|
Arendal IV Workshop Report
At administrative and technical level
it was agreed to go ahead with a project proposal for a system-wide
initiative (SWI) to be presented at Centers' Week as a concept
note. Svein Tveitdal wrote the concept note with input from Workshop
participants and all present accepted the final draft.
Two technical sessions were held. At
the first, the participants exchanged ideas, data, and software.
This included an informal visit to the CIAT Geographic Information
System (GIS) facilities, which was useful in making contacts for
future collaboration. A more structured section was dedicated
to different ways of estimating and mapping poverty, at which
presentations were made by the Instituto Nacional de Estadistica
e Informatica (INEI), the Global Resource Inventory Database (GRID)-Arendal,
and CIAT. This created a large amount of interest.
Both in Workshops and in plenary sessions
a series of project concepts was developed. They will go forward
independently as collaboration between centers, whether or not
the SWI goes ahead. Arendal IV was a great success thanks to the
hard work put in by all those participating.
The Arendal process came about some
years ago from a meeting in Arendal, Norway. Representatives from
the Consultative Group on International Agricultural Research
(CGIAR) centers attended the meeting to work out the best way
of assisting the centers in their aims of studying agricultural
geography, crop distributions, and climate data, and putting together
the relevant datasets. The Norwegian government representatives
were enthusiastic about the work done at that first meeting and
agreed to fund an ongoing project led by GRID-Arendal to help
the rest of the CGIAR centers. This project incorporated a series
of visits to centers to assess each center's needs for GIS technology
and data. Part of the project incorporated various subcontracts
to produce datasets. Another part developed a system of metadata
documentation of the data available to the centers.
The original project was renewed for
a further 2 ½ years. Three meetings were held in Arendal,
Norway: Arendal I (1992), II (1995), and III (1996). This year,
1997, it was decided to change the venue and to choose one of
the CGIAR centers with a serious capacity for GIS geographic analysis.
The meeting at CIAT being the 4th meeting of the Arendal project
was therefore titled Arendal IV even though it was held in Cali,
Colombia.
The Arendal project has now run for
almost 5 years on Norwegian government funding and will terminate
sometime in May, 1998. The project commissioned a position paper
"Towards the Efficient Use of GIS Analysis by the CGIAR System
and its Partners", which was presented at the GRID-Arendal
steering committee meeting in Cairo, in May 1997. The paper presented
four possible options for continuing the project, of which the
overwhelming acceptance was for a new SWI.
Arendal IV was originally planned as
a two-part meeting. The first objective was to put together a
project proposal for an SWI , in which a broad group of donors
would be interested, to carry on the Arendal process. The second
was to hold a technical meeting of GIS experts from the centers
to discuss details of datasets and techniques that are available
or needed. As it turned out, the two halves were not kept separate.
The group working on the SWI project benefited from discussions
in the technical sessions and the GIS technical staff from centers
were interested in adding input to the SWI. In this document they
have been separated for the sake of clarity.
The first day was taken up wholly by
a small group working on a project proposal for an SWI. Day two
started the technical sessions. However, interspersed within the
technical sessions was the opportunity for discussing the project
proposal for an SWI. On the last day, the centers representatives
spent the day in working groups developing project proposals for
collaboration between a number of centers. These proposals, although
not strictly part of the SWI, were projects with high priority
in the CGIAR system for collecting and managing important data.
As such, if the SWI goes ahead, they will form subprojects with
collaborating centers within the overall SWI.
Members: Daniel van R. Claasen (UNEP),
John Dodds (ICARDA), Norberto Fernandez (ROLAC), Carlos Garces
(IIMI), Robert Hijmans (CIP), Peter Jones (CIAT), Ramon Lastra
(IPGRI), Carlos Larrea (MOSTA), Were Omano (KARI), Paul O'Nolan
(ISNAR), Svein Tveitdal (UNEP/GRID-Arendal), Markus Walsh (ICRAF),
and Jeff White (CIMMYT).
The group met on the day before the
Workshop to put together a first draft concept note on the initiative
that could then be discussed and improved upon by CGIAR representatives
during the Workshop itself. The proposal made by the project secretariat
was used as a basis for discussion.
