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

Accueil du site → Doctorat → Allemagne → 2015 → Assessment of Woody Biomass and Solar Energy Resources with Remote Sensing and GIS Techniques : A Regional Study in the High Mountains of the Eastern Pamirs (Tajikistan)

Universität Bayreuth (2015)

Assessment of Woody Biomass and Solar Energy Resources with Remote Sensing and GIS Techniques : A Regional Study in the High Mountains of the Eastern Pamirs (Tajikistan)

Zandler, Harald

Titre : Assessment of Woody Biomass and Solar Energy Resources with Remote Sensing and GIS Techniques : A Regional Study in the High Mountains of the Eastern Pamirs (Tajikistan)

Auteur : Zandler, Harald

Université de soutenance : Universität Bayreuth

Grade : Doctoral thesis, 2015

Résumé
Energy issues have been a main concern of geographical research in the Eastern Pamirs of Tajikistan. Dwarf shrubs (Krascheninnikovia ceratoides, Artemisia spp.), as the only woody vegetation, are of central importance in this context by representing a key thermal energy resource. But despite their relevance for sustainable development, neither an assessment of woody biomass quantities nor an evaluation of potential alternatives has been conducted. Remote sensing and GIS techniques are considered as appropriate tools to study these objectives. However, common space-borne remote sensing methods reach their limits in such arid environments characterized by scarce vegetation cover. Therefore, the main research goals of this dissertation are to evaluate and extend existing remote sensing approaches and test different sensors for woody biomass quantification in drylands to contribute to the clarification of global earth observation problems. Furthermore, related empirical results are intended to shed light on the ongoing regional degradation debate. Finally, the feasibility of photovoltaic energy as an alternative local energy resource for sustainable development should be assessed. Field data represented the basis for the study by providing spatially allocated biomass amounts using an allometric model, climate measurements, and complementary information. A large number of remote sensing variables, potentially relevant for woody biomass prediction, according to the literature, were derived from the Landsat OLI, RapidEye, EO-1 Hyperion and ASTER sensors. Several spectral variables were experimentally adapted to account for interfering background signals. Various techniques and models were applied to compare their performance in spatial biomass prediction. An interdisciplinary analysis including external survey data was used to contrast dwarf shrub availability, accessibility, and demand. An integrative study of field measurements, a spatial solar radiation model, framework scenarios, and literature based cost calculations provided the mean for an evaluation of the local photovoltaic energy potential and anticipated environmental effects. The results show that remote sensing based biomass quantification is possible even under the difficult arid conditions of the research area, but relatively high modeling errors have to be taken into account (RMSE 1000 kg/ha). Statistical models with adequate selection procedures and shrinkage techniques proved to be important in this high dimensional setting. A performance assessment demonstrated that common vegetation indices are not successful and variables adjusting for soil effects are necessary in this region. The comparison of sensors indicated that a large spectral range, comprising plant as well as background information, is advantageous in dryland vegetation modeling. The hyperspectral sensor revealed an increased potential for woody biomass prediction, with the ability to reduce the relative RMSE by a maximum of 20 percentage points compared to multispectral data. Narrowband indices, calculated from the short wave infra-red spectral domain, showed to be particularly suitable for dwarf shrub detection. A conservative biomass model enabled the comparison of available dwarf shrub stocks with harvesting amounts in a case study village by taking prediction errors and harvesting practices into account. Associated results suggest that locally, biomass quantities are sufficient to meet thermal energy demand on the medium term. However, restricted accessibility may limit future energy supply, and long-term sustainability is questionable due to the low regeneration rate of regional dwarf shrubs. The implemented spatial radiation model performed well in deriving solar energy amounts. The assessment of photovoltaic energy resources as substitutes for woody biomass showed that the generation of thermal energy is feasible within reasonable cost limits when restricted to certain basic applications. The estimations of the environmental effects of potentially increased photovoltaic infrastructure showed that it would result in a considerable mitigation of degraded areas and an amplification of carbon sequestration. This demonstrated the benefits of solar photovoltaic energy as an alternative renewable energy resource in peripheral arid high mountains. This dissertation provided contributions to the utilization of remote sensing and GIS techniques in drylands and high mountain regions. It was thereby shown that they offer valuable tools to resolve environmental research issues, but are also subject to major restrictions that require field based method adaptions. This study indicates that upcoming satellite sensors, earth observation products, and sophisticated statistical models will have much potential for regional and global research on natural resources in arid environments.

Mots clés  : Arid ; Environment ; Vegetation ; Modeling ; Landsat ; RapidEye ; Hyperion ; High dimensional ; Vegetation indices ; Hyperspectral ; Multispectral ; Solar radiation ; Central Asia ; Drylands ; Dwarf shrubs ; Space-borne ; Satellite

Présentation

Version intégrale (17 Mb)

Page publiée le 13 mars 2020