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

Accueil du site → Doctorat → Allemagne → 2016 → Monitoring dynamics of semi-arid forests with multi-sensor time series

Technische Universität Berlin (2016)

Monitoring dynamics of semi-arid forests with multi-sensor time series

Gärtner, Philipp

Titre : Monitoring dynamics of semi-arid forests with multi-sensor time series

Monitoring von semi-ariden Wäldern mit Multi-Sensor-Zeitreihen

Auteur : Gärtner, Philipp

Université de soutenance : Technische Universität Berlin

Grade : DOKTOR DER NATURWISSENSCHAFTEN - DR. RER. NAT. 2016

Résumé
Dryland degradation is a challenging environmental problem in the context of global change and China is among those countries that are most severely affected. The far-west province of Xinjiang Uyghur Autonomous Region experienced ambitious agricultural development and land reclamation projects which caused major environmental degradation, loss of forest cover and the advancement of desertification. Efforts from the Chinese government to restore the degraded floodplain ecosystem are ongoing. This dissertation aimed to enhance the monitoring of forest dynamics in the floodplains of the Tarim river in Southern Xinjiang, by examining the suitability of multisensor time series to assess forest disturbance and restoration response. The main focus was on the applicability of very high, high and medium spatial resolution satellite imagery to improve forest disturbance and forest regrowth monitoring. First, I studied the growth and decline of individual tree crowns with bitemporal change detection of very high spatial resolution satellite images (Chapter 2). Second, I investigated the dynamics of forest disturbance caused by an insect pest outbreak. It was examined whether forest disturbance maps produced with additional synthetic high resolution images would improve the accuracy of disturbance detection (Chapter 3). Finally, I used a medium spatial resolution Landsat time series to monitor trend shift dynamics of the floodplains forest, shrubland and grassland areas. A special focus lay on the spatial pattern of longitudinal and transverse linkages to known river discharges (Chapter 4). Results showed predominantly positive growth at all investigated spatial scales. At local scale, findings confirmed increased P. euphratica tree crown growth. The applied OBIA approach proved to be useful in the semi-arid forest setting, producing moderate accuracies. Forest disturbance mapping, with added synthetic scenes, improved significantly when compared to the original data set. The most important factor for the accuracy increase was the timing, rather than the number of images involved in the analysis. Images which were recorded at the end of the insect disturbance period performed best during the disturbance detection. This stage was found particularly important in distinguishing defined defoliation severity classes. Finally, the trend shift analysis showed increased rates of forest, shrub- and grassland growth in times when water deliveries were conducted. The absent of discharge had a substantial interrupting effect on the prevalence of trend shifts. Vegetation showed resilience after a drought year, with above average growth in subsequent years. Longitudinal effects, with more pronounced vegetation reactions, was found in the upper zones and less apparent reactions in the lower sections of the river catchment area. Transverse impacts showed a delayed growth response of ∼six month in areas adjacent to the river channel. This dissertation demonstrates the value of multi-sensor time series analysis for monitoring forest dynamics. The expressed findings increase knowledge and enhance understanding towards disturbance effects and response dynamics in semi-arid forest ecosystems, and shall help to improve future management decisions.

Mots clés  : forest monitoring multi-sensor remote sensing semi-arid ecosystem Tugai forest Tarim river populus euphratica Landsat WorldView2 Waldmonitoring Multi-Sensor-Fernerkundung semi-arides Ökosystem Tugai-Wald Tarim Fluss China

Présentation

Version intégrale (66,76 Mb)

Page publiée le 1er novembre 2017