This document summarizes all the outcome of the three years PhD programme at the University of Pavia. As suggested by the title, the focus is on mutual combination of remote sensing and crowdsourcing for vulnerability estimation and damage assessment. In particular, collected pre-event information should be used to integrate and improve seismic damage extraction from satellite data. 2015 was the year of the Sendai framework for Disaster Risk Reduction (2015-2030). In this framework, the proposed work contributes in the enhancement of disaster prevention by producing relevant risk information and in the enhancement of effectiveness, meaning a quick response in the aftermath of the event. Chapter 1 gives a brief introduction to all the concepts included in the document (e.g. risk, vulnerability) and the phases of the disaster cycle. Datasets used are also described, including spectral and spatial features. Chapter 2 is related to information extracted before a disaster. In particular, the focus here is on the contribution provided by remote sensing. First, a list of indicators derived during the SENSUM project and linked to vulnerability is presented. Next, a set of algorithms is proposed in order to process satellite data and extract those indicators. Built-Up Area extraction is one of the main topics and different methods are proposed; the underlying assumption is that, due to different environmental and spectral conditions, it is really difficult to define a one-size-fits-all method capable of considering all the possible variations involved. A tailored and improved version of the algorithm is also available as service in the ESA GPOD system and directly linked with Landsat 8 and Sentinel-2 repositories. Another proposed workflow takes care of the results obtained from each single processed year and produces a map of the evolution in time of the area of interest. Other algorithms are also explained, focusing for example on the extraction of building footprints and their density, therefore requiring very high resolution optical data as input. Crowdsourcing is addressed in chapter 3 where a generic framework for the collection of geo-tagged reports is proposed. The basic idea is that the great diffusion of smartphones has created a large and dense network of observers; data provided by volunteers can actively integrate what is derived from remote sensing. Two different examples are proposed. The first one, called SEGUICI Vegetation report, was created in the framework of the SEGUICI project to collect data related to different crops and their stage of growth in order to feed a water consumption and requirements model. The app is released on the Google Play store. The second mobile app, called CLOOPSy (Copernicus Land cOver crOwdsourcing Platform for Sentinel-based mapping) has been developed in the framework of the MyGEOSS contest and is designed to collect data related to land cover for validation purposes. The product is available from the Google Play store and soon it will be available on the Apple App store. Chapter 4 gives an insight on the methods developed to extract seismic damage from SAR imagery. In particular, two different approaches have been investigated: the first one uses post-event-only VHR SAR data combined with pre-event information. The other one is based on change detection, a widely used technique to extract damage from different sensors; different attempts were made in order to determine the sensitivity to different spatial resolutions and sensors. The combination with pre-event data has also been defined and tested. Conclusions are illustrated in chapter 5.
|Titolo:||Stima della vulnerabilità e valutazione del danno tramite combinazione di telerilevamento e crowdsourcing|
|Data di pubblicazione:||22-feb-2017|
|Appare nelle tipologie:||8.01 Tesi di dottorato|