Cover Image
close this bookC.I.S.F.A.M.: Consolidated Information System for Famine Management in Africa - Phase One Report (Centre for Research on the Epidemiology of Disasters - World Health Organisation, 1987, 33 p.)
close this folderCHAPTER 3: Information Systems, Databases and the CISFAM Project: Overview of General Findings
View the document(introduction...)
View the document3.1 GRID Technology Applied to CISFAM Data: Image Processing Potential
Open this folder and view contents3.2 CISFAM Informatic Considerations


CISFAM addresses issues in famine information systems within the health service sector, in the Sudano-Sahelian countries. Properly formulated, such a limited system can report, at international and national levels, on epidemiological, infrastructural, socio-economic profiles, providing corrective feedback and objective basis for effective resource allocation.

During the last decade, there has been considerable progress in developing statistical information systems in Africa, but serious deficiencies remain in terms of coverage and the availability of reliable and timely indicators of the human condition. Economic indicators have generally been available and have dominated both international and national development strategies despite their inadequacy to address distributional problems that are frequently the critical issues in development. Even where priority has been placed on monitoring and evaluation systems, it has been largely unrelated to long-term capability building.

The statistical services in Africa, due to serious resource constraints, are limited and on occasion non-existent. In the last two decades, famine interventions implemented by different agencies were dependent on data they collected themselves on the field or simply put in operation activities guided by the most publicised pieces of information. The governments often finding themselves in the position of having a multitude of agencies demanding (or as the case was, not demanding) information, lost control and relief assistance was inequitably distributed over the country.

CISFAM is founded on the concept of Management Information Systems (MIS) for health. Although, this concept is no longer controversial, a certain fuzziness exists about the elements and purposes of such systems. Data of many kinds are fed into a MIS, but unless selectively compiled and processed, they convey little other than archival data. The value of data and statistics in a management information system is judged by their usefulness in programme evaluation and policy formulation. In Africa, however, the value is questionable because data are subject on omissions and misreporting.

For comparative analyses in a multi-sectoral database like CISFAM, additional problems are introduced by the use of non-comparable definitions in data collection procedures and by differences in tabulation. Given the incomplete, fragmentary and defective nature of the data available, their interpretation must remain somewhat tentative and uncritical acceptance of a single set of data may be very misleading.

The constraint on monitoring development and health is not the inability to provide reliable and relevant indicators. This can be done with a degree of confidence but the key element is the appreciation of the part of the international donors and national policy makers of the importance of collecting and using relevant data to monitor conditions and plan programmes on conditions affecting human development and the will to use tools, techniques and materials that are already available.

Like most computer based information systems, CISFAM has a series of basic features: data capture, storage and retrieval, analyses, output and display. Data capture involves putting information into the computer and organising it in memory. The key element is how a soft ware handles each piece of information. Retrieval is the reversal of storage, recovering ordered data from the computer memory, mass storage disks or magnetic tapes for use. It also involves searching for information with certain characteristics, for example all hospitals less than 20 kilometeres from a paved road or with maternity wards. A special feature of CISFAM is its capacity to analyse and display information. This involves the retrieval of data files or parts of them in any combination and their analyses to generate tables, graphs, maps or charts. These display possibilities help planners manipulate and prioritise needs for resource allocation.

The project depends on two basic techniques of data use: overlays and statistics. In overlaying, the GRID software is particularly remarkable. For any defined area, if the information exists, data sets can be overlaid in an electronic version of stacking maps with different information for the same area. The greater the number of data sets, the greater the number of possible comparisons. For example, in a study of malaria control, the software will be able to compare water distribution, health centres, domestic animal distribution, and endemic malaria incidence by overlaying these sets one by one. In emergency health planning, this has special implications in pinpointing vector-breeding sites. Presently, both FAO and GRID use Landsat and the Advanced Very High Resolution Radiometer Data to pin-point locust and other vector breeding areas.

3.1 GRID Technology Applied to CISFAM Data: Image Processing Potential

One of the most important developments in computer technology has been emergence of Geographic Information Systems (GIS). Many types of GIS are currently in use. For CISFAM the most useful GIS system will integrate tabular (i.e., vital statistics) and thematic (i.e., ecological) maps with digital satellite data (i.e., vegetation) in a geographical context which will allow the analysis of physical, biological, economic, and social information by appropriate specialists and associations in era of epidemiological concern.