The objective of the meeting was to
look at ways of keeping the Arendal Project running. The history
of the 4 years of the project was reviewed. The main objectives
had been: to establish long-term linkage between the CGIAR and
the United Nations Environment Program (UNEP), to strengthen environmental
information management capacities, and to use a coordinated and
integrated approach. The project's Phase II outputs were data
and information management, institutional development, and networking
activities. The Norwegian government might add a small amount
of funding for continuing, but needed a "sustainable strategy".
UNEP has also shown interest in continuing the project but in
what relation is unclear. The collaboration has been successful
and Director Generals (DGs) and Deputy Director Generals (DDGs)
supported its continuation. Many models are possible, from rotating
annual meetings to major projects. Developing technologies and
indicators were highlighted, for example, land degradation. The process of developing an SWI had been slow because of the need to better understand it and because of the need to inform CGIAR directors of the potential of continued collaboration. The presentation given at the Cairo meeting was then reviewed. People need to be convinced that GIS is much more than just the gathering of software and hardware. Four options for the SWI operation can be chosen from: An SWI devoted to GIS: similar to that of integrated pest management (IPM), Participatory Research, and the Soil Water and Nutrient Management program (SWNM) An association with ecoregional centers UNEP support for a new project Adding GIS to an existing system-wide initiative
The consensus from Cairo was in support
of the first option.
GIS has many parallels with biotechnology.
It must be emphasized that GIS is a set of tools. The end is to
resolve problems, particularly those which focus on CGIAR priorities.
Experiences with SWIs to date have not been effective, but even
so it is the relevant way forward in this case. A DDG or DG is
needed to be the spokesman at the Inter Centers Week in Washington.
In order to sell to donors these days the full support of the
National Agricultural Research Systems (NARS) is needed. The system-wide
review will look both at "new science" and "information
technology". Innumerable other players are in the game. How
can these resources be accessed? The question of intellectual
property rights must also be addressed - data are becoming as
sensitive as germplasm.
The objectives of the initiative are
to establish center-wide networking, mechanisms, and services;
diagnostically help research planning; assess the impact of research
or development, either ex-ante or ex-post; and to further research
used as a strategic tool. Activities would include development
and information management, institutional development, and networking.
A geographic information network of CGIAR centers and other international,
regional, and national organizations would thus be set up. Outputs
would include spatial and thematic datasets well documented and
usable by CGIAR centers; electronic data lists and network data
sharing and dissemination; and increased capacity in IARCs and
NARS in GIS, internet, and databases (spatial information management).
It was pointed out that the CGIAR has
comparative advantages for this initiative. The CGIAR has the
catalytic role of serving the NARS. It is working with others
with greater expertise and databases already available - public
and private - and building on it. Also, with networking the CGIAR
can expand broadly to disseminate what it does.
It was agreed that networking and the
involvement of NARS must be in the concept note. The Technical
Advisory Committee (TAC) has described the CGIAR as "a global
information system for international agricultural research",
which bolsters this idea. Others (e.g., Food and Agriculture Organization
[FAO]) are doing similar work so the idea would be to steer rather
than compete. Data should be made available, especially to the
providers, and not taken away. Overlaps in projects exist and
are numerous in networking aspects, for example the 50% overlap
in over 500 projects in the Caribbean. It was agreed that a gap
analysis would be beneficial. The CGIAR needs to avoid duplication
and work more on complementing other work being done. This would
be cost effective.
The form of a short concept note for
presentation at Centers Week was agreed upon. The involvement
of DGs was thought essential. The actual organization of the initiative
would give opportunity for innovation. All centers want to be
users (potentially). The need for access to data and techniques
is basic. Some centers are known for expertise (e.g., CIAT for
use of GIS), other centers would like influence of some sort,
on the Steering Committee for example. Those centers not represented
at the Workshop were to be informed and involved if possible.
It was proposed to write a first draft
that evening for perusal next day and to be given out to Workshop
participants for feedback.
In his opening remarks at the Workshop,
the following day, Dr Grant Scobie, the DG of CIAT, spoke about
the SWI. When planning for the its future developments, he advised
the participants to bear in mind the following points. The CGIAR is only 4% of research in developing countries. Its work should therefore be complementary to the rest. The role of the system-wide program is to ensure a common agenda and shared goals to avoid duplication. On the downside, it has potentially high transaction costs. What is the output? We must strive for efficiency. Examine the question of impact. Donors want evidence of the impact of their investment in research. Can impact be sustained? Is the rate of return as good? This is particularly important in the GIS area. Some think of GIS as a set of tools - input rather than output.
