Though Zimbabwe has few traditions as a “fish eating” nation, with the opening of Lake Kariba over three decades ago a fisheries industry was started. It has recently been realised that, in order to boost this industry, and to keep pace with the rising demands for fish, the many small dams in the country (over 12 500), which are used presently mostly for livestock and irrigation, could usefully be exploited. As well as food protein, jobs and alternative income sources would be created, especially in the rural communal areas.
The Min. of Natural Resources and Tourism (MNR) has overall control of the larger water bodies and fisheries in Zimbabwe, and the Dept. of Agriculture, Technical and Extension Services (AGRITEX) oversees activities involving use of the small dams. It operates a small Fisheries Unit (FU) who have responsibility for fostering fisheries development. As a possible practical measure which could contribute to better management of this development, it has been suggested that a Geographical Information System (GIS) be installed. Apart from working out the most appropriate “level” for the installation of the GIS, the terms of reference (TOR) for this consultancy also sought to formulate proper objectives for a FU - GIS; to estimate the availability and quality of data obtainable; to establish the time necessary to collect and compile this; to work out whether there were advantages in adding nutritional data to a FU - GIS and to forecast the results which could be obtained from a GIS.
A total of 15 specific uses for a GIS were formulated and these would basically help to:-
In order to make the GIS functional, it was most important to establish what data was needed and a list was drawn up. Clearly, the data could be in several forms, i.e. mainly maps, digital or tabular data. The availability and quality of the data was checked. About half of the mapping data found was not only of a good quality but very useful and easily obtainable. However, digital data was, for the most part, unobtainable except for a few map outlines, some remote sensing tapes and images plus a potentially valuable DAM DATABASE which was in the process of being made up by the Min. of Energy, Water Resources and Development (MEWRD). Tabular data, as well as its lack of disaggregation, was characterised by its “extremely fragmentary nature”, and hence the quality for a GIS was very poor.
Several factors, both external and internal, were identified as being reasons why it was extremely difficult to estimate the time and costs to collect, compile and input the data to a GIS and to do the analyses and reporting. Generally it would appear that progress would be very slow and this was attested to by the large number of planned GIS's in Harare which had not yet been made operational. From the FU - GIS viewpoint, the main delays are likely to be caused by getting access to a digitiser, from delays in getting the DAM DATABASE completed and operational on a PC system (with a sufficient number of fields) and the paucity of any comprehensive tabular data, at least until the results of the 1992 census are published. Certainly we cannot envisage any output from a fisheries GIS for at least six months.
Regarding the collaboration with a nutritionist to assess the advantages of adding nutritional-related data to the FU - GIS. After careful consideration no real advantages could be found, i.e. since it would appear much more logical, in a country which does not particularly suffer from protein deficiency problems per se, to add nutritional data to a more general “agricultural” GIS, which could be sited in AGRITEX and administered by a reformulated and streamlined “National Nutrition Co-ordinating Committee”.
Given the relatively poor standard and availability of data for a FU - GIS, and given likely delays in both digitising and in the implementation of an upgraded AGRITEX GIS, then the results from any FU - GIS would be not only slow to materialise but they would need to be treated with some caution. Indeed, some of the data sources contain information which is either so generalised, or so dated, that it is inevitable that a serious degree of error propagation would occur. This runs the danger of getting GIS's a bad name when in fact it might be a poor choice of inputs which is to blame. Despite this “gloomy scenario”, it is considered that the longer term outlook for the reasonable functioning of a fisheries GIS are sufficiently promising to recommend it, and that a start should begin soon in implementation in order that a potentially protracted learning curve and spatial data handling assimilation can commence.
Finally, after a considered review of the various options for implementing a fisheries GIS, it was recommended that FU obtain their own system, which they would run with a large degree of collaboration with a future AGRITEX GIS. This was seen as sensible, largely in view of their undoubted need, their enthusiasm and determination to succeed and the importance of physically having a system where it is needed. It was also seen as essential to work closely with AGRITEX in the overall adoption and implementation decisions.
Since full details of the project are provided in the original project document for ZIM/88/021, then only brief details are given here plus added details on the consultancy.
Zimbabwe, being a land-locked country having few natural-lakes, has never been a nation of “fish eaters”. However, with the opening of Lake Kariba in 1959, the opportunity to establish a major fishery was taken. Fish production has expanded significantly so that it now stands at some 23 000 tonnes p.a. This is still insufficient to meet national demands, which are expected to reach more than 40 000 tonnes p.a. by the year 2 000. Even then per capita consumption of fish will be low. Consumption itself shows a geographic variability with the rural areas having very poor access to fish supplies vis a vis urban concentrations. Current aquaculture output is probably less than 1 000 tonnes p.a. This is produced mostly by larger scale commercial farmers around Lake Kariba, by some trout farmers in the Eastern Highlands or by farmers in rural areas who have constructed small ponds. In the latter productivity is low because of lack of access to inputs or to extension advice. Kent and Josupeit (1989) give some further background information on fisheries and its structure in Zimbabwe.
