Predictive models to estimate sediment volumes deposited by debris flows (Vargas state, Venezuela): an adjustment of multivariate statistical techniques

Debris flows are a moving mass composed from water and solids mixture, mainly in form of sediments, with a high destructive power. The debris volume that is transported and deposited outside the drainage system of a watershed has a great importance in the definition of its hydrological response. The objective of this work was to propose predictive models generated through the adjustment of multivariate statistical techniques, to estimate the sediment volumes deposited by debris flows. Measurements and calculations of the morphometric parameters of the watersheds and drainage networks have been performed with the support of GIS software and spreadsheets. The relationships between morphometric parameters and sediment volumes have been analyzed by applying multivariate statistical techniques such as linear correlation analysis. The principal component analysis and multiple linear regression analysis have been performed with principal components, which allowed the generation of predictive models. From the predictive models generated for the sediment volumes deposited by the debris flow event of December 1999, raised results closer to reality with better Pearson’s correlation coefficients from those related to the gradient and shape of the watershed relief and extension of the drainage network morphometric variables. For the estimation of deposited sediment volumes due pre- and post-1999 event conditions, only the predictive models generated with the gradient and shape of the watershed relief variable have had good results.