Impact Factor (IF) - Thomson Reuters Web of KnowledgeSM)

2020: 1.500 - 5 years IF: 1.659

2019: 1.258 - 5 years IF: 1.610

2018: 1.152 - 5 years IF: 1.315

2017: 1.000 - 5 years IF: 1.000

2016: 0.938 - 5 years IF: 1.010

2015: 0.641 - 5 years IF: 0.673

2014: 0.628 - 5 years IF: 0.652

2013: 0.390 - 5 years IF: 0.504

2012: 0.605

2011: 0.468

2010: 0.309

2009: 0.136

An international Journal published under the auspices of:

Recognized by:

Supplements of Geografia Fisica e Dinamica Quaternaria
Volume VII - 2005


Proceedings of the Conferences “Mountains and Plains”


L’uso di tecniche di classificazione non supervisionata per la mappatura automatizzata di forme a partire da modelli digitali del terreno – Unsupervised classification techniques for automated mapping of landform elements from digital elevation models

Pages 35-40


We report the preliminary results of a methodology for the automated mapping of some landform features. O ur investigation has still to be refined and extended to a wider area to give reliable results for the study of the dissection al and tectonic histo ry of the study region, but it is already possible to discuss some advances we have achieved in automatic landform classification for the study area. The Trigno River catchment was selected to test the met ho d. In the study area, due to the almos t complete lack of river terraces, the reconstruc tion of old stages of valleys development must be attempted using othe r characters of th e valley-side slopes. Among others, the widespread, gentle ero sional glacis that over hang the pr esent rhalwegs at different elevat ions can be used as indicators of discontinuous downcut ting and as proxy of ancie nt base levels. Most the methods that are currentl y used to characterize terrain features by means of DEMs extract the information contained in each cell of the raster and are able to quantify the local surface geometry very accura tely (within the resoluti on limit of the DEM used ). However, to make discriminations among land forms which differ in terms of spa tial and contex tual prop rieties, it is essential to take into account also the properties of neighbour cells. In this sense, we propose a procedure that , instead of using a classic neighb ourhood analysis, partitio ns th e DEM by submi tti ng slope angle and elevation value to a k-rneans algorithm for the identi fication of homogeneous landform elements . In thi s way, we expect to obtain patches of cont iguous cells whose boundaries reflect chara cteristic landform eleme nts of the stu dy area. Mea n values of (i) slope angle, (ii) tangenti al curvat ure and (iii) profile curva ture of the cells belonging to each pa tch, are then used as «fea tu re vectors» for an unsupervised landform classification of the area under examination. For the Trig no River test area a lO-cluster solution has been chosen by a tria l and error procedure. The resulti ng automated mapping is judged very good as it recogni zes ten different un its, each of them match s very well with ones of the geomorphological un its th at have been surveye d in the test area. A comparison with a cellby-cell unsupervised classification shows that the proposed method produces more accura te results. In particular a good discrimin ation between summit erosio nal surfac es and valley bott om were achieved. Furt her de – velopments, aimed to the refineme nt of the classificat ion result s, are also briefly discussed.

→ Download Abstract PDF

→ Download Full Text PDF