The CGIAR centers have a changing agenda.
In the past it was productivity. It is now greatly changed, a
complex and expanded agenda. GIS can make a great contribution
here in natural resource management. The urgency is obvious with
world population growth and the question of food supply.
Svein Tveitdal then gave a brief background
of himself and of the GRID-Arendal project for those unfamiliar
with either. He recapped on work done to date and pointed out
that since the GRID-Arendal project is about to end, it was thought
that some of the initiatives should be continued. The SWI had
thus been proposed. The previous day's executive session was summed
up, giving a plan of what was proposed. The concept paper first
draft was circulated as a starting point for discussion. Inputs
were requested from those present.
In the GRID-Arendal project, the emphasis
so far has been upon awareness building, database building, and
basic project activities. A lot of techniques extant are almost
finished so data liberation has been implemented on subcontracts.
The project is being used to initiate major project proposals
talked of but not acted upon. In the SWI initially all three points
would be used.
It was advised that on the SWI the datasets
available should be put together so as to prioritize those that
were missing. Risk assessment is a big part of the poverty issue.
The continuity of an individual's work is important, as is the
avoidance of replication in the future. Institutionalization can
be an important part of the SWI. It should also be able to tap
into individual outside expertise.
The point of avoiding loss of data was
emphasized, using the example of the Kenya Agricultural Research
Institute (KARI) where the maize database has stopped. Availability
will prevent this happening. Loss of data has occurred in the
past and institutionalization has the potential here to make a
great difference.
A 4.30 pm meeting continued work on
the concept note. Comments on the first draft had been received
and Svein Tveitdal had arranged a meeting with Grant Scobie to
go through the document on the last day of the Workshop. The ideas
given to date were further discussed. The acronym Geo-Information
Network (GIN) was suggested and unanimously agreed upon. Various
options were put forward as to the graphics to be used to sum
up the SWI. The following diagram was drawn up for consideration.
Other suggestions for diagrams were
asked for.
It was urged that innovative measures
be used wherever possible, for example, having a virtual steering
committee through electronic means. Agreeing that this was a good
idea to follow up, it was thought that not all meetings could
be replaced thus.
Regarding the involvement of the NARS,
caution was advised against overselling the technology as cultures
are not changing as fast.
Official Minutes on Participants' Reaction
to "Concept Paper on SWI on GEO-Information, 2nd Draft"
On the second day of the Workshop, comments
were asked for respecting the 2nd draft of the Concept paper.
Most participants expressed the view that the document was basically
good as it stood. It was added that it was extremely important
to lay out for each activity what the benefits and impacts would
be. A structured method of making publication available was also
called for. The procedural question of how the centers would collaborate
needed to be clearly set out: where the data would be lodged,
in what form, how advertised, what products et cetera. A structure
was asked for which would be useful across the system and could
be used by the NARS. It could then be built up slowly as a global
dataset. Combined synergistic products should also be worked upon.
Concern was voiced on the possible interpretation
of "in kind" funding, also on the issues of access to
and ownership of data.
The problems should be more clearly
identified, such as data quality and overlapping. Likewise the
opportunities should be stated, the fact that there is no competition.
Using the World Wide Web (WWW) to bring GIS to desktops in this
way is revolutionary. There will be a spillover to other dataholders
in the centers. Enrolling outsiders such as FAO should be explicitly
stated in the document.
Transaction costs should be minimized
for example the Committee could be a virtual one. It must be decided
whether the SWI is run by the CGIAR system or is a project from
outside. If it is coming from within, it will have links out,
especially in projects that require specializations. The advantages
of having standardized system databases and documentation should
be included. A metadatabase system could be developed to use for
both internal and external databases.
Consulting with a director general who
understands about funding is vital. The original GRID-Arendal
project had funds for limited projects. It is important to know
where TAC stands re the SWI. This links to whether an external
group is used or it is done internally. This point should be open
to manipulation depending on what the Director General thinks.
There may be a value in looking at a consortium approach as a
possibility if the SWI is a problem. If it really is an SWI then
an internal lead center is the best way to go forward.
The Concept note should state which
approach is to be used. If it does not go forward, then try for
a consortium approach. It is difficult to expect an external organization
to manage the project. Using one to do some of the work may alleviate
political problems within the system as to which center does which
part.
Concept Note, 3rd draft
On Thursday, the last day of the Workshop,
Svein Tveitdal gave an update on the Concept Note for the SWI.
