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

Accueil du site → Doctorat → États-Unis → 1995 → Using kriging to predict distribution of arid vegetation, with discussion of co-kriging field data and satellite imagery

New Mexico State University (1995)

Using kriging to predict distribution of arid vegetation, with discussion of co-kriging field data and satellite imagery

Horton, Jane Dorothy

Titre : Using kriging to predict distribution of arid vegetation, with discussion of co-kriging field data and satellite imagery

Auteur : Horton, Jane Dorothy

Université de soutenance : New Mexico State University

Grade : Doctor of Philosophy (PhD) 1995

Résumé
The regionalized variable, vegetation canopy class, was kriged to predict distribution of selected plant species. A 5000 ha site representative of the northern Chihuahuan Desert was selected to test prediction of spatial distribution of vegetation. This method predicts based on the variation over increasing linear distances separating pairs of observations of a variable. Estimates of canopy class of a plant species at unsampled sites were weighted by the spatial variation of canopy class of sampled sites. Canopy class was measured with a modified Daubenmire (1970) method with randomly located transects 53 and 47, respectively, in 2 pastures of the USDA-ARS Jornada Experimental Range, Las Cruces, New Mexico. For all transects, 100 Daubenmire plots, 50 x 50 cm, were placed and canopy for all live species occurring in the quadrats was classified. Classes were 0, 1-5, 6-25, 26-50, 51-75, 76-95, and 96-100%. Of the 56 species identified, 18 species were selected for analysis, based on their presence in 30 or more plots. Semi-variograms were calculated for these 18 species per pasture. Out of a potential of 36 models, 19 were identified. Some plant species, such as black grama (Bouteloua eriopoda), snakeweed (Gutierrezia sarothrae), and honey mesquite (Prosopis glandulosa) occurred in both pastures but exhibited different parameter estimates for each pasture. These differences may be driven by plant growth form, seedling establishment requirements, historic utilization patterns by herbivores, climate patterns and (or) edaphic characteristics. Results show that kriging can predict vegetation distribution. Co-kriging LandSat Thematic Mapper image and vegetation data was confounded by scale and year differences in data sets. Ordinary kriging successfully predicted canopy classes of dominant species. Ordinary kriging should be applicable for predicting vegetation distribution in similar landscapes, when applying the following guidelines : a preliminary nested survey to reveal general spatial variation needed for designing representative sampling methods, estimation of cover class for only 4-6 dominant species, estimation of cover class for bare ground, match sampling scale with that of remotely-sensed data (for co-kriging analysis).

Mots clés : Ecology, Applied sciences, Remote sensing, Biological sciences, Range management, Biostatistics

Annonce (NMSU Library)

Accès au document : Proquest Dissertations & Theses

Page publiée le 11 mars 2015, mise à jour le 20 octobre 2017