Ιδρυματικό Αποθετήριο [SANDBOX]
Πολυτεχνείο Κρήτης
EN  |  EL

Αναζήτηση

Πλοήγηση

Ο Χώρος μου

Development of synthetic velocity - depth damage curves using a weighted Monte Carlo method and logistic regression analysis

Vozinaki Anthi-Eirini, Karatzas Giorgos, Sibetheros Ioannis , Varouchakis Emmanouil

Πλήρης Εγγραφή


URI: http://purl.tuc.gr/dl/dias/9A803C62-51BC-4C8E-94AB-6DB3F5BC0AD1
Έτος 2014
Τύπος Πλήρης Δημοσίευση σε Συνέδριο
Άδεια Χρήσης
Λεπτομέρειες
Βιβλιογραφική Αναφορά A. E. Vozinaki, G. P. Karatzas, I. Sibetheros and E.A. Varouchakis. (2014). Development of synthetic velocity - depth damage curves using a Weighted Monte Carlo method and Logistic Regression analysis. Presented at EGU General Assembly 2014. [Online]. Available: http://www.researchgate.net/profile/Anthi-Eirini_Vozinaki/publication/262006760_Development_of_synthetic_velocity_-_depth_damage_curves_using_a_Weighted_Monte_Carlo_method_and_Logistic_Regression_analysis/links/02e7e53652508d3301000000.pdf
Εμφανίζεται στις Συλλογές

Περίληψη

Damage curves are the most significant component of the flood loss estimation models. Their development is quitecomplex. Two types of damage curves exist, historical and synthetic curves. Historical curves are developed fromhistorical loss data from actual flood events. However, due to the scarcity of historical data, synthetic damagecurves can be alternatively developed. Synthetic curves rely on the analysis of expected damage under certainhypothetical flooding conditions.A synthetic approach was developed and presented in this work for the development of damage curves, whichare subsequently used as the basic input to a flood loss estimation model. A questionnaire-based survey tookplace among practicing and research agronomists, in order to generate rural loss data based on the responders’loss estimates, for several flood condition scenarios. In addition, a similar questionnaire-based survey took placeamong building experts, i.e. civil engineers and architects, in order to generate loss data for the urban sector. Byanswering the questionnaire, the experts were in essence expressing their opinion on how damage to various croptypes or building types is related to a range of values of flood inundation parameters, such as floodwater depth andvelocity.However, the loss data compiled from the completed questionnaires were not sufficient for the construction ofworkable damage curves; to overcome this problem, a Weighted Monte Carlo method was implemented, in orderto generate extra synthetic datasets with statistical properties identical to those of the questionnaire-based data.The data generated by the Weighted Monte Carlo method were processed via Logistic Regression techniques inorder to develop accurate logistic damage curves for the rural and the urban sectors. A Python-based code wasdeveloped, which combines the Weighted Monte Carlo method and the Logistic Regression analysis into a singlecode (WMCLR Python code). Each WMCLR code execution provided a flow velocity-depth damage curve for aspecific land use. More specifically, each WMCLR code execution for the agricultural sector generated a damagecurve for a specific crop and for every month of the year, thus relating the damage to any crop with floodwaterdepth, flow velocity and the growth phase of the crop at the time of flooding. Respectively, each WMCLR codeexecution for the urban sector developed a damage curve for a specific building type, relating structural damagewith floodwater depth and velocity.Furthermore, two techno-economic models were developed in Python programming language, in order to estimatemonetary values of flood damages to the rural and the urban sector, respectively. A new Monte Carlo simulationwas performed, consisting of multiple executions of the techno-economic code, which generated multiple damagecost estimates. Each execution used the proper WMCLR simulated damage curve. The uncertainty analysis of thedamage estimates established the accuracy and reliability of the proposed methodology for the synthetic damagecurves’ development.

Υπηρεσίες

Στατιστικά