He had met with a positive response from Dr Scobie, who had volunteered
to take the new draft to Centers Week where he would involve colleagues
and talk to TAC. A process to take the note to a wider audience
had been agreed upon and hopefully final endorsement would be
given at Centers Week. The input from Wednesday's discussion had
been included, especially on activities. As regarded the funding
strategy, there appeared a reluctancy to commit core funding to
something shared with donors. The concept note now had a center
input as to what is already being done, such as cataloguing, storing,
and data management. A cost recovery component had been included
to cover the possibility of loss of core funding for such necessities.
The Workshop
Dr Grant Scobie, CIAT's Director General,
gave a warm welcome to all participants. In CIAT's 30th anniversary
year he felt it was particularly appropriate to be hosting the
UNEP-GRID Workshop. CIAT likes to share experiences and learn
from others. Its involvement in GIS goes back a long time.
With regard to the proposed SWI, Dr
Scobie's advice is given above.
Finally, he hoped the Workshop would
be fruitful and wished participants every success.
Collaborative Project Proposals
Projects would go forward under the
aegis of the SWI. A short session later in the Workshop would
look at what such projects might be.
It was pointed out that projects should
focus carefully on strategic issues that affect all centers. Allegiances
could be shared for a single goal. When working out projects,
different combinations are possible, they are not necessarily
to be run by the SWI but need the SWI to maintain viability. The
modules are output-oriented. The need is for the module but the
database required to produce it is not as obvious to a donor.
Both aspects have to be covered. Three modules to produce outputs
were suggested:
Extend all crop and livestock mapping
to the whole of the tropics
Look at NARS capabilities and the possibilities
for upgrading them
Decision-support systems (DSS) - how
to use GIS as DSS tools.
These modules were considered, along
with others, in working groups on the last day of the Workshop.
Review of Center Needs
A format was suggested to see what needs
were in common and to work out what each center has got/uses/is
doing/needs. A list was put together with input from various participants
and suggestions made as to its organization. Some conflict of
ideas arose and it was suggested that technical issues be laid
out, and that each center name three items from the list that
were best done at system level.
The list was seen as containing three
areas: socioeconomic, bridging the gap between genetic resources
and NRM, and datasets. It was also seen as three different areas:
datasets, decision support systems, and data management.
Presentation of Data Exploration
Tool
John Corbett presented the Data Exploration
Tool developed at ICRAF and the Blackland Research Center. The
spatial characterization tool itself deals with generic questions
about spatial variability and homogeneity, addressing issues that
come up in the working environment. It gives a vision of what
is intended and is a manifestation - doing specific actions. It
comprises a query tool and a dataset package. The models help
package data in a more biologically correct order.
Corbett said that the Blackland Research
Center receives a lot of similar questions that are individually
a lot of work to answer. So they put together a set of generic
questions which end users can use to go into the database. Once
in place, it becomes more elaborate as other generic questions
are added.
To determine effective environments
we have taken climatic surface data through cluster analysis.
Each resulting polygon is of a high status probability of similarity.
Sufficient data is now available to be far more specific. For
the objective specific classification, we initially use the effective
environments then group them with another set. In the example
being shown at the Workshop, we used soil information. We then
run the model, the one shown here is for maize. Thus we get the
yield simulation and can compare it with other results.
The GIS Awareness Package
Examples were handed out of the GIS
Awareness Package produced by UNEP-GRID-Arendal and available
on the Internet, http://www.grida.no/Cgiar/htmls/awpak.html. The
package comprises work from various sources with two types of
end product. All 43 examples collected are available on the WWW.
A printed form of the package is to be disseminated at Centers
Week. The UNEP-GRID-Arendal program has a special homepage on
the Web and one of the new products available at this site is
the GIS Awareness Package. Maps can be downloaded direct onto
computers. The database is not being continued but it would be
a good output to carry this on. The CGIAR system should be in
a position to maintain initiatives such as this.
Some of the authors of examples in the
Awareness Package then explained their contributions.
No. 22. GIS and breeding for drought
tolerant maize in sub-Saharan Africa
(Jeff White)
This example came about because a scientist
wanted to understand conditions in sub-Saharan Africa for work
he was doing on maize. He needed data on drought intensity. The
data and basic tools are available. The data is being left in
UNIX and the final product is being produced on CD-ROM. The same
datasets will be useful for others. The flow goes from a breeder
with a specific problem, through the CIMMYT laboratory with connections
to the USA, to other users within the system. The big challenge
here is maize breeding with a 5-to-10-day interval requirement.
