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

Accueil du site → Doctorat → Allemagne → Eine raum-zeitliche Modellierung der Kohlenstoffbilanz mit Fernerkundungsdaten auf regionaler Ebene in Westafrika

Universität Würzburg (2011)

Eine raum-zeitliche Modellierung der Kohlenstoffbilanz mit Fernerkundungsdaten auf regionaler Ebene in Westafrika

Machwitz Miriam

Titre : Eine raum-zeitliche Modellierung der Kohlenstoffbilanz mit Fernerkundungsdaten auf regionaler Ebene in Westafrika

Spatio-temporal modelling of the carbon budget in West Africa with remote sensing data on a regional scale

Auteur : Machwitz Miriam

Université de soutenance : Universität Würzburg

Grade : Doctoral Thesis 2011

Global warming associated with climate change is one of the greatest challenges of today’s world. One regulating factor of CO2 exchange with the atmosphere is the vegetation cover. Measurements of land cover changes in combination with modeling of the carbon balance can therefore contribute to determining temporal variations of CO2 sources and sinks, which is an essential necessity of existing and prospective political instruments like CDM (Clean Development Mechanism) or REDD (Reducing Emissions from Deforestation and Degradation). The need for quantifiable terrestrial carbon stocks is especially high for regions, where rates of land cover transformation and population density are high and knowledge on vegetation productivity is low. One region which is characterized by these criteria is West Africa. Therefore, carbon stocks in this region are seriously endangered by land cover change like the expansion of agriculture and forest logging. Large areas already act as carbon sources on a yearly basis instead of their previous function as carbon sink. Since only a few studies have analyzed the terrestrial carbon stocks in Africa and especially regional analysis in West Africa are missing, the following study focuses on regional scale modeling of the actual terrestrial carbon stocks. Additionally, the potential carbon stocks of unmanaged land cover and the potential for CO2 payments have been analyzed in this work. To quantify and assess carbon fluxes as well as the loss of carbon, net primary productivity of vegetation has been modeled, based on the plants characteristics to fix carbon from the atmosphere during photosynthesis. Modeling vegetation dynamics and net primary productivity has been realized by using MODIS 250m time series for semi-humid and semi-arid savanna ecosystems in West Africa. This study aimed to quantify CO2 exchanges of the Savanna regions in the Volta basin by applying and adapting the Regional Biomass Model (RBM). The RBM was developed by Jochen Richters (2005) at a resolution of 1000m for the Namibian Kaokoveld. In this study the model was optimized to the scale of 232m to consider the heterogeneous landscape in West Africa (RBM+). New input parameters with higher accuracies and resolution were generated instead of using the global standard products. The most important parameters for the modeling are FPAR and the fractional cover of herbaceous and woody vegetation. To enhance the MODIS FPAR product, linear interpolation and downscaling algorithms were applied. The main objective of the downscaling is a better representation of the finely scattered vegetation by the 232m resolution FPAR. The second optimized parameter, the fractional cover of herbaceous and woody vegetation was represented by the Vegetation Continuous Fields product (VCF) from MODIS in the originally version of the RBM. This global product reflects the vegetation structure of West Africa poorly, since few high resolution training data is available for this region, and the dynamic savanna vegetation can hardly be classified by not regionally adapted methods. Additionally, the data is only available with 500m resolution. Therefore, in this study a new product with 232m resolution was developed which represents the spatial heterogeneity well and, due to the regional adaptation, shows higher accuracies. The percentage cover of woody and herbaceous vegetation and bare soil on 232m MODIS data was calculated in a multi scale approach. Based on very high resolution data, represented by Quickbird and Ikonos with 0.6-4m resolution, and high resolution data from Landsat with 30m resolution, the percentage coverage was estimated for representative focus regions. These classifications were used as a training data set to determine the percentage coverage on the 232m scale with MODIS time series for the whole study region. Based on these optimized and adapted input parameters, the net primary productivity was modeled. Data from a meteorological station and an Eddy-Covariance-Flux allowed a detailed validation of the input parameters and of the model results. The model led to good results as it only overestimated the net primary productivity for the two analyzed years 2005 and 2006 by 8.8 and 2.0 %, respectively. The second aim of the study was an analysis of the potential for long term terrestrial carbon sinks. Classifications of the actual and of the potential land cover were calculated for this analysis. Considering the overall long time CO2 fixation behavior of trees, which depends on their age, longterm carbon stocks for 100 years were simulated. As carbon fixing could be paid by emission trading, which is in future depending on the political Post-Kyoto programs, potential alternative income was calculated with different price scenarios for the three countries. A comparison with the gross domestic products of these countries and with developing aid, showed the significance of CO2 trading in this region

Mots clés  : Carbon dioxide capture and storage ; Fernerkundung ; Klimaänderung ; Volta-Gebiet — Fernerkundung ; Klimawandel ; Kohlenstoffmodellierung ; Nettoprimärproduktion Carbon modelling ; climate change ; net primary productivity ; remote sensing


Version intégrale (144 Mb)

Page publiée le 22 janvier 2016, mise à jour le 5 janvier 2019