The use of remote sensing to complement conventional investigations is again demonstrated in this application. It is felt that costs of rural water supply programmes would be reduced by more accurate location of feasible sites for boreholes and wells.

IV. SOME CONCLUDING REMARKS

As with any new technology, the introduction of remote-sensing techniques in developing countries requires education and training of technicians to read and analyse the data. (Even in the advanced countries like Japan and the United States there is a shortage of trained or qualified personnel to make judicious use of remote-sensing data.) Education and training for developing countries should have two objectives. First, it should aim at mastering airborne and space-earth observation techniques and applications and using them for the inventory and management of national resources. Second, it should create national expertise to negotiate systems specification and systems implementation of remote sensing suited to the particular needs of developing countries.⁸ There is a complete lack of facilities for the lower-level personnel even though courses exist for high-level training (some are for M.Sc and Ph.D degrees).

In spite of the attractiveness of remote-sensing techniques for resource investigation, a number of countries have expressed concern about the political implications of their use - the possible threats posed to national sovereignty, security and general well-being. The international community has to ensure, through legislation or conventions, that access to information is controlled and that its misuse is avoided.

Until recently, systems developers, owners and users have been mainly concerned with the technological applications of the techniques. Hence the technology “push” dictated the availability and economic value of information extractable from space observations. The potential economic benefits have only been partially obtained in the absence of systems geared to the optimum use of information.

At present, there is a growing feeling that definite strategies and policies need to be formulated prior to increasing investment on research and development and systems design. The degree of centralisation of collection, processing and interpretation of data is a matter for the national and international remote-sensing agencies to resolve.

NOTES AND REFERENCES

1. B. N. Haack: “Landsat: A tool for development”, in World Development, Vol. 10, No. 10, Oxford, October 1982.

2. Other proposed remote-sensing satellites are the French SPOT and Japanese ERS-1.

3. J.A. Howard: “Remote Sensing in developing countries: FAO’s international experience”, in Remote sensing for developing countries, Proceedings EARSel-ESA Symposium, Igls, Austria, April, 1982.

4. See A. Berg and J.M. Gregoire: “Rice production in West Africa (Mali and Guinea: Niger-Bani Project)”, in Remote sensing for developing countries, Proceedings of EARSeL-ESA symposium, Igls, Austria, 1982.

5. These are results from actual ground measurements and observations.

6. See paper by J.P. Zariryadis: “Application of satellite data to hydrogeological investigation in Damaragan - Mounio Niger”, in Remote sensing for developing countries, Proceedings of EARel-ESA Symposium, Igls, Austria, 1982.

7. In these areas all the drillings were unsuccessful as underground water could not be reached.

8. See C. Voute: Education and training on remote sensing, invited review paper presented at IGARRS, Munich, June, 1982.