No. 7. Environmental and Sustainability
Indicators for Latin America and the Caribbean (Manuel Winograd)
A description of production indicators
for Latin America is available. We built a database and GIS interface
to develop a product for policy makers. The pressure indicators
are to be released in November on CD-ROM. The concern is to give
policy makers information for their use. A homepage has been created
to interact with users but there have been no replies during the
first year. This is one of the frustrations of the work, in the
real world there is little cooperation.
We have made contacts with Ministries
et cetera and visited some policy makers to interact on what was
needed and what was offered. A small publication of 50 pages was
issued on indicators. The policy makers' reaction was that they
had no time to read it. However it has been used.
There are different scales in policy
makers, from the World Bank to the local ministry. We have developed
a national project with Colombia both at State and ecological
levels to test the use and efficacy of the indicators. We used
Arc View 2-1, which gave some problems with maps but was used
because it is the most easily available in Latin America. Data
from the National Aeronautics and Space Administration (NASA)
satellite was used to identify areas and we developed a map and
overlaid it with a map from a local survey. This was to show policy
makers that a map is an interpretation of reality. They do not
seem to care about the quality of the data. This was to show them
the difference.
A second step, of the project is being
developed in which a big component is capacity building and training.
No. 5. Diversity of wild potato species
in Bolivia (Robert Hijmans)
We used a germplasm database to map
locations where wild potato species might be found. This was done
to locate areas for preservation in situ, where it should be done,
and to produce collection sites. The data has a few problems.
Some big biases occur on collection (road bias, and restaurant/hotel
bias). We correct for this in different ways. We can ration between
the number of accessions and the number of species in them, or
look at the number of expeditions undertaken.
No. 38. Simulation models for studying
limiting factors in potato production (Robert Hijmans)
This example is important to work in
the CIP. Breeders are looking for new versions of potatoes, and
wanting information on adaptation. This example on drought stress,
using CIAT's climate data, can determine where the crop will grow
and its potential yield. We compare potential yields with irrigation
and natural rainfall. The idea then is of finding where drought
is probable and that is the area to work in to develop resistant
types.
This was a small-scale project and is
now being extended. Socioeconomic datasets are being linked in
through GIS. It is a difficult process. We start on the world
scale and work down as far as we can.
No. 16. Using SAR images to locate gaps
in the riparian forests of the Colombian Tropical Lowlands (Nathalie
Beaulieu)
This is a straightforward application
responding to the need for information. In many areas of the Colombian
Tropical Lowlands, livestock dominates and the riparian forest
is important in these areas. The Corporacion Colombiana de Investigación
Agropecuaria (CORPOICA) has a program to try to promote the use
of trees in pasture. Gaps occur in the riparian forest and the
idea was to identify these as potential for reforestation by farmers.
They are areas where there is river but no forest. We used a LANDSAT
TM image and radar imagery, which show the clear contrast between
forest and pasture. Classification was thus easy. The mapping
was completed and sent to CORPOICA so that they could focus on
the areas needing attention. Subsidies in Colombia encourage reforestation.
How this system works with streams is not clear, because Colombia
owns the streams. We have to establish if such a program can be
promoted, which is a policy issue.
This was a case of having the data on
hand and putting it together for another use; we had already digitized
the stream lines for another study. The work was cost efficient
because of this and production was rapid. The ecological value
is high and it plays an important part in maintaining biodiversity.
An estimate of the quantity and impacts
of soil erosion in Costa Rica (Glenn Hyman)
This item does not appear in the printed
Awareness Package, which is not yet up to date only including
43 of the 49 examples collected. It is a similar piece of work
to the preceding one and was done in collaboration with the Centro
Agronómico Tropical de Investigación y Enseñanza
(CATIE). We took hydrological models based on the stream power
theory and Darcy's law. We worked in areas where we know the sediment
and agricultural chemical movement and can reference to the buffer
area. We used soil, land-use, and slope maps and then developed
the model taking into account the differences. We overlaid with
land use from satellite imagery to show nonforest area by streams.
These areas are protected by law but the law is not enforced.
We are looking at the possibility of extending the model to the
rest of the watershed.