Plate 2: UNEP/GRID model for population density estimation

The above maps have been developed based on available health resource and population data reported to the various international organizations. They were generated as an experiment to test the model, pending acquisition of additional data. The iterations were made successively, as a function of constraints to produce topographical distributions of the variables of interest. The top photograph shows the population and major road network, based on the model explained below. The second is the possible health distribution as a function of the facilities available and the estimated population density.

Plate 3: Possible Health Distribution in Senegal


In the absence of detailed data, the development of a population density dataplane for Senegal has been accomplished at UNEP/GRID using a very rough estimation technique. The latitude and longitude of major transportation route and 118 towns in Senegal were digitized from 1:500 000 scale maps obtained by CRED personnel from the Institut Graphique National, Paris. Each of the towns was assigned a rank from one (village) to five (metropolitan area) based upon the relative size of each as depicted on the maps.

The Earth Resources Laboratory Application Software (ELAS) was utilized on the computer systems at UNEP/GRID to develop six raster or gridded cell (pixels) dataplanes generated at 30 seconds of latitude I longitude resolution (approximatively one square kilometre), one for the transportation network and one for each of the five ranked town sizes. Dataplanes for distances away from each town and road was then developed using ELAS overlay DIST. The final estimated population density dataplane was then generated by the summation of each of these distances indexed through tables with different weights for the roads and relative size of towns as contained in the following chart using ELAS overlay DBAS. The results are shown in the attached print where relative population densities have been color coded.



Understanding and controlling famines, mass migrations, abnormally high mortality, collapse of social and economical structures and epidemic diseases may be effectively accomplished when vital statistics in tabular form are linked to thematic forms of environmental and socio-economic parameters, using qualified population density distribution, allowing policy makers to take informed and productive decisions.


Two factors have a bearing on judgement of what statistics to collect: first, technology can almost certainly be counted on to cut down costs of producing and collecting figures (e.g. processing); and second, data compilation is only the start of a whole chain of processes, which include interpretation of the data collected, statistical analysis and research, which are very dependent on man, even if they make use of machines. Not withstanding these problems, every African country is trying to develop, within its limitations, the collection of better data. A further problem, however, is how to reconcile the needs for international comparability, and hence conformity with the standards laid down by the international agencies, with the specific domestic needs of an individual country.

The ELAS system software was developed mainly for processing of raster based satellite imagery, although it has routines for digitizing polygons and processing vector based data. Conversion between two data formats are also possible. The main routines of ELAS are as follows:

(i) Image analyses, histograms, scattergrams, other image area statistics;

(ii) Supervised and unsupervised image classification

(iii) Map transformations in 24 projections;

(iv) Geo-referencing of images and registration of maps and images using control points and non-linear distortion techniques;

(v) Logical and mathematical function of up to 20 different maps or images.

All of ELAS database is converted and stored in the latitude-longitude projection with a grid resolution of 30 seconds. This corresponds to pixel areas of 927 m × 927 m at the equator. At this resolution only one map in the African database contains 256 × 512 × 512 = approximately 67 million pixels.

Plate 4: Desertification Hazard in Senegal

Using data from CISFAM, some maps have been digitized, along with the communication networks. Health data are partially digitized and work is on-going in developing models and images with the data.

3.2.1. Computer Software Use and Development: An Overview of Low-cost Options

With the introduction of the personal computer, data processing equipment has become more affordable, easier to install and maintain, and no longer demands specialized skills for full utilization. These characteristics tend to facilitate the use of computer systems in developing countries, where information technology continues to be an essential tool for development. Increasingly, the personal computer is being applied to the planning and managing of development. The result is high-quality information for use in decision-making to help determine what projects are needed, to aid in implementing and monitoring them, and to evaluate their results. On a much wider scale than ever professional and technical personnel in developing countries have available the computing power needed to organize and analyze large quantities of data.

Appropriate software is the key, in fact, to the effectiveness of personal computers in the development process. Such software has been prepared by United Nations specialized agencies, multilateral and bilateral donor agencies, universities and foundations. In most instances, it is available free of charge from the organizations that produced it, on request of developing-country governmental agencies. In addition to appropriate software, decision-makers and users must have the personal computer hardware. With these two elements in place, expanding the use of information technology is a matter of connecting the information need, the software package and the personal computer. Because of the way software is designed, no package works on a personal computer without an operative system. One or more supporting packages may be required, depending on the application programme.

