Exploration de la fusion des informations pour améliorer la fiabilité locale d'une carte forestière
|Authors:||Orzanco, Maria Gabriela|
|Advisor:||Lowell, Kim; Fortin, Marie-Josée|
|Abstract:||In the traditional forest maps in Quebec, sharp lines represent stand boundaries, forest strata are considered equally concordant in relation to field/cartographic attributes and the woody volume estimations issue from successive regrouping of field plots. The general objective of this research is to explore how different methods can be combined to characterize stand boundaries, forest strata and woody volume estimations, from geomatic and ecological points of view, by map and field information fusion. In order to qualify boundaries, their contrast and spatial contexts are quantified. The integration of these aspects allows the detection of forest sectors where boundary identification could be easiest. A significant relationship exists among the contrast of examined variables with the existence probability of boundaries. Other boundaries are identified from the field plots. These ones correspond to homogeneous plot clusters and to strong spatial discontinuities in the field variables. The two boundary types coincide spatially when neighbourhood differences are intense. The woody volume boundaries are, in this scale, spatially dependent on forest and topographic boundaries. Regression trees are used for grouping field plots considering minimum volume variability. The most effective tree model is associated with field variables. The groups formed with this method present less spatial variability in comparison with the plot groups presently used in Quebec. The forest strata are qualified by their accord map/plot level. For many forest strata the local accord is high while the neighbourhood accord is low. The local accord is attributed to internal strata homogeneity and/or the fact that plots poorly represent the forest stratum. The neighbourhood accord is associated with positional uncertainty of field samples, of boundaries or both. Finally, in order to permit cartographic and field boundary fusion at the element level, the resolution of heterogeneity between field and forest map data must be addressed first. The spatial resolution and semantic ecological significance of cartographic units are the most important of these heterogeneities.|
|Document Type:||Thèse de doctorat|
|Open Access Date:||12 April 2018|
|Collection:||Thèses et mémoires|
All documents in CorpusUL are protected by Copyright Act of Canada.