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

Accueil du site → Doctorat → Pays-Bas → Tree transpiration mapping from upscaled sap flow in the Botswana Kalahari

University of Twente (2009)

Tree transpiration mapping from upscaled sap flow in the Botswana Kalahari

Chavarro-Rincón, Diana

Titre : Tree transpiration mapping from upscaled sap flow in the Botswana Kalahari

Auteur : Chavarro-Rincón, Diana

Université de soutenance : University of Twente

Grade : Doctor 2009

In arid and semi-arid environments such as the Kalahari, where groundwater represents the main source of water supply, water extraction by vegetation from unsaturated and saturated soil must be carefully identified. Previous studies in the Kalahari have shown that groundwater can be discharged from more than 60m depth in the form of transpiration by deep rooted vegetation. Investigations carried out by the Botswana Government concluded that most of the tree species in the eastern Kalahari make predominant use of soil water from depths of more than 3m, i.e. below the root zone of shrubs and grasses. These findings shed light on how certain species remain green during the dry season, but they also raised concerns on the significance of tree transpiration compared to the potential groundwater recharge, thus highlighting the importance of tree transpiration mapping.
In this study, a methodology for the mapping of tree transpiration based on the upscaling of sap flow measurements during the dry season is presented. For that purpose, nine Kalahari tree species namely, Acacia fleckii, Acacia erioloba, Acacia luederitzii, Boscia albitrunca, Lonchorcarpus nelsii, Terminalia sericea, Burkea Africana, Ochna pulchra and Dichrostachys cinerea were investigated. The selection of these species was based on their frequency of occurrence in the study area.The methodology is presented in two major steps : i) sap flow investigations and ii) upscaling and mapping. The former was based on ground measurements and follow-up laboratory experiments, and the latter on GIS and RS techniques. This approach is considered adequate for savanna ecosystems in which different species coexist, as it allows the identification of species-dependent transpiration dynamics based on the examination of inherent sap flow patterns. The method was developed for the particular conditions of the Kalahari but it can be used in its general form in other environments.
In the first step, sap flow measurements were acquired using thermal dissipation probes (TDP), in which the effect of natural thermal gradients (NTG) was analyzed. Complementary, the issue of conductive sapwood area determination was reviewed at the light of laboratory experiments using X-ray tomography and nuclear magnetic resonance, NMR. Sap flow was calculated as the product of sap flux densities measured with TDP, and sapwood area in groups of 18 to 24 specimens per species. The obtained sap flow patterns combined with micro-climatic indicators suggested unusual water-use habits e.g. nocturnal sap flow, downwards stem flow and air moisture harvesting, in four out of the nine species investigated. Such events had to be considered before the upscaling process.
In the second step, mean sap flow of each measured tree was correlated directly or indirectly with a RS-measurable parameter, namely canopy area (CA), for the establishment of species-specific upscaling functions. Five multi-spectral airborne images at three different resolutions, together with an IKONOS satellite image of the study area were classified using object-oriented approach to allow species differentiation. Accuracy assessment of the classification determined the best resolution in which tree species can be discriminated. Species-specific upscaling functions based on the correlation between CA and sap flow were applied to the classified images, resulting in transpiration maps.
The results showed that among the nine measured species, the presence of the evergreen Boscia albitrunca defines areas of high water extraction. The total dry-season transpiration flux in the five images ranged from 0.021 mm/day for the image with predominant occurrence of Dichrostachys cinerea and Terminalia sericea, to 0.165 mm/day for the image where Boscia albitrunca was abundant. Transpiration fluxes obtained from the RS-based upscaling approach yielded more realistic values than energy balance methods previously applied in the same area and during the same season. The results presented in this study showed that the combination of ground measurements, laboratory experiments and GIS/RS procedures constitutes a robust approach for the mapping of natural processes such as tree transpiration.


Version intégrale (61,5 MB)

Page publiée le 22 mars 2011, mise à jour le 25 janvier 2018