F unctional data analysis methods including functional cluster analysis and functional linear modeling are discussed. The methods are used to describe and compare the shape of elevation cross-profiles taken from three Himalayan valleys. Typical methods for the analysis of these profiles are discussed in a nonlinear regression framework along with the use of model selection criteria. Curve registration is used to align important features in the profiles. Functional cluster analysis is used to group profiles by shape, with the shape based on the estimated curvature of each profile. Functional linear models are then used to explain the variability in the observed shapes of the profiles.

No particular mathematical or negotiation skills are required. This talk is open to everyone.