The reference area was based on the
knowledge of what was happening in the area. CATIE has plot experiments
there, so data was available.
The Poverty Process
The Geography of Poverty in Ecuador
Carlos Larea
This work was done about 1 year ago.
A map of poverty in Ecuador was constructed and some studies of
poverty in Latin America made in collaboration initially with
the University of Toronto. It was felt, for the purposes of this
Workshop, that illustrating methodology problems was more significant
than analyzing the contents of maps or showing more elaborate
ones. How can we in an internationally comparable form, compatible
with existent information, reach proceedings established to provide
poverty maps for the third world, especially Latin America? Methods
that are commonly used and can be used on a regional level will
be discussed schematically. The case of how problems of deficiency
of information in Ecuador were solved will be explained.
Definition: Structural situation of
basic needs deprivation
Poverty was defined as a situation that
affects a household, structurally impeding (not transitorily)
the satisfaction of the basic needs of those composing the family.
The family has needs in health, education, housing, and nutrition.
So the household poverty situation was taken as a base.
Why poverty is important in Latin America
Here we have the reasons behind the
work being done. The World Bank made a conservative estimate of
the incidence of poverty in 1995, estimating 35%, or 164 million
people, as falling below the poverty line. Poverty is more generalized
in Africa and southern Asia. Rural areas are the most affected.
In Ecuador, two thirds of the poor live in the country. This is
a general phenomenon in Latin America.
Poverty has been increasing in most
countries since 1982, which was a severe crisis year in the region.
There have been technologic and socioeconomic changes but the
general tendency in Latin America is an increase in poverty since
that date. The tendency of concentration in the instruction of
income also exists. Changes particularly affect the small-scale
farmer. For example, the expansion of new export agriculture excludes
the small holder who does not have capital or economic resources
to face the situation. So one of the groups most strongly affected
by the situation is that of country people.
Methods to measure poverty
There are two basic ways of measuring.
The direct method is by measuring unsatisfied basic needs. Data
for this is available from census, poverty maps, and the World
Bank survey done over many African, Asian, and Latin American
countries. The survey does not permit an understanding of a situation
because it is a small, micro-regional sample.
Which are the basic needs? The method
has limitations because most of the information is available from
census and is thus incomplete and biased. The census consists
mainly of housing and education data and really reflects housing
infrastructure, which is only one dimension of basic needs. Also,
change is slow; a census is usually taken every 10 years. It does
not reflect short-term changes or politics.
The indirect method uses consumption
data or income poverty lines. But, poverty lines differ. The minimum
line is where the basic needs of a household are satisfied. Because
the income side is sometimes hidden, consumption is easier to
measure, but is also often underreported. This type of data is
usually taken from samples and applicable only to large geographical
areas. It is the most used method but deficient in that it is
done by questionnaires (e.g., Chile, Colombia done systematically)
and is mostly not done on a small scale. Thus it is difficult
to compare regions at the second and third level.
Projection method used in Ecuador
The shortcomings of data in Ecuador
were overcome in various ways. For the sources, the Life Standards
Measurement Survey (LSM) of 1994 and the 1990 Census were used.
The LSM is a World Bank Survey done
over many African, Asian, and Latin American countries. It has
a common formula and extensive questions (2700) done over 15 days
with each household. It covers 5000 households in urban and rural
sectors and allows in-depth analysis of each household. As stated
above, it does not permit an understanding of a situation because
it is only a small, micro-regional sample. The Census has few
questions and is schematic but has the advantage of being a simple
national coverage. We define the common questions of the census
and survey.
As an indicator we used per capita consumption.
This was calculated with prices adjusted to regional changes and
to be comparable. We then used the projection method of mathematical
modeling from shared variables. This shows the determinant factors
of family consumption. We used models that were regionally distinct.
This is important because in Latin America rural and urban situations
are different and can be different between regions. Eight models
were made for Ecuador in two stages: first, labor income and family
labor income; and second, family consumption. Separate, regionally
defined models were made and regression models to check.
Results
The results are the maps being shown
at this Workshop. This process is a possible way of getting viable,
believable maps.
Poverty characterization: Why poor households
are poor
Specific basic needs were analyzed using
multivariate statistical methods. Factor and cluster analyses
were then made.
For education, housing, health, and
nutrition, a composite basic needs index was made.