The government of Zimbabwe has a policy to achieve food self-sufficiency and to raise agricultural output generally. This policy is particularly directed to the communal areas, not only to provide food security but also to provide employment and to generally raise the standard of living. Small scale aquaculture aimed at increasing fish yields by purposeful management, could make an important contribution towards achieving these aims by:
The institutional framework covering fisheries production in Zimbabwe is basically under the auspices of two government departments. The Ministry of Natural Resources and Tourism (Dept. of National Parks and Wildlife) has the major responsibility for all water bodies and fish stocks in Zimbabwe. They maintain direct control of the major lakes and recreational fisheries, and they support some aquacultural activities. The Ministry of Lands, Agriculture and Rural Settlement (Dept. of Agriculture, Technical and Extension Services - AGRITEX) has practical control over the 12 500 small dams and the fish ponds, i.e. since they have a well developed extension service in the field who are able to promote a range of agricultural activities, many of which are linked directly to the small dams (since they primarilly constructed for water supply needs). Within AGRITEX there is a small Fisheries Unit who are concerned with fostering fisheries development, primarily in those dams situated in the communal areas. The FU itself is housed in a small self-contained compound about 1.5 kms from the main AGRITEX complex. Appendices 1 and 2 show the personnel employed and the main goals of the FU respectively. Present computing equipment includes two Apple/MACINTOSH SE PC's plus an IMAGEWRITER 11 printer.
The project (ZIM/88/021) arose from the government's recognition that aquaculture production was failing to realise its potential, i.e. due mainly to there being a lack of trained personnel, inadequate supplies of fish seed and a lack of facilities for demonstrating and fostering fish farming techniques under Zimbabwe conditions. (There are only six government stations either rearing fish or doing research.) The government recognised the need for specialised intervention, particularly in the form of technical and fisheries infrastructure, so the UNDP/FAO were judged as appropriate agencies to provide assistance. It is intended that by the end of the project (end of 1992) :-
The project document itself provides detailed strategies on how these objectives will be achieved. It is important to note that the project will-not operate in isolation, i.e. it will maintain contacts with the following FAO projects:
The consultancy has arisen with the perceived need by the Chief Technical Adviser to the project, of implementing a form of management/information control. At the project's initiation, the useful data that existed in the FU tended to be of an “extremely fragmentary” nature. Since 1989 an effort has been made to implement more sophisticated management techniques such that many routine procedures have now been computerised. However, during the past year it has become apparent that the potential scale of the whole project is vast, principally from the point of view of there being so many many dams spread over an enormous geographical area. The rapidly emerging field of Geographical Information Systems was seen as a potential way of coping with the likely information demands of the project. Briefly, GIS is a means whereby spatial information can be stored and manipulated, in a great variety of ways, by the use of powerful computer software packeges allied to the appropriate hardware systems. Data must therefore be input in digital form, using if desired a variety of methods, and output from the system can be in the form of a screen image, hardcopy written to printers, plotters or to film or data stored on tapes or discs. The implementation of GIS requires that a number of decisions be made - hence this consultancy.
The original TOR's were dated 29th May 1991 and contained six specific areas. However, on 15th July, 1991 a further area was added, in a non-structured form, and this was interpreted as per point 1 below. This interpretation may appear different from the intention of the originalpoint but I think that point 6 will also convey the essence of what was required. Since this additional point would logically take precedence over the others it will be listed first.
| Point 1. | Establish the objectives for implementing a GIS, and to convey further possible uses of such a system. |
| Point 2. | Working closely with staff of the Fisheries Unit of AGRITEX, determine the availability and quality of the data on spatial factors that control the potential of fisheries in small water bodies and the development of fish farming in ponds in Zimbabwe. |
| Point 3. | Estimate time and costs to collect, compile and input these data to a GIS and to do the analyses and reporting. Check to see what other projects or entities in AGRITEX could use the same data and investigate the possibilities for collaboration. |
| Point 4. | Look into collaboration with the consultant on “Strengthening the Use of Fish and Fisheries in the Alleviation of Undernutrition” on mutual data needs, collection and analyses. Assess the advantages of adding the nutritional-related data to the GIS. |
| Point 5. | Forecast the results that could be obtained from the GIS analyses including their reliability given the quality of the data available. |
| Point 6. | Evaluate alternatives for GIS implementation including use of FAO Central GIS, contracting for the GIS with a commercial firm, integration with an already existing GIS in Zimbabwe (e.g. FAO Forestry Dept. project) and a PC-based system in the AGRITEX Fisheries Unit itself that could be compatible with the proposed Departmental GIS using ARC/INFO. |
| Point 7. | Write a brief report of your findings and recommendations as an AGRITEX Fisheries Unit Technical Report. |
The report will be structured such that the subject matter of each Point will form a seperate major section.
Before leaving FAO in Rome a briefing was given on the background to the consultancy and on consultancy expectations. It was originally planned that work in Harare should be carried out in tandem with a nutrition consultant working on project GCP/INT/467/NOR. However, because of delays in getting requisite clearance to proceed with the consultancy, this did not happen. This meant that some of Point 4 requirements were met by means of preliminary discussions in Rome with the nutritionist before departure for Harare.
The means of achieving the aims of the other Points was via three main strategies, i.e. meetings with relevant personnel, reading appropriate reports and by “interacting” with the FU staff. Obviously a good deal of day to day coordination was necessary, as was preparation for most meetings. The main people or groups consulted are listed in Appendix 3 and the main reports studied are shown in Appendix 4. The Chief Technical Adviser (FU) and/or one of the Specialist officers (who had some background in GIS) also attended many of the meetings so that they could be conversant with progress and as an aid to this consultant being made aware of any “local factors”. Regular Friday afternoon meetings were held under the chairmanship of the Deputy Director (Technical) of AGRITEX at which progress was discussed.
Upon arrival at the FU of AGRITEX exploratory discussions were held with the Chief Technical Adviser (CTA) and the Senior Fisheries Officers. This was clearly a two-way process - the consultant finding out some of the background to fisheries production and organisation in Zimbabwe, and the FU being introduced (or more comprehensively explained) to some of the merits (or otherwise), and the potential, of GIS as it might apply in the FU situation. Mid-way through the consultancy it was possible to formulate a fairly structured list of possible ways in which a GIS could be utilized by FU. This is as follows:
In summary, a GIS would help to:
In the wider context, as long as systems compatibility was adequate, then it would be hoped that FU data/output could be matched up or shared with data held by other groups or departments, i.e. with regard to such functions as nutritional surveillance, land use planning, farm output, employment needs, water use zonation, etc.
