Landschafts- und Landnutzungswandel im nigerianischen Tschadbecken : qualitative und quantitative räumliche Analysen durch Auswertung multitemporaler und multisensoraler Fernerkundungsdaten im Rahmen eines regionalen geographischen Informationssystem

Landcover and landuse changes in the Nigerian Chad BasinQualitative and quantitative geographic analyses by evaluation of multi-temporal and multi-sensoral remote sensing data in the context of a regional geographic information system

Die Dissertation wurde im Rahmen des von der Deutschen Forschungsgemeinschaft geförderten interdisziplinären Sonderforschungsbereiches 268 "Kultur- und Sprachentwicklung im Natrurraum Westafrikanische Savanne" angefertig
Die Dissertation wurde im Rahmen des von der Deutschen Forschungsgemeinschaft geförderten interdisziplinären Sonderforschungsbereiches 268 "Kultur- und Sprachentwicklung im Natrurraum Westafrikanische Savanne" angefertigt. Dabei war die Kartierung und Analyse von Landbedeckungs- und Landnutzungsveränderungen in einem Ausschnitt des nigerianischen Tschadbeckens Zielsetzung vorliegender Arbeit. Trotz eines recht großen allgemeinen Interesses am Tschadsee, dem größten See Westafrikas, fällt es schwer, topographische und thematische Information über diese Region zu erlangen. Viele Zusammenhänge bezüglich seiner Vergangenheit und seiner zukünftigen Entwicklung sind bis heute nicht geklärt. Soweit Daten zur Landbedeckung und Landnutzung bereits verfügbar waren, wurden sie aufbereitet und im Rahmen des regionalen geographischen Informationssystems FirGIS strukturiert gegliedert aufgenommen. Als hauptsächliche Datenquelle wurde jedoch auf Fernerkundungsdaten der Region zurückgegriffen. Sie liefern meist die einzige großflächig vorhandene, aktuelle und vor allem multitemporale Information über das Untersuchungsgebiet. Die Auswertung der Fernerkundungsdaten erfolgte mit Methoden der digitalen Bildverarbeitung. Aktuelle und historische Luftbilder wurden visuell interpretiert und anschließend vektoriell digitalisiert. Satellitenszenen wurden vorwiegend digital direkt im Rasterdatenmodell bearbeitet. Nach der Datenaufbereitung mit verschiedenen Techniken der Bildverarbeitung erfolgte die Auswertung mittels digitaler Klassifizierung. Weitere wichtige Bearbeitungsschritte bei der Ergebnisfindung waren unterschiedliche multitemporale Darstellungsweisen, die Berechnung von Vegetationsindizes sowie Verschneidungstechniken zur Aufdeckung von Veränderungen. Die Auswertung aller Daten erfolgte nach einem einheitlichen mehrstufigen Klassifizierungssystem, das für die Verhältnisse im Tschadbecken entwickelt wurde. Die aus den Fernerkundungsdaten abgeleiteten Ergebnisse gliedern sich nach vier verschiedenen Dimensionen der Zeit. Die Ausgangsbasis bildet eine digitale aktuelle Kartierung der Landbedeckung und Landnutzung eines mehr als 3500 km² großen Ausschnittes des nigerianischen Tschadbeckens im Maßstab 1 : 25 000. Aufbauend auf der aktuellen Landnutzungskartierung wurde für einen etwa 1100 km² großen zentralen Teilausschnitt aus dem Gesamtuntersuchungsgebiet das dynamische Verhalten der Landbedeckungs- und Landnutzungseinheiten im Jahresverlauf untersucht. Dazu konnten fünf SPOT-XS-Satellitenszenen der Vegetationsperiode 1995/96 herangezogen werden. Als Zusammenfassung aus den phänologischen Analysen aller im Untersuchungsgebiet auftretenden Objekte und ihrem entsprechenden unterschiedlichen Spektralverhalten zu verschiedenen Zeitpunkten wurde eine detaillierte Landbedeckungs- und Landnutzungskartierung der Saison 1995/96 angefertigt. Sie ging aus einer komplexen hierarchischen Satellitenbildklassifizierung mit differenzierten Zuweisungsregeln hervor. Mit den Erfahrungen aus der Untersuchung zur saisonalen Dynamik wurden in einem weiteren Schritt alle für das Teiluntersuchungsgebiet zur Verfügung stehenden Satellitenszenen ausgewertet, um Erkenntnisse über interannuelle Unterschiede der zu verschiedenen Zeitpunkten aufgenommenen Satellitenszenen zu gewinnen und deren Ursachen aufzudecken. Als ideale Grundlage wurden multitemporale Vegetationsindizes verglichen, Veränderungskarten abgeleitet und Niederschlagsdaten zur Interpretation hinzugezogen. Eine 24jährige Zeitreihe der klassifizierten Vegetationsindizes ermöglicht eine zusammenfassende Interpretation. Die Auswertung historischer Luftbilder von 1957 und der Vergleich mit den Luftbildern von 1990 ermöglichte schließlich die Beurteilung langfristiger Entwicklungen im engeren Untersuchungsgebiet. Die Ergebnisse sind in Form einer historischen Karte, als Veränderungskarten und quantitativ festgehalten. Ein wichtiges Ergebnis der vorliegenden Arbeit besteht in der Strukturierung aller verwendeten Grundlagendaten zum Thema Landbedeckung und Landnutzung im Tschadbecken einschließlich seiner Nachbargebiete innerhalb des regionalen geographischen Informationssystemes FirGIS. In ihm sind neben den Ausgangsdaten alle Ergebnisse und die zum Thema gehörigen Zusatzinformationen enthalten. Es ermöglicht damit eine zukünftige Fortschreibung durch thematische, räumliche oder zeitliche Ausdehnung. Das Geoinformationssystem FirGIS bietet außerdem die Möglichkeit, bereits vorhandene wie auch zukünftige Information anderen Interessenten vor allem auch vor Ort zur Verfügung zu stellen.
show moreshow less
