COST Action: IC0602
STSM title: Risk analysis models for landslide management
Reference : ECOST--IC0602-100610-000000
STSM dates: from 14-06-2010 to 27-06-2010
Location: School of Telecom Engineering
Universidad Rey Juan Carlos, Madrid, SpainThe Stsm in risk analysis.
During my time spent at the King Juan Carlos University, I was supervised by Professor David Rios Insua, doing work on a probabilistic approach to risk analysis for landslide management to be applied in the Daunia Apennine area in Southern Italy. This study refers to environmental risk analysis. Landslides constitute a major threat in this area and, indeed, in many other parts of the world.
We have sketched a methodology which includes three phases:
During this visit we have mainly focused on the first stage using methods of logistic regression, and multinomial regression analysis to predict the probability of a given type of landslide over a given planning period. With an exploratory view, we have undertaken first a classical statistics approach, with the aid of SPSS. The logistic regression models have worked well, but not that much the multinomial regression ones, which require further work.
We have also started a Bayesian approach with the aid of WINBUGS. The models are now running and we are studying convergence and fit issues at the moment to undertake then forecasting tasks.
We have also started to sketch a paper about this work, which we hope to finish at the end of September. This will be a part of my PhD thesis at Politecnico di Bari to be completed by the end of this year.
STSM title: Risk analysis models for landslide management
Reference : ECOST--IC0602-100610-000000
STSM dates: from 14-06-2010 to 27-06-2010
Location: School of Telecom Engineering
Universidad Rey Juan Carlos, Madrid, SpainThe Stsm in risk analysis.
During my time spent at the King Juan Carlos University, I was supervised by Professor David Rios Insua, doing work on a probabilistic approach to risk analysis for landslide management to be applied in the Daunia Apennine area in Southern Italy. This study refers to environmental risk analysis. Landslides constitute a major threat in this area and, indeed, in many other parts of the world.
We have sketched a methodology which includes three phases:
- the prediction of landslide susceptibility
- the prediction of landslide impact
- the choice of optimal prevention policies.
During this visit we have mainly focused on the first stage using methods of logistic regression, and multinomial regression analysis to predict the probability of a given type of landslide over a given planning period. With an exploratory view, we have undertaken first a classical statistics approach, with the aid of SPSS. The logistic regression models have worked well, but not that much the multinomial regression ones, which require further work.
We have also started a Bayesian approach with the aid of WINBUGS. The models are now running and we are studying convergence and fit issues at the moment to undertake then forecasting tasks.
We have also started to sketch a paper about this work, which we hope to finish at the end of September. This will be a part of my PhD thesis at Politecnico di Bari to be completed by the end of this year.