Given the time available for the consultancy relative to the potential breadth of this Point, it is only possible to gain a generalised, though hopefully perceptive, view of the data situation in Harare. Initially, it was important to establish what spatially related data would be required. Clearly, for data to be of use to a GIS it had to be suitable for conversion to digital format, i.e. data lacking geo-referencing or maps lacking projection type would be unsuitable. Table 1 represents the main factors relative to fish production for which spatial data may be required. Clearly the list would not be exhaustive.
| TABLE 1 - Spatially Related Data Required for Use by FU - GIS |
| Soil types River location or drainage patterns Relief or slope Water temperatures Water quality Population distribution or density Transport routes Patterns of land holdings Agro-ecological zones or farming types Administrative boundaries Dam site locations Market locations Agglomeration locations Availability of fish feeds, manures, seed, etc. Location of extension workers Prevalence of predators (crocodiles and hippo's) Prohibited areas for pond culture or dam use Economic indicators of wealth Nutritional indicators |
Data on any of the parameters listed may be available in different forms. The main forms, upon which the rest of this section will be subdivided, are classified as - a) maps and photographic data, b) existing digital data and c) textual or tabular data.
Appendix 5 summarises the main relevant maps which are readily available and it also comments on their quality. All of the 1:1 000 000, 1:250 000 and 1:50 000 DSG sheets are currently held by the FU. The cost of these maps is typically low and the Dept. of Surveyor-General (DSG) catalogue (with updates) gives further details of these and other maps. The catalogue also details the aerial photograph collection. Most of the country has been surveyed by air since 1986, and photos are available at 1:25 000. There is earlier photography at scales varying from 1:40 000 to 1:80 000 for over half of Zimbabwe. Since it was impossible to compile a list of other potentially useful maps, of which undoubtedly there would be many, contacts with any of the following could be useful:
Further enquiries could also be made via the Secretary - National LIS/GIS Committee; the Chief Topographer - AGRITEX or senior personnel at DSG. Additionally, the Central Statistical Office (CSO) produces some useful maps in its “Zimbabwe in Maps - A Census Atlas” (1982)
The quality of the mapping listed in Appendix 5 can be divided into three approximate categories:-
Where positional or linear data is required, e.g. roads, rail, rivers, contours and major administrative boundaries, then it is very good, i.e. since most of it is derived from aerial surveys and thus features can be accurately located.
Where data being mapped shows relatively permanent boundaries, then the quality should be quite good, though boundary changes will inevitably have occurred, e.g. with regard to resettlement lands, and some of the district boundaries have recently been changed. Administrative boundary changes cause particular problems for GIS because of attribute allocation in any temporal analyses.
Where data is showing either information which can be highly variable at the micro-scale, e.g. soil boundaries or agro-ecological zones, or where maps plot numerical data by various classes, e.g. population density, then the quality must be treated with extreme caution. Indeed, where possible it would be adviseable to seek more detailed information. It is also possible to prove that different editions of one map give totally different information, e.g. as an examination of the three latest editions of the “Agro-Ecological Zones” map shows.
Two further qualitative points to remember are that:-
At a scale of 1:1 000 000 any mapped surface will be generalised, and grid references on these particular DSG maps are not plotted.
Some of the data is based upon surveys which are more than ten years old.
Appendix 6 summarises the availability and quality for some of the other parameters which could be useful for the FU - GIS, but for which data could not be directly inferred from easily available maps, i.e. “proxy” data would need to be derived. It can be readily seen that for most of these parameters careful consideration would need to be given as to how the data was going to be used or plotted. Additionally, some of these factors might need to be treated differently at different scales, e.g. availability of manure would be a micro-scale consideration whereas water temperature variations might only be noted at the macro-scale.
It would appear that there is no map evidence to give clues as to prevalence of predation or to any economic indicators. It is possible that areas which would be banned for fisheries production, e.g. National Parks, could be easily plotted but it would first be necessary to compile a list of areas prohibited for fisheries development. The Early Warning Unit for Food Security (within AGRITEX) is able to provide some small scale mapping which might give a clue to food production potential for various crops (see Figure 1), i.e. as an indicator to nutrition, but this data would need to be treated very cautiously since it is highly specific in what it plots, whilst it necessarily is highly generalised in its plotting.
Though there are now several sourcesof readily digitised data in Zimbabwe, it is unlikely that much of it would be of potential use to the FU, e.g. the Dept. of Meteorological Services collect data from geo-stationary satellites. However, there are certainly three potentially useful sources:-
Few fields on the database are currently “comprehensive” and the data compiled needs sample verifying by field research. It is anticipated that a large number of new fields could be added, e.g. relating to limnology, biology, fisheries-exploitation and fisheries-extension (Appendix 10 gives a more detailed list). Regular updates of the database will be passed on to AGRITEX by MEWRD (perhaps every three months) and the presently missing data should “gradually accrue”. The personnel compiling the DAM DATABASE at MEWRD “could not vouch for its accuracy” and also said “it might not be too reliable”.
