2.3 The Land Unit map and its legend

The concept of the land unit map and its legend follows three main steps:

• the preliminary work;
• the field survey;
• the final land unit map and legend.

The preliminary work

In this phase a literature review and a preliminary legend and land unit map was prepared.

To obtain a general view of the geography, climate, geology and geomorphology, agriculture, natural vegetation and soil characteristics of the study area, a preliminary bibliographical study was carried out. Globally, there was not enough data, particularly for the climate, geology and geomorphology. Also, recent climatic data was not available. This insufficiency constitutes a major problem because, in Senegal as well as in all the sahelian countries, rainfall is considered to be the limiting factor for the vegetal production.

As for the geology component, an old geological map at a scale of 1:500, 000 was available. Fortunately, during the fieldwork in Thiès, a hydrogeological map of Cap Vert peninsula (to which part of the study area belongs) at a scale of 1:50, 000 was obtained.

The topographical maps at a scale of 1:50, 000 were used to draw the boundaries of the study area.

A preliminary legend was developed and the photointerpretation revealed seven landscape systems and thirteen landforms according to the morphology. Whilst the satellite image interpretation, using the vegetation type and coverage, revealed seventeen land cover/land use types (Table 4).

Then, a preliminary land unit map was produced by on-screen digitization using the processed satellite images as a reference base.

To facilitate the fieldwork planning, the road and track network of the study area was integrated to the preliminary land unit map. Meanwhile, the location and number of the relevés in each land unit is based on the associated legend code and the land unit area.

The field survey

The field survey was carried out from the 24^th of March to the 16^th of April in the Thiès region and 74 relevés were done. Using for the first time the Field Handbook (IAO, 2003), for each relevé, the following information were recorded:

• geography: the coordinates (acquired through the GPS). The location, elevation range, aspect and slope position (from the topographic maps) were specified according to the NRCS-USDA methodology (1998);
• geology and geomorphology: the geomorphologic processes type, area, activity and degree. The lithology, nature of the degradation processes and the parent material were collected following the IAO methodology which derives from McDonald et al. (1998) with major modifications;
• vegetation: the land use in the agricultural area (total area, technical knowledge, land improvement, crop types, distribution and farming system). The land cover type in natural vegetation (randomly sampled in the land unit, to define the total coverage of the woody, shrubs and herbaceous species, their list and physical characteristics) were collected according to the modified Eiten (1968) methodology and for the human influence following the FAO classification (1990);
• soil: mainly according to the FAO (1990) and the NRCS-USDA methodologies; the site surface description (coarse fragments, erosion, sealing, cracks) and the soil profile characteristics (texture, structure, moisture, color, rock fragments, pH, efflorescence, biological features) were recorded and the samples were taken for further laboratory analysis.

The final land unit map and legend

The information collected during the fieldwork revealed an unexpected complexity of the study area. It concerned mainly the land use/land cover that is strongly related to the amount of rainfall. In this particular condition, the possibility to use different kinds of satellite images, with different acquisition date and different spectral characteristics, was very useful.

The most useful image was the Aster, true colour composite 321 (RGB), that gave an acceptable distinction of all the features. The Landsat 5 TM, true colour composite 321 (RGB), was used to estimate the thickness of the sand cover on the Thiès cuesta backslope and the Landsat 7 ETM+ was used for the distinction between the vegetation and the geological features. To differentiate the agricultural areas, the Landsat 7 ETM+, false colour composite 453 (RGB), was used too. For the geological features enhancement, the false colour composite with the Aster infrared bands 468 (RGB) was utilised.

The data collected during the fieldwork were corrected and inserted in the database. Then, according to the new available information, the photo-interpretation and the on-screen digitizing were carried out once more in order to improve the preliminary legend and land unit map.

Thus, in the final legend (Table 5), the description of some landscape systems and landforms were modified.

The four landscape systems of the first legend: the sloping area, plateau with ferruginous crust, plateau without ferruginous crust and the isolated reliefs, were interpreted only from the geographical and lithological point of view. This differentiation was emphasized by the satellite images. Then, the structural and morphological analysis showed that the plateau, the sloping area and the isolated reliefs emanated from the morphological evolution of a similar feature, the Thiès cuesta, in which they were regrouped in the final legend. Also the land cover/land use was corrected and organized to match the real situation, which is subjected to annual changes. This gave thirteen land use/land cover categories whilst the preliminary legend had seventeen. For instance, the agricultural zone was divided into four types instead of the previous six (the "enclosed field of cassava" was kept to express a transitory stage).

This was followed by a last photo-interpretation, which was completed by the on-screen digitizing, to draw the final land unit map (Figure 4). After that, names were given to the polygons according to the main facets.