Abstract :
[en] Rangeland or natural arid pastures of Morocco are defined as ecosystems where there is a natural or semi-natural vegetation composed of steppes, shrubs and grassland used mainly for livestock production as climate and soil are unsuitable to agriculture. The arid rangelands of Morocco occupy an area of over 33 million hectares between the isohyets of 100 and 400 mm. These areas represent livelihoods for thousands of people and protect the country from desertification. The main objective of this study is to provide scientific community and decision-makers methodological tools for assessing arid rangelands, based on geomatics and biophysics data.
This study is divided in four parts:
1. The first part presents an overview of the threat of the desertification, emphasizes on the causes and consequences of rangeland degradation in Morocco, from literature sources, statistics, climate data and maps. Morocco rangelands are located in ten different pastoral zones that differ from each other by their floristic composition, soil and climatic conditions. According to Globcover map these rangelands are mainly composed by bare soil, herbaceous vegetation, shrubs and deciduous savanna mosaic / shrub or forest. The largest pastoral zones are: the Saharan zone, Pre-Saharan, the Oriental plateaus and the Valley of Moulouya. These zones are the most degraded with respectively 97, 89 and 69% of their total area. Available information on several pilot areas shows that the Moroccan rangelands are degraded due to many factors, which include overgrazing, cultivation, population increase and climate variation.
2. The second part of the study demonstrates the usefulness of remote sensing for assessing drought in arid rangelands of Morocco. Bi-weekly TERRA Moderate Resolution Imaging Spectroradiometer (MODIS 250 meters) data were used for this purpose. A Preliminary mapping by using Landsat TM5 of major land cover types was carried out to extract the pasture area. A comparison of annual and seasonal Normalised Difference Vegetation Indices (NDVI), Vegetation Condition Index (VCI) and rainfall during the time period of 2000–2008 were carried out. Results show significant correlations of either NDVI (r= 0.72**) or VCI (r=0.42*) with past season (3 months) rainfall. NDVI variation is a good indicator of vegetation changes and consequently can give a reliable indication on drought conditions in the study area. NDVI values lower than 0.2 are indicative of drought occurrence. NDVI values between 0.20 and 0.28 indicate average weather conditions and values higher than 0.28 correspond to humid conditions.
3. The third part presents an original knowledge-based approach for mapping the degradation of rangelands in North Africa. The study area is located in the high plateaus of eastern Morocco which include 3.5 million hectares of arid rangeland steppes. The approach consists in using datasets derived from Landsat TM satellite imagery, lithology, phytogeographic data and field indicators. The field indicators are: the steppes composition, perennial vegetation cover, annual perennial production, grazing level and the prevalence of rangeland cultivation. Results show that the knowledge-based approach is a valid method for evaluating rangeland degradation. The proposed knowledge-based approach discriminated between rangeland categories that would not have been discernible using only remote sensing. Overall classification accuracy of rangeland degradation obtained using this approach was 93%. This approach revealed that 11, 36 and 30% of the study area have shown very severe, severe and moderate degradation level, respectively.
4. The fourth part concerns the assessment of Alfa grass (Stipa tenacissima) tussocks at various degradation levels of Alfa grass steppes in the high plateaus of eastern Morocco, based on field hyperspectral data (350 – 2500 nm) and digital images during fall and spring seasons. Digital images of Alfa grass tussocks were taken using a digital camera to classify of the tussocks according to their proportion in green leaves (green, mixed and dry tussock). Assess software (Image Analysis Software for Plant Disease Quantification, APS 2002) was used to obtain the proportion of greenness in each tussock. Hyperspectral data of three states of tussocks (Green, Mixed and Dry tussock leaves) were collected within three degradation levels of Alfa grass steppes (Slight, Moderate and Severe degradation) with the ASD FieldSpec® 3 spectroradiometer. Paired t-test, Normalized difference spectral reflectance (NDSR) and Stepwise Discriminate Analysis were used to discriminate between various tussock status and different Alfa grass steppes. The results indicate that Alfa grass had shown an intraspecific variability in reflectance spectra. The proportion of green leaves in Alfa grass tussock strongly influences the spectral response. The discrimination of different Alfa grass tussock status was better during fall than spring. The spectral behavior of Alfa grass tussock is problematic for the mapping and the assessment of Alfa grass steppes by conventional remote sensing techniques.