Although there are many government departments, including especially the CSO, who are producing vast quantities of statistical tabular data, there is very little which is of use to a FU - GIS. Most data in tabular form suffers from a lack of disaggregation, and where it is disaggregated it is usually only done to Provincial level (there are 8 provinces in Zimbabwe). There are simply not the data collection structures necessary throughout most of rural Zimbabwe, and thus much of the data relies either on sample surveys or it is data for which collection procedures are relatively simple, e.g. mineral production, trade figures, financial data, etc. Indeed the consultancy found no readily accessible data, at lower than provincial level, which was of use to FU. Having said this, it is likely that a more thorough investigation would reveal numerous small “snippets” of useful data, i.e. especially from within AGRITEX and certainly some data would be forthcoming from the FU (Appendix 11 gives some examples of FU data which could be included on the DAM DATABASE, and data for a separate “pond database”). To some extent this paucity of data will be ameliorated when the results of the 1992 census are made available, i.e. this is planned to have data disaggregated, where possible, to village level -or at the “worst” to ward level.
There are a number of external factors relating to this consultancy which collectively make questions concerning time and costs to collect, compile and input the data to a GIS, plus doing analyses and reporting, extremely difficult to answer. Briefly these are:-
However, making certain presumptions which will be indicated, tentative estimations on GIS implementation time and costs can be arrived at. Again it will be convenient to examine this section by type of data.
Here the concern is largely with map collection and digitising considerations. The collection of all main paper DSG maps is already complete and has been paid for. Some of the requisite maps have been digitised by the DSG (see Section 5.3) and these could be obtained free of charge with about 48 hours notice. The NRSF have other, though unlisted, digitised outlines (see Section 5.3.) which should be available quickly and cheaply. However, additional digitising, which could be needed for perhaps 50% of required maps, is a problem. AGRITEX do have a digitiser but they do not know whether it is functioning and they do not employ digitising staff. The submitted AGRITEX draft proposals (ZIM/91/005) do include “a capacity to prepare and produce digitised base maps” (p.24), but when this proposal would be fulfilled is difficult to predict (see also Section 9.4). When the equipment did arrive it is still difficult to predict digitising times because the amount of access to the machine would be critical. Given the small amount of longer term work it would not be viable for FU to purchase their own digitising equipment, especially since present FU personnel could not give time to this task. Few other government owned digitisers have been located and those that do exist are intensively used, e.g. the DSG could not guarantee any work being completed within a year, though by then a new donor sponsored (CIDA) system may speed things up. There is at least one private company (Systron) who can do commercial digitising which might be worth investigating. No companies or government departments have automated scanning facilities.
FU could acquire some of the map outlines from FAO's copy of World DataBase 2, though these would require expensive (US$1 000) projection changes and their detail is very poor for the FU type work. A further possibility would be to arrange, through FAO Central GIS, to get the 1:1 000 000 DSG sheets scanned, i.e. if the maps were deemed suitable or if the colour plate masters showing the individual“layers” were used - these can be obtained on film from the DSG. Unless the AGRITEX digitiser can be fixed and a person be employed to use it, then there is little hope of getting requisite digitisation completed, at an affordable price, in less than say 6 months.
As previously mentioned, the DAM DATABASE is presently being converted so that it can run on a PC system. Here again there are major temporal problems. The conversion process has been proceeding now for more than two years (this process is described in detail in Wood-Sichra, May 1991, Appendix L1–2), and it is not being done by AGRITEX so they have little control on its progress. As of August 1st 1991, only 55% of the 12 500 entries had been transferred, and it was inferred that it would be at least six months before all entries even had a geo-reference. This too is creating problems. The geo-referencing entry has been given as a six figure grid reference, proceeded by the sheet letter index code. The map sheet number have been entered as a separate field. In the case of Zimbabwe this data is insufficient for national mapping purposes since any six figure G.R. is only unique to either the Western or Eastern half of the country and the digitising software would not cope with the sheet letter index code without a programme being written. Even when this had been done there are only a few fields in the database which have any entries, and these fields are only up to 80% complete, and quite possibly inaccurate. Unless it was the subject of a special project, it would take years to complete the database with the desired number of fields. It is recommended that the database be eventually transferred from “KNOWLEDGEMAN” to a more widely used and easily supported database. It is doubtful whether it would be presently possible to speed up the conversion process because of the specialised nature of the work and the lack of access to alternative computing equipment.
If digital-Landsat data tapes were required fron NRSF then these could be obtained quickly and, for the present, at no cost. However, only a limited temporal coverage is available, though they are presently acquiring about seven new Zimbabwe scenes per year from South Africa.
Preparation of tabular data could theoretically begin almost immediately since AGRITEX does have database storage packages, and it is likely that access to a computer for this work could be found. Indeed FU could obtain suitable software for use on its Apple/Macintosh's, since conversion to IBM PC is no problem. However, as has been shown (see Section 4.4) there is really very little data which is worth storing at present. The collection, compilation and input of useful data, e.g. broadly on say pond size and location, stocking inputs and yields, or indeed on any other useful field, would be an extremely slow procedure given the extent of the task and the personnel limitations, i.e. years for completion rather than a year. Obviously a start could be made in specific districts, and this should be planned for. Estimates on time and costs to collect, compile and input other tabular data for a GIS, e.g. nutrition or socio-economic, would simply be a “guestimate” because of the diffuse nature of the problem.
Once data had been collected, compiled and input to the GIS, then, providing that the GIS operative was familiar with the software, any analysis and reporting should be executed relatively quickly, i.e. certainly in relation to data compliation times. Depending on the types of analyses required, this should be a matter of weeks and times should rapidly reduce with experience and once procedures had been verified. For these tasks it is anticipated that FU would-have complete control over these procedures and would use their GIS trained specialist.
