Time series analysis of high resolution remote sensing data to assess degradation of vegetation cover of the island of Socotra (Yemen)
The island of Socotra has long been in geographical isolation, hence nearly 30% of the plant species are believed to be endemic to the island. Until the end of 20th century there was only very little and incomplete information and literature about the vegetation on the island. This isolation broke down in 1990 with the country unification in which then the island received much attention. Subsequently the scientific knowledge of the local flora slowly increased, but many of plant species are now reported to be confined into small populations, hence being particularly vulnerable to habitat loss, overgrazing, as well as urban expansion. 1. The overall objective of this research attempted to assess and examine the trends of vegetation changes since 1972 to 2010 with the use of Landsat MSS, TM and ETM+ images and to investigate the related driving factors, such as rainfall, grazing pressure changes and underlying spatial variability of the landscape. This is to answer the overall question: Is there a trend in biomass, cover and species composition on Socotra Island over the last 40 years? If so, is that trend associated with the rainfall patterns? What are the drivers behind the vegetation change? And then how can we define changes in patterns or changes in this study area? 2. From a methodological point of view, our approach of systematically using remote sensing technology data proved scientifically an applicable tool to improve our understanding of the spatial complexity and heterogeneity of the vegetation cover as well as to provide a conceptual method with specific data for monitoring the changes over this time period. Our data obtained from these different Landsat sensors during the study period were - after many sophisticated processing steps - essentially able to provide time series information for Normalized Difference Vegetation Index (NDVI) data and to assess the long term trend in vegetation cover in the island. 3. Moreover, our approach combining supervised maximum-likelihood and unsupervised classification with the pre- and the post-classification approaches besides the knowledge based classification was table to provide sufficient results to distinguish and to map nine (9) terrestrial vegetation cover classes. The overall accuracy (compared with ground truth data) was about 91%, 77%, 70% and 72% for the images 2005, 1994, 1984 and 1972 respectively. Consecutively, the GIS analysis allowed estimates of highly valuable information as absolute areas and relative coverage of particular vegetation classes over the island with their spatial distribution and also their ecological requirements. Analysis of climatic conditions and NDVI 4. As a results of the complex topography of the study area and the wide climate range, with the guidance of prior knowledge of functional relationships between site parameters, ecosystem and the specific form of biological production, our work resulted in a division of the entire area into six variously sized ecosystem units, which were enough to properly depict the spatial heterogeneity of the rainfall and vegetation and to assist reflecting the influence and reaction between environmental parameters as well as it might have significance both for development of resources and for conservation of environment.
Use and reproduction:
All rights reserved