Remote sensing and geostatistics for the assessment of spatial distribution of savannah woody species biodiversity for upscaling transpiration in Serowe, Botswana

G.B. Demisse, Y.A. Hussin, I.C. van Duren, M. Lubczynski, O.T. Obakeng

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

Abstract

Relatively little attention is currently given to the issue of biodiversity loss in savannah ecosystems. Only a limited range of methods have been developed for the assessment and management of this issue. The aim of this study was to contribute to a better understanding of dry savannah woody species biodiversity and to test if species biodiversity assessment can be performed using remote sensing and geostatistics. The research was implemented in an experimental site in Serowe area, Botswana. A total of 169 sample plots were collected and analyzed. An importance value index (IVI), which is the summation of relative frequency, relative density and relative basal area, was used to select the dominant species of the area. The suitability of IKONOS satellite imagery and geostatistical techniques for savannah woody species biodiversity assessment were tested. A total of 33 woody species, which belong to 25 genera and 18 families, were identified. Two vegetation types were identified as: dense savannah (SAD) and open savannah (SAO). The trees stem density was significantly higher (t-test, t=-4.39, d.f=167, p<0.001) in SAD than in SAO. According to the IVI criteria, the dominant species in the area were: Terminalia sericea, Dichrostachys cinerea, Ochna pulchra, Acacia fleckii, Boscia albitrunca, Burkea africana and Grewia retinervis (with IVI value of 65, 64.7, 44, 37, 22.7, 12.9 and 11.5, respectively). Low correlation was observed between IKONOS imagery data and tree canopy cover percentage of the area. The dominant species were identified on pan-sharpened IKONOS imagery with reasonable classification accuracy. Three kriging methods, universal kriging (UK), simple kriging (SK) and ordinary kriging (OK), were tested for mapping of the savannah woody species. In the case of UK, the trend surface model explained only 14% (p<0.001) of the variability in woody species density prediction and UK was found less applicable. The comparison of SK and OK showed that OK was more accurate and practical for savannah woody species mapping. This research revealed that both remote sensing and geostatistical techniques can be applied for savannah woody species biodiversity assessment.
Original languageEnglish
Title of host publicationAARSE 2006
Subtitle of host publicationProceeding of the 6th AARSE International Conference on Earth Observation & Geoinformation Sciences in Support of Africa’s Development, 30 October - 2 November 2006, Cairo, Egypt
Place of PublicationCairo, Egypt
PublisherNational Authority for Remote Sensing and Space Science (NARSS)
Number of pages8
ISBN (Print)1-920-01710-0
Publication statusPublished - 2006
Event6th African Association of Remote Sensing of the Environment (AARSE) Conference 2006: Earth Observation & Geoinformation Sciences in Support of Africa’s Development - Cairo, Egypt
Duration: 30 Oct 20062 Nov 2006
Conference number: 6

Conference

Conference6th African Association of Remote Sensing of the Environment (AARSE) Conference 2006
Country/TerritoryEgypt
CityCairo
Period30/10/062/11/06

Keywords

  • ADLIB-ART-1359
  • NRS
  • WRS

Fingerprint

Dive into the research topics of 'Remote sensing and geostatistics for the assessment of spatial distribution of savannah woody species biodiversity for upscaling transpiration in Serowe, Botswana'. Together they form a unique fingerprint.

Cite this