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

Accueil du site → Doctorat → Autriche → Image analysis as a methodological tool for soil erosion studies

Universität für Bodenkultur Wien (2015)

Image analysis as a methodological tool for soil erosion studies

Bauer, Thomas

Titre : Image analysis as a methodological tool for soil erosion studies

Auteur : Bauer, Thomas

Université de soutenance : Universität für Bodenkultur Wien

Grade : Doctor rerum naturalium technicarum (Dr. nat. tech.) 2015

Résumé
This cumulated doctoral thesis deals with the development and application of two methods for the estimation of commonly used parameters in soil erosion studies. An object-based image analysis method for soil cover estimation by residues and living plants on arable land was developed. The classification algorithm is not implicitly based on the information of single pixels. It rather uses objects resulting of several pixels with similar characteristics to analyse the percent cover of different soil cover classes (residues, plants, stones, soil). This method was calibrated under laboratory conditions and validated during field campaigns. The validation demonstrated that the object-based soil cover estimation method is faster, better reproducible and the accuracy is comparable to a manual method. The generation of digital soil surface models by using photogrammetry was developed as a method to analyse soil surface conditions. After elaboration of the methodological approach and proving accuracy of the method, different cameras and settings were tested. Furthermore the calculation of several soil surface roughness parameters was carried out. The developed photogrammetric technique for soil surface estimation was later utilized for analysing soil volume, soil volume change and bulk densities. 37 replicates and a resulting coefficient of determination of 0.98 for a linear regression (y =-0.15 +0.97x) verified the usability of the photogrammetric method for soil volume detection. The results of the photogrammetric measured bulk densities were compared to the excavation method and the core method. The photogrammetric bulk density estimation was able to produce accurate soil bulk densities for several soil textures. Even for soils with high stone content comparable results were obtained. As a next step soil surface roughness changes and soil consolidation driven by rainfall were photogrammetrically monitored for different field sites and long term management practices. Soil physical properties and different management tools influence initial soil surface roughness. Long-term conventional and reduced management practices can be distinguished by comparing results of the photogrammetric soil consolidation estimation. Fertilizer applications over long periods (mineral, organic, no fertilizer) did not lead to a different soil consolidation or soil roughness. The introduced methods allow a simple and accurate estimation of parameters that are frequently used in soil erosion studies

Mots clés : Bodenerosion / Bodenbedeckung / Bodenrauigkeit / Bodensetzung / Rohdichte / Photogrammetrie / Langzeitversuch / Bodenoberfläche — Soil erosion / soil cover estimation / soil surface estimation / soil consolidation / bulk density estimation / photogrammetry / soil roughness change

Présentation (österreichischen Bibliothekenverbundes)

Version intégrale (13,4 Mb)

Page publiée le 19 novembre 2015, mise à jour le 13 mars 2019