Besides the ELAS system, operational in CISFAM, there are two other possibilities that require lighter support mechanisms for the international CISFAM focal point. These are as follows:

1) The ARC/INFO is a system running on the following hardware:

. PRIME 750 CPU with 6 Mb memory
. 2 × 675 Mb fixed disks
. 1 CALCOMP 1044 Pen Plotter and 9100 Digitizer
. 1 Textronix 4109 colour display

The ARC/INFO digitizes and processes vector based cartographic information and has many routines for operations research applications (optimizing networks etc.). It supports raster-based data and can do overlays. Whereas ELAS can overlay twenty raster maps, ARC can deal with two at a time. The INFO routine can carry out database inquiries on attribute data, which has special implications for the type of information available in health and other population-based data.

2) A PC option for similar analyses and processing is the ERDAS system. This designed to run on IBM and IBM compatible personal computers. The support hardware required are:

. IBM PC AT with 85 Mb fixed disk and 1 floppy disk units
. Cartridge Tape Unit
. Colour Image Display, 512 × 512, 19”
. CALCOMP digitizer
. Colour inkjet plotter

The ERDAS system supports both vector and raster-based data and is very user-friendly. It has a well developed digitizing and editing routines and a large number of image-processing routines.

Table 3
Software Available for Low-cost Information Systems: Sources and Types






Processes farm survey data for farmers, agricultural experts and government officials

Familiarity with word processing, spreadsheet and/or databases helpful. Users must receive one month of initial training at FAO.

IBM PC (256 KB) with 2 × 360 KB diskette drives, IBM portable PC or PC XT; IBM monochrome display and Graphics Printer or equivalent DOS 2.10; PC-SORT


Produces computer models of agricultural development projects

Familiarity with word processing spreadsheet and/or databases essential. Ability to programme helpful

IBM PC (256 KB) with 2 × 360 KB diskette drives or PC XT; IBM monochrome display, IBM Graphics Printer or equivalent DOS 2.10


Projects population growth and change

No computer expertise required. Comes with an user’s manual giving step-by-step instructions on using the programme and a sample database on diskette suitable for training.

IBM PC (256 KB) with 2 × 360 KB diskette drives or PC XT; IBM monochrome display; IBM Graphics Printer; or equivalent. DOS 2.10


Processes data from census questionnaires

Familiarity with word processing, spreadsheet and/or databases essential. Ability to programme helpful.

IBM PC XT with expansion unit; IBM Monochrome display; IBM Graphics Printer or equivalent DOS 2.10; COBOL Compiler, ASCII text editor


Analyzes economic and demographic data to estimate economic growth.

Familiarity with word processing spreadsheet and/or databases essential. Ability to programme helpful

IBM PC (256 KB) with 2 × 360 KB diskette drives or PC XT; IBM Monochrome display; IBM Graphics Printer, or equivalent. DOS 2.10

3.2.2. Hard and Software Capabilities and Needs: CRED

The current useable hardware at the Centre for Research of Disaster Epidemiology is as follows:

1) PRIME 2250 (300 Mb capability with 150 Mb dedicated to the CISFAM Project);
2) TEXTRONIX 4691 Pen Plotter;
3) TEXTRONIX 4109 Smart Terminal

The short term hardware requirements for CISFAM operations are:

1) Image Processing Display System (IPDS) 512 × 512

. One Image Plane
. Four Graphic Plane

2) Hardcopy Report Quality Printout

. Matrix camera
. Dunn camera

Both cameras use signals of blue, red, green on IPDS to produce:

. A4 size Polaroid print and transparencies
. 35 mm colour slides and prints.

The long-term requirements are:

1) Ancillary Data Digitization
2) Remotely Sensed Data Processing Capability

The short-term hardware acquisition options are the following:

1) Upgrade PRIME 2250

. Gould/DeANZA equipment with CONTROLLER, REFRESH, CRT
. ERDAS IBM-PC complete system with add-on digitizer.


2) DIAD Systems Ltd. Complete system for PC Image Processor.

The procurement and installation of the hard and software required to make CISFAM operational at CRED, Brussels, in terms of using different types of data and image-processing, were explored. In view of the time and budget constraints vis-a-vis the CRED staff training, administrative and technical processes involved, this was undertaken via the co-operation of UNEP/GRID Project.


Estimation of Population Density, the UNEP/GRID model applied to Senegal integrating major cities and transportation routes.