|
Focus
|
| Mapping
Desertification for Decision Makers |
Eddy
De Pauw
|
 |
|
How to bring together diverse and multi-thematic information in such a way that it can be used by policy makers for policy and management interventions in specific areas with specific degradation problems? The answer: use of Geographic Information Systems, or GIS.
Planners and decision-makers are not particularly fond of thematic 'pixel' maps, which they find difficult to interpret and combine for practical use. Instead, they want maps with a limited number of cartographic units, in which specific problems and potentials are clearly stated, so recommendations for action can be formulated.
To address this need, ICARDA’s GIS Unit has been developing the concept of 'agricultural regions'. These are integrated spatial units in which available water resources, climate, terrain, and soil conditions combine to create unique environments, which are associated with distinct farming systems and land use and settlement patterns. Agricultural regions are therefore holistic spatial entities with their own 'personalities'. Unlike thematic maps, they are real, not artificial constructs.
Agricultural regions can be characterized and quantified in terms of any biophysical or socioeconomic factor that has a spatial dimension. Within this spatial framework, desertification features can easily be accommodated as database elements.
To test the feasibility of the concept, we mapped the agricultural regions of Syria using a combination of remote sensing and expert knowledge: satellite maps, interpreted by experienced field researchers. The boundaries between adjacent mapping units were delineated by visual interpretation of recent satellite imagery, plus secondary information including geological, soil, landform, and climate maps. Figure 1 shows the results: a provisional map of Syria’s agricultural regions. The limited number of mapping units (27 in this case) is typical of this kind of synthesis map. The legend consists of ‘labels’ to which large attribute tables can be attached, including those related to desertification.
An example of part of the attribute table, related to desertification, is shown in Table 1.
| Table
1. Sample extract from the attribute table for the agricultural regions
of the limestone hills in north-west Syria  |
The next step was to develop a similar spatial framework for the whole of Central and West Asia and North Africa. Obviously, the same approach (manual interpretation of remote sensing data, with validation by experts) was impossible for this huge area. Instead, we developed a ‘proxy’ method based on the overlay capabilities of GIS. Four spatial themes were combined: climate, soils, landforms, and land use systems.
 >> Click here to enlarge |
The overlaying process may seem simple – the GIS software does it automatically – but it is actually fairly tricky. The data layers selected for overlaying need to match each other in resolution. Otherwise, if you combine layers with very different patterns of spatial variability, you might end up with many new mapping units that have no basis in reality. To avoid this, appropriate classifications had to be developed. And after overlaying, several ‘cleaning’ procedures were necessary to remove ‘spurious’ units.
After all these steps, we still ended up with 677 agricultural regions. This is both good and bad news. The good news is that complex dryland environments can be represented in a realistic way by combining a carefully selected (but limited) set of biophysical data layers. The bad news – at least for desertification researchers – is that there are many dryland environments and, therefore, the task of outscaling research from benchmark sites will be more difficult, and take much longer, than we had imagined.
| ___________________________________________ | |
| Dr Eddy De Pauw (e.de-pauw@cgiar.org) heads the GIS Unit at ICARDA, providing technical support to scientists from various disciplines. |
© 2008 International Center for Agricultural Research in the Dry Areas (ICARDA).
See
copyright and disclaimer information.