There is little doubt that some of the basic 1:1 000 000 digitised map outline data could be successfully shared though this would be a one-off procedure. Since this is the case then it would be pointless for both FU and other AGRITEX entities to each digitise their own set of boundaries - they should get together to agree outlines required and standards. Collaboration with AGRITEX would be likely to speed up the process of achieving these outlines. The DAM DATABASE would also be useful to AGRITEX, especially their irrigation section, since a primary purpose of the dams is to provide water for irrigation and for livestock. It is anticipated that AGRITEX could get their own copy of the DAM DATABASE, i.e. separate from the FU copy. Collaboration with other entities in AGRITEX would bring very few benefits at present, though it will be recommended that before the ZIM/91/005 proposals are fulfilled, FU should get together with the other entities to discuss possibilities.
As mentioned in Section 3, it was not possible to be in Zimbabwe at the same time as the nutrition consultant. Our discussions in Rome provided an indication of the nature of the data required, as did a reading of the draft document “Strengthening the Use of Fish and Fisheries in the Alleviation of Undernutrition” (GCP/INT/467/NOR).
What was pursued with regard to this Point was initially to contact as many people/departments as possible in order to ascertain the availability of the many types of data which may be relevant to GCP/INT/467/NOR (and to ZIM/88/021). However, as the nutritionist found, and as was found with regard to tabular data collection for fisheries, the quality of data from the GIS input viewpoint, was of very limited use. The expected problems of disaggregation, fragmentation, inaccessability, sample data only, lack of collection procedures, unstructured data handling systems, insufficient personnel or capital to manage/run computing systems and disorganised management structures were noted as the primary causes. Down to provincial level there was certainly a fair amount of data, but no recent, comprehensive, post 1982 tabular data on factors relating to food supply, incomes, health or nutrition at sub-provincial level was discovered. To be fair, there is certainly some data for some regions, e.g. the Chief Nutritionist (MOH) pointed out that, with written permission from the Secretary of Health, then sub-provincial data on “Mother-Child Health” would be obtainable, and the Early Warning Unit in AGRITEX had quite a lot of 10% sample survey crop yield data relating to particular sampled communal areas. There seems to be a problem that the efforts needed to locate and obtain some data may well outweigh any advantages gained from its acquisition!
I next sought to assess the extent to which we (GCP/INT/467/NOR and ZIM/88/021) required similar data. There are a number of data fields of mutual interest. We would both be interested in the location of communal areas, in any economic-income indicators, in locating areas having actual aquaculture potential or protein supply needs, in a measure of population density or distribution, in patterns of land holdings, in agroecological zones of farming settlement types, in administrative boundaries, in market locations, in locations of extension workers, and in other nutritional indicators. It would certainly be FU's goal to incorporate these parameters into their GIS.
In many senses “aquaculture” and “nutrition” in Zimbabwe are not obvious “partners”. There would not appear to be any relative lack of protein, except perhaps amongst the poor. This would be the case anywhere and where this occurred there would also be a more general protein-energy deficiency situation. Additionally, as the Chief Nutritionist (MOH) pointed out, “over the vast majority of Zimbabwe there is no fish eating tradition and most of the rural-poor would not initially know what to do with a fish if they were given one!” This might be an argument which negates the work of ZIM/88/021, though fish production is hoping to be started as a means of exploiting further the rural dams and to create jobs and an additional income source, i.e. as well as a food source. Having said this, it would appear to be very important that nutritional considerations are related to a GIS, and this was agreed by those relevant people who were contacted.
What needs to happen, as again agreed by the nutrition people contacted, is that there should be a streamlining of the presently fragmented organisation dealing with nutrition, i.e. it is possible to presently identify at least four organisations involved. A clearly defined, single organisation needs to evolve having links to at least the MOH, AGRITEX, the Meteorological Services and perhaps the CSO. The organisation might best be housed with AGRITEX (again mutually agreed), since they currently have most involvement, i.e. with the Early Warning Unit and the National Steering Committee on Food and Nutrition, and since they have a huge extension service who might be available for data collection or monitoring purposes. The new organisation should be entrusted with the designing its own geo-referenced data collection system. Data collected could then be stored and available for input to the AGRITEX GIS, i.e. which should evolve from draft project proposal ZIM/91/005 when implemented. It might be necessary to concentrate the limited resources presently available on a few sample areas, which clearly should include areas considered to presently be at risk nutritionally. This would allow models to be created showing how best nutritional data could be integrated to other databases so as to provide adequate nutrition surveillance. Though this may sound idealistic, little headway will be made regarding nutrition surveillance until adequate organisational and data storage and processing facilities are in place.
Finally, it can be inferred that very few real advantages could be seen to adding nutritional related data to a FU -GIS, but there were many advantages to adding it to more generalised AGRITEX GIS databases.
Given the somewhat pessimistic findings in the previous Points, then the subject matter of this Point can hardly hope to fare any better, i.e. given that the quality, certainly of much of the socio-economic data available, is generally poor and/or that access to the data is difficult. However, in the recommendations some directions towards obtaining positive rewards will be provided.
In Section 4 a detailed list showing possible ways in which a GIS could be used by FU was proposed. Clearly, given optimum conditions, then the fulfilment of these objectives would be what was required of the GIS, i.e. these are the results which could be obtained fron the FU - GIS analyses. A perusal of this list will show that virtually all the objectives rely on the DAM DATABASE and/or the digitised outline maps. Though the lack of digitising facilities would not directly affect the “quality of the data” (as stated in Point 5), without many of the digitised outlines then there simply cannot be many “results” (as stated in Point 5). And when the DAM DATABASE is completed in perhaps six months time, having only a complete set of locational references, then this can only provide a list showing dam locations plus the numbers of dams in administrative districts - facts which can be inferred anyway from perusal of the 1:250 000 or 1:50 000 DSG maps, or which could be handled by existing AGRITEX information systems databases. If certain digitised map outlines can be obtained from either the DSG or the NRSF, then some limited analyses will be possible either using merge or overlay techniques to create new “layers” or to form the bases of correlation and other statistical analyses. When the location of dams are added, then some movement can be made towards getting some answers to more serious management problems or plans, bearing in mind, of course, the suspect nature of some of the DAM DATABASE.
