4.4 Land changes

In the last 15 years the study area has been influenced by different anthropic factors. In the North-East part a big quarry was created and along Oued Cherrat valley the condition of some area degraded. Some eucalyptus plantation was cut and, in a plot of the Ben Slimane forest vegetation regrowth occurred after a forest fire. In the South-East part, other areas were afforesteted with Pinus halepensis and Eucalyptus spp. and in 1985 and 2000 satellite images this variation is very evident. In vertisoils, agricultural fields have been converted to irrigated plots and in other soils tree orchards (olive and pomegranate) have been planted.

To detect land cover changes a multitemporal data merging was applied (Lillesand and Kiefer, 2000). The work was based on normalised 32-bits NDVI images of 07/27/1985 and 08/13/2000 Landsat scenes; preference given to summer images, is related to a clearer differentiation of land changes not disturbed by saturation induced by crops in agricultural land units and by herbaceous layers in natural land units.

The scatter plot of 1985 NDVI versus 2000 NDVI shows an ellipsoidal pixels distribution related to a high value of the correlation coefficient (R²=0.82) with two clouds of pixels, one in the upper and one in the lower part outside the ellipse indicating areas of changes, respectively a land cover increase in the upper part and a decrease in the lower one. On the other hand the pixels distributed along the main ellipse axis are related to no changes. A new image from 1985 NDVI was created according to the linear regression equation; this represents the theoretical variation of 1985 NDVI. This image was then subtracted from August 2000 NDVI. The scatter plot of 1985 NDVI versus the subtractive image presents an horizontal axis that can be used to mark the threshold values between degraded (below), restored (above) and quite stable (inside the ellipse) vegetation.

The subtractive image was classified with the Pseudo Colour Transformation (PCT) in the three mentioned classes according to these ranges:

• DN<59 decrease of land cover;
• 60<DN<119 stable cover;
• DN>120 increase of land cover.

Then a median filter 7x7 was applied to reduce noises.

To obtain an image with land change only inside the study area, the rasterization of the study area boundary vector layer was performed. The raster image was converted to value 0 (outside the boundary) and 1 (inside the boundary) and then multiplied by the subtractive image. The result was a land change image with PC only inside the boundary.

At this stage, an automatic vectorialization was performed using the PCI utility RTV to obtain the land change lines and point vector layers. The layer was then exported to Cartalinx to convert the lines in polygons and then exported again as shape file for Arc View uses.

Results and discussion

The results obtained by image numerical classification allowed us to distinguish some changes in our study area in the last 15 years (Figure 42). These changes are related to changes in environmental conditions the influence of multiple factors, mainly human activities and climate changes. In the map, only major areas were labelled to show the respective cause of change: afforestation, after fire regeneration, quarry exploitation, and vegetation for the golf course. Surfaces of different units are listed in Table 60.

The Ben Slimane region is characterised by several activities that are becoming more and more important particularly agriculture, pasture, forest utilisation, and commerce. The geographical context allowed the town to play a very important role for the local population and for the one of neighbouring towns; it resulted in drastic land use changes and development of commercial activities, not so common in the past. These were able to improve the living conditions of the local population. Analysing the results of NDVI processing, it was possible to distinguish different zones related to different reflectance. Colour composition of satellite images (1985 and 2000), aerial photographs, fieldwork forms and recent cartography (Ministère chargée des Eaux et Forets, 1999) allowed us to understand most of the changes inside the study area. Climatic conditions were analysed during the period from 1985-2000, and constituted a helpful tool from to a better understanding of observed changes: the first half of the ‘80s was characterised by a very severe drought followed by a light improvement of the situation especially in the last years. This improvement could have favoured changes in forest species density and composition and also in land use for agriculture. Observed vegetation evolution in the period from 1985-2000 could be mainly due to the growth of natural thuya population and to the afforestation in the SouthEast area of Ziaida municipality.

In the municipalities of Moualine El Ghaba and Ain Tizgha, the causes of reflectance increase in some areas could be due to vegetation regeneration after-fire, irrigation and the golf club, the presence of the hunting reserve, and cork oak degradation to the detriment of the secondary species.

In the fields used for agriculture, vegetation progressive trend can be essentially linked to agriculture practices; in fact the local population dug wells aiding the diffusion of small irrigated gardens for legumes and fruits. In addition, eucalyptus plantations close to farms became frequent to protect animals from the sun. The decrease of vegetation areas was also caused by infrastructures related to quarries, urban extensions and other factors related to the kind of vegetation in the farms and to drought. Areas for about 223ha were transformed into quarry; this value could seem very high but the emission of white powder from the quarries which deposited in the neighbouring areas could have strongly affected the accuracy in determining the boundaries by satellite images. This quarry was built to satisfy the needs of the Ben Slimane province regarding urban development and station of water (66ha during the period 1985-2000).

Inside the SOGETA (SOciété GÉnerale des Travaux Agricols) national farm, characterised by vineyards, the observed regression of vegetation cover was due to the removal of a part of the vineyards. It also seems that dayas surface increase, to the detriment of vegetation cover and vegetation variation close to the water spring of Ain Khneg Ennamer, is probably an effect of dry conditions. In the study area, some more changes were noted on small surfaces but identifying their causes appears extremely difficult. The influence of the date of the two images utilised and the satellites sensor calibration can not be ignored to fully understand detected changes; their effects should be well considered to validate this kind of process.