It is most important to work out why
the poor are poor rather than, for example, how many poor there
are in a region. We looked at what element should be addressed
in a particular area to improve their conditions. Poverty is an
indicator of results and can be the result of various different
causes. An interesting complement would be to determine the typology
of poverty, the factors leading to poverty in a particular area.
If the cause is health then different politics can be used to
help. An integrated analysis is required, not one that just informs
on the extension (descriptive) of poverty but one that indicates
the causes or the types of poverty.
In Latin America we are talking about
absolute poverty. We are basing analyses on the basic cost of
a food basket for a family for calorific and protein needs. This
is one half of the satisfaction of other needs (education, housing
et cetera). The cost of living index can be used instead. The
needs for nutrition can be worked out.
We have not used environmental indicators
but the association is there. For example, poverty is linked with
soils, and with deforestation and erosion, which lead to a lack
of cultivatable lands. An integrated database would allow comparisons
and needs to be done. Life expectancy data would be a useful factor
in the analysis, but mortality by age is required and no information
is available on this at the micro-regional level. Deaths are not
registered, only births. Questionnaires could be used with women
to determine numbers of children born and alive et cetera.
Poverty Mapping, Honduras
William Bell
The Inter-American Development Bank
(IDB) funded this 2-year project. The first staff members were
hired in July but we have been developing the database for over
a year.
Development of GIS Application
We too are looking at Poverty Indices.
It is difficult to get quality data. For the Wellbeing Ranking
Index we are trying to link with the work of anthropologists and
sociologists. We are using Fondo Hondureño para Inversion
Social (FHIS), Living Standards Measurement Survey (LSMS), and
census datasets to be able to compare between different regions
in a country. A tiled roof on a house might be an asset in one
area and a problem in another-the perception of house quality
differs greatly cross-country. We are trying to link down to the
village level and trying to link the Wellbeing Ranking Index to
census data. We are also using data analysis. We aim to find the
key determinants of rural poverty.
Current Hypotheses
From the GIS point of view the range
of determinants of rural poverty includes; market failure, human
capital, information/technology, and infrastructural/ regional
economic diversification. Points such as access to market can
be easily checked. Information on all these aspects can be obtained.
We are trying to look at a totally different way of measuring
poverty. The World Bank views it from an economist's viewpoint,
the IDB from the sociological viewpoint, but we want to try to
join the two.
Development and modification of current
poverty indices
Many indices exist, some can be easily
obtained from census data but most cannot and we need additional
data. For the review of present poverty indices we use the poverty
line and unsatisfied basic services in relation to the datasets
of Honduras to build an appropriate index. For the poverty mapping
project we are using the best available data to develop a human
needs index. Important variables are used:
Income variables like monthly expenditures
at household level (extracted from LSMS datasets) and the total
monthly expenditures at village level.
Shelter variables to measure the quality
of housing by examining the construction materials used.
Biophysical variables like land cover,
slope, agroclimatic, soils, and protected areas.
Other relevant variables like road,
electricity, and telephones.
Educational variables like enrolment
rates, number of schools, access to school, literacy rates, student-to-teacher
ratio, and wastage rates.
Health variables like health facilities,
accessibility, infant mortality rates, safe water, sanitation,
number of doctors, and primary health care facilities.
Nutrition variables like per capita
food consumption, and per capita caloric intake at the household
level (extracted from LSMS datasets).
The educational variables can be taken
from the census, which is good on a large scale, but if we want
details we have to go in and collate data from the Ministries.
This takes time and coordination but is necessary. Likewise for
the health variables at village level we need to work in the country.
We have the LSMS but it has not been sampled in a statistical
way, lots of biases exist in the sampling and we do not know exactly
where the samples were taken. Using LSMS data has many limitations.
We try to collect biophysical data. CIAT has a lot of this kind
of information. We have thematic map data for the whole of Honduras
at 30 m. We also have elevation and slopes data, but the latter
needs to be used with care in the tropics. It has to be extrapolated
from other data, applying a correction factor to smaller scale
maps. The relation of slope with poverty is significant. We are
also interested in climatic risk data. We are building classical
GIS maps including roads and some data on telephones and electricity.
This type of data should not be ignored; a lot of information
can be gleaned from a power company map re the distribution of
population on landscape.
Geo-processing
A population analysis is made at village
level to predict the population behavior and therefore relate
it to the existing poverty levels. We focus on changes in relation
to population density and migration patterns over time and space.