To achieve faster and more qualitative results, it might well be worth while concentrating initially on a small-geographic area. Thus for instance the DSG or the NRSF could have perhaps digitised outlines showing various parameters relating to one province or some other bounded region. (Though the FAO Central GIS in Rome might arrange to do the required digitising, for a trial area, the costs of this are likely to be beyond the project's budget). An attempt could then be made to complete several fields on the DAM DATABASE for that particular region, and to verify the data entries where possible. Using these two linked databases then at least a good deal of experimental work could be done towards creating useful analytical models.
Finally, any results must be viewed in the context of the fact that many maps at a scale as small as 1:1 000 000 are extremely generalised, and they may also be using very dated information. These are only two of the main possible sources of error in GIS's (see Burrough, 1986, for a comprehensive list - the fact of error propagation when utilising GIS techniques is too often ignored and Burrough is well worth reading on this subject). It is feared that those potential results relying on dated tabular or mapped data, or on data using generalised classes to categorise a distribution, or on using “proxy” data, would need to be treated with caution. If models or decisions start to be formulated on shaky base material, then the GIS is in danger of being perceived as “useless” and its inherent value may never materialise.
This section will be interpreted in terms of the level of the implementation rather than systems implementation per se, i.e. a PC based system will be presumed and little discussion will be made as to varieties of same. This is largely because AGRITEX has yet to reach a decision on a system, though it is believed that they are considering both ILWIS and ARC/INFO. Realistically, five major alternatives for a fisheries GIS implementation may be considered:-
Clearly, if a complete systems view of GIS is taken then it might be possible to have limited combinations of these, e.g. the digitising or plotting could be done separately. There could also be short term implementation alternatives which differed from longer term ones, and finally it might be adviseable to postpone implementation indefinitely. Some of these points will be considered as each of the major alternatives is reviewed.
In the sense of undertaking all of the work, this alternative can be immediately discounted. The main reason for concluding this is that any GIS which FU might set up would have to be done as a long term continuous process whereby, for the most part, small amounts of the data would gradually “accrue” to the system in the process of database building. The amount of interchange of necessary communications between Harare and Rome would be prohibitive. It is also desireable that local personnel who are both familiar with the intricacies of the FU, or with local conditions generally, should be available and that they should have the opportunity to learn with the process of GIS implementation. FAO Central GIS is more suited to “one-off” projects where adequate data has been gathered and where it is not worth while for a project to invest in its own GIS. This is certainly not the scenario in FU where the demands are more for long term planning and management or for shorter term continual reference.
Having said this, there may certainly be a case for asking FAO Central GIS in Rome to give assistance with digitising or scanning. Section 8 made it clear that this could be a major bottleneck in Harare and it will be recommended that the FU investigates the possibilities of contacting FAO Central GIS for help if more convenient arrangements cannot be made.
This option too can be immediately negated. If the commercial firm were to be overseas then exactly the same reasons as listed in Section 9.1. would apply and it presumably might be much more expensive. Certainly, with the amount of work which would need completing, over a long time period, then the costs would be extremely high - certainly beyond the budget of the project. This option for implementing the GIS with a local commercial firm is out of the question since no local firms exist who could do the work, or indeed any major GIS work.
Again this option can be immediately discounted. After extensive enquiries amongst most of the people “in the know” about GIS, it was only possible to definitely confirm four operational GIS's, though there may be others. One of these, at the DSG, was really only doing CAD type work and it's outdated equipment was already fully utilised for several years ahead. Pooley (1991) describes this in more detail. The second was at the NRSF and they too had quite enough of their own work. The third was at a Dept. of National Parks and Wildlife site in Hwange. Here the very small GIS was also fully utilised and in any case inaccessible (and perhaps not yet operational). And the fourth was a “contrived” GIS in the Early Warning Unit of AGRITEX which operated by using a combination of image analysis, graphics and spreadsheet packages. Other GIS operations, for one reason or another, could not manage to get their own systems working and they were certainly not in any position to take on work for others. Under this section heading it should be mentioned that there is to hand a draft “Project Management Plan”, dated April, 1991, which outlines proposals for a CAN$4 million worth of GIS to be supplied by a Canadian agency (CIDA) to the Dept. of Natural Resources (DNR) in the Ministry of Environment and Tourism (about CAN$250 000 of this would be for upgrading the DSG - GIS facilities). This is to be used mainly for environmental monitoring. Since the DNR has overall control over water and fish in Zimbabwe, a case could be made for siting a FU-GIS here. Although I was informed that this project would commence in September, this was treated with some scepticism elsewhere, it being questionnable whether DNR would have access to sufficient data to maintain such a GIS. Additionally, a “user needs” study has yet to be accomplished ! It would certainly be adviseable for FU to keep track on this development in case any integration possibilities arose.
