Combined use of airborne laser scanning and hyperspectral imaging for forest inventories (OFEV, WHFF project 2013.18)
The thesis proposes as a general objective the estimation of forest inventory parameters (e.g. trunk location, height, basal area, crown spread, species, etc..) from the combination of airborne laser scanning and hyperspectal imaging data. The results of this research are expected to be useful for the operational domain in several ways: provide tools to better survey areas which are unmonitored in field inventories (e.g. private forests, low accessibility areas), act as a decision aid to determine where field inventories are necessary and where they can be partly or totally substituted by remote sensing, simplify the visualization, quality assessment and field data integration of the produced tree models.
The overall aim is composed of two interrelated research topics: tree biophysical parameter estimation and the optimization of manual processes (quality assessment and field data input). The first research topic mainly involves the development of automatic point cloud segmentation algorithms, the addition of hyperspectral information to the segments, and the direct/indirect estimation of tree attributes from the 3D structure, spectral signature and contextual information of the segments. The second topic involves the development of interactive tools which tightly couple the automatic biophysical parameter estimation and field observation processing chains.
Methodology for energy strategies selection at the regional scale: mitigating air quality impacts while minimizing the deployment costs. A Cuban case study
This research proposes, as a general objective, the design of a methodology for selecting the most promising energy strategies with a view toward air pollution impact mitigation and deployment cost reduction. This methodology is to be applied at a regional scale and in regions with limited data access. The results are expected to be useful as a decisional aid by producing computer-aided tools in this field. The overall aim is composed of three interrelated research topics: air quality modeling, integrated assessment methodologies and process energy system design methods. The first research topic primarily involves the use of numerical tools to describe the causal relationship between emissions, meteorology, and atmospheric concentration. The second topic involves the development of tools for integrating the contribution of emission sources on air quality impacts and the implementation costs of potential emission reduction measures in order to find optimal application rates of technologies. The final topic involves the development of urban energy system design and the application of process integration and multi-objective optimization concepts.
Biogeoinformatics for the management of farm animals in Switzerland
In the context of both severe selection in farm animals and of current and potential effects of climate change, it is crucial to implement a suitable and sustainable management of the breeding practice in Switzerland, supported by a judicious use of geographic information technologies. To reduce further loss of genetic diversity and to protect and promote the diversity of farm animal genetic resources (FAnGR), this research will propose a novel monitoring approach of Swiss FAnGR including characteristics of the Swiss transhumance system (alping), whose effects will be investigated (milk production). Phenotype and genotype data, but also the location data of more than 8’000 alps from the Braunvieh cattle breeders’ association will be used as a case study and processed by a combination of population genetics and spatial statistic tools. The monitoring based on different categories of criteria (genetic, demographic, geographic, socio-economic, cryo-conservation, etc.) will enable the determination of a level of endangerment for Swiss livestock species and breeds according to FAO recommendations.
Moreover, investigations dedicated to the study of the alping system will permit to i) study lactation curves varying in time to point out phenotypes and genotypes that are worth preserving (valuable traits related to production quality and quantity) and to ii) define present and future environmental conditions showing the highest probability to constitute sustainable breeding conditions, i.e. favoring a sufficient grass quality.
To achieve these goals, the research described here comes within the scope of biogeoinformatics: biological data will help to monitor diversity and to identify key genomic regions associated with important phenotypes (e.g. fat, protein or lactose proportion in milk) and related to selection; geography will make it possible to analyse the spatial distribution of animal characteristics and to link the latters with environmental variation; and finally computer science is required to process the corresponding large volumes of data (GWAS, spatial statistics, pedigree analysis, etc.) and to integrate the different thematic criteria to facilitate decision making (e.g. breed or species prioritization).