Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Master → Autriche → Deduction of vegetation parameters from remote sensing for the integration into a regional SVAT (soil-vegetation-atmosphere transport) model in central Northern Namibia

University of Vienna (2012)

Deduction of vegetation parameters from remote sensing for the integration into a regional SVAT (soil-vegetation-atmosphere transport) model in central Northern Namibia

Mayr, Manuel

Titre : Deduction of vegetation parameters from remote sensing for the integration into a regional SVAT (soil-vegetation-atmosphere transport) model in central Northern Namibia.

Auteur : Mayr, Manuel

Université de soutenance : University of Vienna

Grade : Master of Science (MS) 2012

Résumé
In Owamboland, Namibia, the resources available to humans are limited due to climatic and hydrological conditions. Prolonged population growth further increases the pressure on resources like water and arable land. This is most notably expressed by changes in land use and land cover (LULC), or the transformation of vegetation. The study, therefore, aims to qualitatively and quantitatively assess the present distribution of vegetation in a pilot area. Data on LULC, species composition and leaf area index (LAI) from gap-fraction were collected at the end of the dry season in 2010. Techniques of remote sensing were applied using high-resolution RapidEye data to spatially aggregate LULC and LAI data. The supervised LULC classification using a Maximum-Likelihood Classifier (MLC) shows an overall accuracy of 89.53% (Khat statistics=85.41%) revealing high fragmentation and patchy structure of the study area. An empirical approach was used to upscale in situ LAI measurements. The Difference Vegetation Index (DVI) was found to perform best with in situ LAI using a linear model (R²=0.712). The modelling results reveal mean true LAI to be 0.582 (±0.254). Especially for sparsely vegetated sites, the model overestimates LAI compared to MODIS LAI (RMSE=0.47), and to in situ LAI (mean=0.277 (±0.051)) by about 128%. Model uncertainties can be attributed to soil background contamination, spectral insignificance of senescent vegetation and a uniform model for all vegetation types. The present study highlights significant LULC changes in the research area and points to the limits of the optical spectrum when applied to senescent vegetation in sparsely vegetated ecosystems

Mots clés : leaf area index / LAI / empirical model / RapidEye / land use / land cover / LULC / vegetation / Namibia / Owamboland — Blattflächenindex / LAI / empirische Modellierung / RapidEye / Landnutzung / Landbedeckung / Vegetation / Namibia / Owamboland

Présentation

Page publiée le 16 décembre 2015, mise à jour le 15 mars 2019