We isolate the principle features of poor villages, households,
and municipios. We look at employment characteristics at the village
level, at who are gainfully (or not) employed.
Working definition of poverty
"Poverty is a social, economic,
or environmental disharmony that limits an individual from the
realization of their full aspirations and satisfaction."
(Tonny J. Oyana, 1997, personal communication). This definition
is wide scale, complex, and far reaching. When dealing with poverty
it is hard to generalize.
Selected datasets
We are working at household level data
(census) aggregating it up to village level and quality controlling
it at household level. We have a lot to learn on the visualization
systems for detection of errors. We are aggregating out from village
to municipality to department level to then be in a position to
move backwards and forwards. Millions of records are kept in the
main database. The data needs to be understood.
The kind of data and sample data that
we have are:
Population census data: at household
level for 1974 and 1988.
Agricultural census: at household level
for 1974 and 1993.
World Bank LSMS: sample data (household
level)
FHIS data: sample data (municipio level)
Wellbeing data: sample data (village
level)
Education data: selected variables (village
and municipio levels)
Health data: selected variables (village
and municipio levels)
The datasets have limitations such as
evolution of boundaries, definition of a village, temporal aspects,
inconsistencies in coding, and the sector approach to data collection.
Point data and sample village data do not integrate well because
people do not take a global positioning system (GPS) with them
when they go to take sample data (to note where they are taking
the data). Integrating across different levels of data is an interesting
challenge. The amount of information available varies by population
density. To sample, borders are needed but the division between
villages on the census is unclear and maps of village boundaries
are unavailable.
In the well-being data, we wanted to
try and encapsulate some of the variability within the country.
We took six variables and tried a cluster analysis and did a sample.
In our three study areas we looked to see how much of the variability
of the country we would get in 90 samples (sample size can be
increased). Rural sociologists are doing the interviewing work.
We now have village level data and are
putting them together to see the relationships. We have some population
change from 1974-88. Some of the villages have changed (definitions),
and the number of aldeas has increased. These have to be taken
into account on changing boundaries. We are now doing exploratory
mapping. A map is much more descriptive than a set of tables;
it can be quick to do once we have the data, yet useful, especially
to the planning process. The data is the big problem as outlined
above. No easy way exists to deal with such a disparate set of
data.
Getting data back to the data providers
We are trying to get information back
to the people who give us the data; we must do this if we want
their future help. We are building a CD-ROM that can be used for
display or as a query capability at the national level. At the
municipality level we also make hard copies. At the village level
we take posters, or take part in some kind of participatory method.
In our three study areas we make sure we have a well-defined mechanism
to get information back to the people who work with us.
Poverty - West Africa
Svein Tveitdal
Background
The TAC approached GRID-Arendal for
help in producing reliable statistical and cartographic products
on poverty and land-use potential. These activities are integrated
into an ongoing collaborative project between UNEP and the CGIAR,
"Use of GIS in Agricultural Research".
The poverty mapping activities have
been conducted in a relatively short period (March-May 1997) starting
with a small Workshop discussing scope, data sources, and outputs
of the study. Jake Brunner of the World Resources Institute and
Uwe Deichmann from the UN Statistical Division gave invaluable
inputs to these preliminary activities.
Objectives
The aim was to generate reliable statistical
and cartographic products to communicate the relationship between
rural poverty and land use potential in West Africa, in order
to provide information to ensure optimal use of research investment.
Further, the project served as a pilot
study to investigate an appropriate approach to identify the location
of poor people on a global basis with a reasonable investment
of time and resources.
Approach
To measure poverty a variety of factors
needs to be taken into account. A standard indicator frequently
used is the pure economic value "gross national product (GNP)
per capita". Using this attribute makes sense if the study
is to be comparable at a global level. However, GNP data is not
available universally at the desired accuracy. We therefore chose
to represent poverty following the example of a World Resources
Institute (WRI) study in progress entitled "Human Development
versus Aridity in West Africa". We use the indicator variables
of the Human Development Index (HDI) and the actual data comes
from the Health and Demographic Survey (HDS).
Because none of the HDI indicators are
explicitly captured in that survey, surrogate variables are used
from the DHS data to infer poverty levels. The four surrogate
variables used in this study are: child mortality, adult female
literacy, primary school enrolment, and children with stunted
growth. These data exist for 2263 sample points within West Africa.
In a second step, we excluded all urban samples to get a better
idea of the situation of the rural poor (1113 samples). |