GEOSCAN, résultats de la recherche


TitreMind the scale! Modeling at multiple scales to predict the distribution of sediment grain size for use in benthic habitat mapping
TéléchargerTéléchargement (publication entière)
AuteurMisiuk, B; Lecours, V
SourceProgram and abstracts: 2017 GeoHab Conference, Dartmouth, Nova Scotia, Canada; par Todd, B J; Brown, C J; Lacharité, M; Gazzola, V; McCormack, E; Commission géologique du Canada, Dossier public 8295, 2017 p. 86, (Accès ouvert)
LiensGeoHab 2017
ÉditeurRessources naturelles Canada
Réunion2017 GeoHab: Marine Geological and Biological Habitat Mapping; Dartmouth, NS; CA; mai 1-4, 2017
Documentdossier public
Mediaen ligne; numérique
Référence reliéeCette publication est contenue dans Todd, B J; Brown, C J; Lacharité, M; Gazzola, V; McCormack, E; (2017). Program and abstracts: 2017 GeoHab Conference, Dartmouth, Nova Scotia, Canada, Commission géologique du Canada, Dossier public 8295
Sujetstechniques de cartographie; océanographie; milieux marins; études côtières; conservation; organismes marins; écologie marine; gestion des ressources; peuplements biologiques; etudes de l'environnement; écosystèmes; benthos; établissement de modèles; sédiments marins; repartition granulométrique; boues; sables; graviers; échantillons prélevés au hasard; levés géophysiques; levés acoustiques marins; levés au sonar; sonar latéral; analyses statistiques; biologie; géologie marine; géologie des dépôts meubles/géomorphologie; géologie de l'environnement; sédimentologie; géophysique
ProgrammeGéoscience en mer, Géoscience de la gestion des océans
Diffusé2017 09 26
Résumé(disponible en anglais seulement)
Sediment grain size is an important habitat-defining parameter for many benthic species. Grain size distribution is affected by a number of oceanographic and sedimentological processes that operate at a variety of scales. Modeled predictions of sediment grain size can aid in benthic habitat mapping, yet the scale of processes is seldom considered in modeling, despite mounting evidence indicating the scale-dependent nature of these processes. The consideration of spatial scale can increase predictive accuracy of distribution models, and aid in understanding of the processes that drive distributions.
Sediment grain size data from 98 grab samples collected from around Qikiqtarjuaq, Nunavut, Canada were used to train distribution models of mud, sand, and gravel fractions for subsequent use in habitat mapping and sediment classifications. Sixteen predictor variables derived from multibeam echosounding data were resampled at eight different resolutions (using the calculate terrain variable then average result over n x n window method; Dolan, 2012), which were tested for statistical importance in predicting mud, sand, and gravel grain size fraction distributions. Optimal resolutions were determined for each predictor variable, which were subsequently used to train compositional distribution models of mud, sand, and gravel grain size fractions in a Boosted Regression Tree model, using a methodology similar to that of Diesing (2015). Results demonstrated that the default multibeam data resolution (5 m) was often not the optimal choice for predictor variables, with coarser resolutions often explaining more statistical deviance. This reinforces the importance of considering or testing multiple spatial scales in distribution modeling. Identifying variable-specific scales aided in understanding the drivers of sediment grain size distribution in the area around Qikiqtarjuaq, and resulted in a high 10-fold cross-validated predictive accuracy (Spearman correlations of rho-mud=.772, rho-sand=.712, rho-gravel=.578). Predictions of mud, sand, and gravel grain size fractions were subsequently combined into a Folk (1954) grain size classification, and were used in species distribution models to support conservation and management of marine resource development around Qikiqtarjuaq, Nunavut, Canada.