The gum arabic belt in Sudan plays a significant role in environmental, social and economical aspects. The belt has suffered from deforestation and degradation due to natural hazards and human activities. This research was conducted in North Kordofan State, which is affected by modifications in conditions and composition of vegetation cover trends in the gum arabic belt as in the rest of the Sahelian Sudan zone. The application of remote sensing, geographical information system and satellites imageries with multi-temporal and spatial analysis of land use land cover provides the land managers with current and improved data for the purposes of effective management of natural resources in the gum arabic belt. This research investigated the possibility of identification, monitoring and mapping of the land use land cover changes and dynamics in the gum arabic belt during the last 35 years. Also a newly approach of object-based classification was applied for image classification. Additionally, the study elaborated the integration of conventional forest inventory with satellite imagery for Acacia senegal stands. The study used imageries from different satellites (Landsat and ASTER) and multi-temporal dates (MSS 1972, TM 1985, ETM+ 1999 and ASTER 2007) acquired in dry season (November). The imageries were geo-referenced and radiometrically corrected by using ENVI-FLAASH software. Image classification (pixel-based and object-based), post-classification change detection, 2x2 and 3x3 pixel windows and accuracy assessment were applied. A total of 47 field samples were inventoried for Acacia senegal tree’s variables in Elhemmaria forest. Three areas were selected and distributed along the gum arabic belt. Regression method analysis was applied to study the relationship between forest attributes and the ASTER imagery. Application of multi-temporal remote sensing data in gum arabic belt demonstrated successfully the identification and mapping of land use land cover into five main classes. Also NDVI categorisation provided a consistent method for land use land cover stratification and mapping. Forest dominated by Acacia senegal class was separated covering an area of 21% and 24% in the year 2007 for areas A and B, respectively. The land use land cover structure in the gum arabic belt has obvious changes and reciprocal conversions between the classes indicating the trends and conditions caused by the human interventions as well as ecological impacts on Acacia senegal trees. The study revealed a drastic loss of Acacia senegal cover by 25% during the period of 1972 to 2007.The results of the study revealed to a significant correlation (p ≤ 0.05) between the ASTER bands (VNIR) and vegetation indices (NDVI, SAVI, RVI) with stand density, volume, crown area and basal area of Acacia senegal trees. The derived 2x2 and 3x3 pixel windows methods successfully extracted the spectral reflectance of Acacia senegal trees from ASTER imagery. Four equations were developed and could be widely used and applied for monitoring the stand density, volume, basal area and crown area of Acacia senegal trees in the gum arabic belt considering the similarity between the selected areas. The pixel-based approach performed slightly better than the object-based approach in land use land cover classification in the gum arabic belt. The study come out with some valuable recommendations and comments which could contribute positively in using remotely sensed imagery and GIS techniques to explore management tools of Acacia senegal stands in order to maintain the tree component in the farming and the land use systems in the gum arabic belt.
Identifer | oai:union.ndltd.org:DRESDEN/oai:qucosa.de:bsz:14-qucosa-67657 |
Date | 18 April 2011 |
Creators | Adam, Hassan Elnour |
Contributors | Technische Universität Dresden, Fakultät Forst-, Geo- und Hydrowissenschaften, Prof. Elmar Csaplovics, Prof. Elmar Csaplovics, Prof. Michael Köhl |
Publisher | Saechsische Landesbibliothek- Staats- und Universitaetsbibliothek Dresden |
Source Sets | Hochschulschriftenserver (HSSS) der SLUB Dresden |
Language | English |
Detected Language | English |
Type | doc-type:doctoralThesis |
Format | application/pdf |
Page generated in 0.0018 seconds