The mapping and analysis of landcover and landuse changes in a part of the Nigerian Chad Basin was the target of this thesis. In spite of a considerable general interest in Lake Chad, the biggest lake of West Africa, it 
The mapping and analysis of landcover and landuse changes in a part of the Nigerian Chad Basin was the target of this thesis. In spite of a considerable general interest in Lake Chad, the biggest lake of West Africa, it is difficult to obtain topographical and thematic information about this region. Until today, many interrelations regarding the past and the future of the lake remain unclear. One reason for this is the comparatively poor data situation. Most of the maps and data, particularly landuse information, are not up-to-date. As far as data or maps of landcover and landuse were already available, they were structured and integrated into the regional geographic information system FirGIS. However, remote sensing data was used as main source for the evaluations. Image data gathered from airplanes or satellites provide the only extensively available multi-temporal up-to-date information of the research area. The analysis of the different remote sensing data was carried out by methods of digital image interpretation, on the one hand by visual interpretation and on the other hand - to a larger portion - by automatic digital procedures. The procedures in detail depended on the question and particularly on the data base structure. Current and historical aerial photographs were visually interpreted and then digitised using the vector data model. In most cases digital pre-processing, such as geographic rectification, contrast enhancement and mosaicing, were applied. Satellite scenes, however, were predominantly processed directly in digital form using the raster data model. In addition to the data preparation by digital image interpretation techniques the evaluations were carried out by means of digital classification. Further important steps of investigation that have been applied include different visualisation techniques of multi-temporal data, the computing of vegetation indices as well as different geoprocessing techniques for change detection. One main conclusion in respect to methodology particularly when using multi-temporal data is that automatic procedures by themselves do not provide satisfying results. Instead, semi-automatic evaluations, which include additional visual interpretations, considerably improved the quality of the results. In addition, such an approach favours the common evaluation of data of different remote sensing platforms, such as aerial photographs and satellite scenes, as well as the comparison of atmospherically non-corrected satellite scenes of different recording dates. The classification of all data was carried out using a common multi-level classification scheme, which was specifically adapted to the conditions in the Chad Basin. The results derived from the remote sensing data were for presentation purposes subdivided according to four dimensions of time. Attachment 1 Starting with a current detailed digital map of landcover and landuse, a 3500 km² subset of the Nigerian Chad Basin is presented. The map originates in aerial photographs of a scale of 1 : 25 000 and is enclosed in this thesis as a hardcopy. The main criterion regarding landcover is the distribution of the sediments: About 40% of the investigation area is covered by sandy sediments, the remaining larger part by clayey sediments. Few small expanses of water cover less than 1%. Built-up areas - predominantly settlements - also take up about 1% of the research area. They are mainly located on the sandy substratum. Agriculturally productive land occupies 53% of the total area and can be found both on the sandy and on the clayey areas. On the latter, both, traditional dry season cultivation (12%) as well as large area irrigation cultivation (22%) are practised. The sandy sediments are used for the traditional rainy season cultivation (18%). However, only one third of the rainy season land is currently under use, whereas the larger part is left as fallow. Further agricultural activity, such as rice cultivation, river irrigation and the cultivation following the seasonally receding Lake Chad (1% all together) is practised on both of the two substrata. Only little more than a quarter of the investigation area (27%) is covered by a mainly sparse tree cover. Though the sand areas show a clearly denser tree cover (41%) than the clay areas (14%). Besides the two-dimensional elements also linear elements such as roads, pathes, rivers, non-perennial rivers and channels were mapped in detail. Figure 45 Based on the current landuse map the dynamic seasonal behaviour of the various landcover and landuse units in an approximately 1100 km² central subset of the research area was examined. For this purpose five SPOT-XS satellite scenes of the vegetation period 1995/96 were consulted. As was expected, the most essential seasonally variable influences were the factors water and vegetation. In addition bush fires which are set at different times of the year, strongly influence the spectral response. As a summary for the phenological analyses of all different objects and their corresponding spectral behaviour at different times of the year, a detailed landcover and landuse classification of the season 1995/96 was developed. The presented