There is a GIS in AGRITEX which is presently producing work, mainly for the Land Use Planning section. It is basically an older ERDAS image processing system linked to an ink-jet colour plotter, and the system was designed principally for Remote Sensing work. Since AGRITEX have still to get their digitiser functioning, and since they have no digitising personnel, then there appears to be extremely limited possibilities for integration here at present. Until now it would be fair to say that GIS was very much in its infancy at AGRITEX - there is certainly little coordination between sections. However, AGRITEX do have plans to introduce a new upgraded GIS, as revealed in draft project document ZIM/91/005 (June, 1991) “Programme of Technical Assistance to the Department of Agriculture, Technical and Extension Services”. The document does not spell out in detail the GIS component, i.e. actual hardware or software types, or whether it would be at work-station or PC level, but inputs would include training provision, full time GIS staffing till 1994 and six months worth of GIS specific consultancy. US$30 000 has been budgeted for GIS software. Although “computers” are mentioned in the equipment budget, digitisers and plotters are not.
Clearly, quite a strong case could be put for locating a FU - GIS with the intended AGRITEX GIS purchase (though not with the existing set-up). The system would be new, there would be a GIS specialist on hand for assistance and the systems equipment itself could be mutually shared. Also, planning has been done from the top and certainly some enthusiasm is there (if not the time) to help with getting the system operational. However, problems could be foreseen in getting sufficient access, since the budget would not appear to be aimed at a networked system, and I foresee substancial demands from the various divisions within AGRITEX, i.e. certainly from crops, irrigation, livestock, planning and social economics. The 1.5 kms distance barrier between FU and AGRITEX main office would be a not inconsequential problem. At the present time we have no real idea as to when the system might be functional. This cannot be envisaged before 9 months to a year, and indeed the GIS portion might not be functional for 18 months.
This would certainly be the preferred option. The reasons for this are that the senior staff are extremely keen for this to happen; they do have a potentially useful resource in the DAM DATABASE; they can envisage, and indeed have the need for, a wide variety of systems applications; it could probably be installed in advance of AGRITEX's GIS; it is always preferable to actually have the system where it is needed and FU ought to be the “masters of their own destiny”. Having said this, some caveats do have to be stated. It is considered almost essential that FU and AGRITEX run the same; or very compatible, systems - indeed AGRITEX shares this agreement (see draft ZIM/91/005 - page 31). This is for the obvious reasons, i.e. centred on compatability and mutual assistance, plus the fact that at the end of the project (ZIM/88/021) there could be concern about the FU - GIS sustainability, and systems compatability could ensure it.
Given that AGRITEX are unlikely to make an early GIS systems decision, and the extreme paucity of suitable FU - GIS based data, there can be no sense in making a hasty adoption decision. The time before adoption can be usefully employed in other necessary GIS related preparations (see Section 11). The cost of equipment will also be a factor limiting the rate of GIS adoption, i.e. since it may need to be spread over several budgetary periods (budgetting guidelines can be obtained from Wood-Sichra, May 1991 - see Appendix 12). It would certainly be envisaged that the computer and software be purchased quite soon, plus a plotter/printer when it was needed, in say 9 to 12 months time. It could be possible to utilise the present small plotter at AGRITEX until some more substancial equipment was acquired. Some suggestions with coping with the known digitising problems will be given in the Recommendations (see Section 11), as is a brief reference to the preferred type of system. Finally, considerations must also be given to further FU staff training, i.e. budgeting and time allowances for this.
Since the “Summary” and “Recommendations” to this report make most of the conclusions fairly obvious, there is no point in re-itterating the material. Here an attempt will be made to “probe a little deeper” in order to shed some light on the overall background to some of the conclusions and recommendations made, and to put into perspective the ability to operate a GIS successfully with regard to conditions as they pertain presently in Harare. In order to accomplish this it will inevitably be necessary to range fairly widely, and perhaps to sound rather negative.
GIS implementation in Zimbabwe reveals a singular lack of success. Judging by the enthusiasm shown for GIS in many quarters, by the numbers of departments who are at some stage in the implementation process, or by the number of groups who were reputed to have GIS's operational, one might have expected a “thriving GIS scene”. This unfortunately, was far from the truth. As far as it was possible to ascertain, only 4 GIS's were actually functioning in the country. It seems important to evaluate the reasons why the intentions and enthusiasm were not coming to fruition. A tentative list may be proposed:
It is for these sorts of reasons that the recommendations may be seen as cautious. Clearly, neither the FU nor AGRITEX are really in a position to make a GIS fully function as it has the potential to. But what might seem the sensible option of postponing the idea of GIS for the present, has deliberately been avoided, since it was felt that GIS could be the “carrot” towards wanting to improve information collection structures, i.e. if it is seen that GIS will only work well with quality data inputs. Additionally, if implementation were delayed until most of the listed problems were solved then a GIS might never become a reality. So, despite the obstacles to success, there is no hesitation in recommending that the FU should move slowly towards the goal of GIS acquisition, taking all possible steps on the way towards familiarisation, collaboration with AGRITEX, database building and careful planning.
Nearly all of the recommendations are directly concerned with the core part of the whole consultancy, i.e. how to proceed with the implementations of a GIS for the Fisheries Unit of AGRITEX (as recommended in Section 9.5), and within this theme recommendations will be made in what constitutes a “three pronged” attempt aimed at overcoming the not inconsiderable barriers to GIS implementation. Other more general recommendations concerning the various Points have been made already throughout the report.
“Prong 1” - Overcoming the DAM DATABASE Problems
Members of AGRITEX fisheries, irrigation and livestock units (all potential users of the database) to formulate a joint strategy aimed at encouraging MEWRD to complete the database quickly. This might involve loans, or hiring of personnel or hardware or aid from the extension service in doing field verification work. Discuss the problem at high level with MEWRD and try to get agreed target dates to work towards.
Working again with relevant AGRITEX personnel, to establish a pilot area for an initial investigation using the database. This would need to be no larger than the area covered by one (or half of) a 1:250 000 DSG map sheet. Make certain that it is an area which has already been debugged and copied by MEWRD, and that it does not straddle longitude 30 E.
Ask DSG and NRSF to transform the pilot area chosen above, from latitude and longitude co-ordinates to UTM Grid References for each of the digitised editions of the 1:1 000 000 DSG sheets which they currently hold (see Section 5.3). ARC/INFO can do this.
From the potential DAM DATABASE fields (shown in Appendix 10), make a list of the most useful ones, or the easiest to get the information on. Again co-operating with AGRITEX, organise a week well in advance - say 3 months - when you could have a “blitz” of the pilot area to gather as much data on the dams as possible, i.e. using extension, FU and/or hired personnel.
If most of the above were done, which could take six months, then a good deal of experimental work should then be carried out within the pilot area on some of the GIS objectives as listed in Section 4.
“Prong 2” - Overcoming the Digitising problems.
The AGRITEX digitising pad needs to be tested. If it does not work ask the GIS expert at the DSG to come and have a look at it and to advise accordingly. He has volunteered to do this. If he cannot get it going, contact the DAM DATABASE specialist at MEWRD. He has contacts with a peripherals hardware firm (C.P.G.) who should come and look at it. There are also other firms who could help, e.g. Infotech, Systron, Realtime.
If it is still not working you might enquire about access to the MEWRD or Early Warning Unit digitisers - or even contacting FAO to do the digitising, i.e. for either the rest of the 1:1 000 000 DSG map sheets not yet acquired or for only the pilot area selected.
Whatever digitiser you may get working locally, or get access to, make arrangements with AGRITEX to get the rest of the work done (see Recommendation 13 below), e.g. this might be suitable work for AGRITEX topographic section. If not you will probably need to hire someone temporarily to do the work - contact head of NRSF or the GIS section of DSG - they may know of someone.
Check carefully the quality, i.e. detail of, the digitising, and the compatability of any outlines acquired.
“Prong 3” - Other Essential GIS Implementation Considerations.
Organise someone (perhaps through the National LIS/GIS Co-ordinating Committee, or someone at DSG or even someone through FAO Central GIS - Rome) to write a programme which allows the six figure grid references on the DAM DATABASE to be uniquely identified - this should not be a problem if Zimbabwe is treated in two halves, i.e. to the West and East of 30 E. For a map of the whole of Zimbabwe, you can join the two half maps together and transform the shape.
Encourage AGRITEX to make a decision, as soon as they are in a position to, vis a vis the draft report (ZIM/91/005), concerning the nature of any GIS systems to be implemented. An early decision would greatly expedite FU planning.
Be making up (or revising) required fields, including a grid reference or unit area location (ward or village name), to be input into any other databases that FU may require, e.g. certainly you would want a pond database. Work out a data collection strategy for these, probably concentrating on purposefully chosen pilot areas. Select a common database software system - one which works on Apple/MAC and IBM (perhaps ORACLE or dBASE) and then commence collecting and storing data.
Acquire all available NRSF and DSG digitised 1:1 000 000 sheet outlines, and any others which they may be willing to provide, as soon as possible, i.e. before they get “lost” or before charging starts. You can then assess the quality and work out additional digitising needs in collaboration with AGRITEX (refer to Table 1 for digitising needs). Commit some effort to searching for more detailed map outlines which could be of use.
Do as much background reading as possible on GIS, especially on implementation (see material left with Mr. Pilime and Mr. Chimowa). More information could be obtained by linking up with the National LIS/GIS Co-ordinating Committee. Compilation of a list of all possible backup support is also worthwhile (Section 6.7 of Meaden and Kapetsky (1991) outlines further ideas). Senior staff in FU and AGRITEX need to be continually updated on GIS so as to be aware of the potential.
Look into the possibility of FU staff (possibly Mr Chimowa) attending one of the NRSF courses, i.e. which has an ARC/INFO input. Enquire about and possibly attend further courses. This will be essential if ARC/INFO is not chosen by AGRITEX. Training and familiarity is vital.
A close eye should be kept on the GIS developments that CIDA are instituting in both the DSG and DNR - there could well be valuable spin-offs given the scale of the operation, especially since DNR do have overall control over the dams and fish
There should be a commitment in terms of manpower, by AGRITEX, to get someone to be employed full time in GIS work - a professional person who could undertake all aspects of GIS, and who could assist FU when necessary.
Systems Recommendation
Although the consultancy has not concentrated on types of GIS systems, it has been discussed with the CTA at FU and it is recommended that FU investigates this further in collaboration with AGRITEX with a view to ensuring compatability and so that budgetary strategies, training plans, equipment siting, etc., is considered. The consultant's views on GIS systems would lean heavily towards the PC ARC/INFO 3.4D system because:-
FAO, through Central GIS - Rome, can send example print-outs to show what can be achieved by ARC/INFO if required.
Having selected the software it is essential to get appropriate hardware as per distributors recommendations. FU should only obtain computer, monitor and plotter/printer plus some essential accessories. Details of computing requirements could be as per page G - 1 of Wood-Sichra (1991) (see Appendix 12), though a 486 chip might be adviseable plus a coloured monitor. Prices will therefore vary. Would recommend that AGRITEX itself obtain at least two Sun workstations - plus ARC/INFO version 6.0, so as to be able to satisfy likely future GIS demands.
Other Recommendation
That, since relatively few mutual data needs were found between nutritional interests and fisheries production, then nutritional related data should be added to the proposed new AGRITEX - GIS.