results were obtained from a complex hierarchical satellite image classification with differentiated assignment rules. Compared to the map made from aerial photographs, the strength of the satellite image classification lies in the mapping of two-dimensional elements like sediments, farming areas and vegetation. For these object classes a higher differentiation due to the multi-spectral information contained in the different satellite scenes is possible and therefore better results in automatic processes are delivered. The comparison of the results obtained from the two examinations, however, revealed only restricted comparability due to the different data sources and the resulting divergent approach and data model used. Figure 6 With the experiences from the examination of the seasonal dynamics, in a further step, all satellite scenes available for the research area were evaluated in order to obtain knowledge about the inter-annual differences represented in the satellite scenes recorded at different times as well as about their causes. These evaluations concentrated on vegetation changes. As an ideal basis multi-temporal vegetation indices were employed for comparison and for derivation of change detection maps and precipitation data were consulted for the interpretation. Only in rare cases inter-annual differences could be looked at isolatedly. More often, inter-annual and seasonal alterations as well as long-term oscillations are combined in the satellite scenes. Nevertheless, a summarizing interpretation in form of a time series over 24 years derived from classified vegetation indices is presented. This time series can be subdivided in three periods: During the 1970s a weak general decrease in vegetation cover has to be recorded; there are no obvious differences between the low-rainfall period at the beginning of the decade and the period at the end of the decade with rainfall at or above average. These could be symptoms for anthropogenous causes for the vegetation reduction. A clear break is represented by the satellite scenes of the 1980s. They show extremely low vegetation cover at this very low-rainfall decade. The satellite scenes of the 1990s give the impression of recently more humid conditions, which however can only be verified for the end of the decade (without satellite cover), when precipitation conditions were increasing again. Figure 52Finally, the evaluations of historical aerial photographs of 1957 enabled the analysis of long-term developments in the investigation area. The results are presented in form of a historical map (enclosed as hardcopy in this thesis), as different change maps and quantitatively in form of diagrams. As far as land cover is concerned, expanses of open water decreased to a fifth of the area of 1957 in the current maps. A drastic fall was also observed for the vegetation cover (decrease of 37% within 33 years). In this case, primarily the vegetation cover with a high (by approximately 76%) and a medium density (by approximately 42%) was decimated. The area used by man has extended clearly. In 1957 an area of 70% of the investigation area was without anthropogenous use, whereas in 1990 it decreased to 31%. The settlement area has doubled in this period; taking into account the modern plants of the irrigation project, the area is even five times greater than in 1957. Unlike for the settlements, the area used for agriculture was not extended considerably, except for the large irrigation project. The currently used area for the traditional rainy season cultivation takes 6% of the total area at both times. Only the extent of the fallow land has been tripled. The dry season acreage has been reduced by a third. Instead, the extensive irrigation cultivation, which operated succesfully only during few years, has effectively gained area. Attachment 2An important result of this thesis is the integration of all used data and information on landcover and landuse in the Chad Basin including its neighbouring areas within the regional geographic information system FirGIS. The information system contains all base data as well as all results and additional information relating to the topics mentioned above. Therefore, a future continuation by thematic, spatial or temporal extension is easily possible. In addition, the geographic information system FirGIS offers the possibility to make already obtained data as well as information gathered in future available for all interested parties especially in the research area itself.
show moreshow less

Download full text files

Export metadata

  • Export Bibtex
  • Export RIS

Additional Services

    Share in Twitter Search Google Scholar
Metadaten
Author:Ina Franke-Scharf
URN:urn:nbn:de:hebis:30-0000001551
Referee:Günter Nagel
Document Type:Doctoral Thesis
Language:German
Date of Publication (online):2003/05/19
Year of first Publication:2000
Publishing Institution:Univ.-Bibliothek Frankfurt am Main
Granting Institution:Johann Wolfgang Goethe-Univ.
Date of final exam:2001/11/22
Release Date:2003/05/19
SWD-Keyword:Fernerkundung ; Landnutzung ; Landschaftsentwicklung ; Nigeria; Tschadbecken
HeBIS PPN:10820667X
Institutes:Geographie
Dewey Decimal Classification:550 Geowissenschaften
Sammlungen:Universitätspublikationen
Licence (German):License Logo Veröffentlichungsvertrag für Publikationen

$Rev: 11761 $