Database Concept Design

Intensive cooperation between the project partners is a prerequisite for de- signing a common research platform. This implies that each partner con- tributes his datasets and knowledge to identify research foci. For this pur- pose each HazNETH partner delegated one or more responsible persons who discussed the natural hazards to be considered as relevant for Switzer- land, as well as each partner’s research interests and suggestions on how to map hazard related phenomena in a database. The parameters relevant to monitor each natural hazard phenomenon were identified and grouped thematically and known relationships between phenomena were docu- mented.

Web Based Information System for Natural Hazard Analysis 5 Information gathered for natural hazards research is complex. Apart from managing very large interrelated datasets of different scale and spa- tial extent, important issues of combining hazards and their effects have to be solved. A well thought-out database concept and implementation will provide the basis for deriving new hazard assessment methods from intrin- sic interactions between the studied hazards. Methods of quantifying haz- ard parameters as well as the uncertainty of such methods and of the data- sets themselves have to be taken into account when designing the database concept.

2.1.1 Data Collection on Three Spatial Scales

Depending on the spatial extent and local distribution of natural hazard phenomena three spatial scales of data collection and analysis were identi- fied (Fig. 3):

• General Level: Switzerland (country)

• Regional Level: Wallis (river basin)

• Local Level: (alpine valley)

Fig. 3. Different spatial scales of data collection and analysis

The Local level (alpine valley) is targeting natural hazard phenomena occurring at a local level including landslides, torrent streams, debris flows, glaciers hazard events, etc. The datasets for this level were gener- ally sampled at a higher resolution than for the other two levels.

The Regional Level datasets were collected for an entire hydrological system, a river basin, which roughly corresponds to the administrative boundaries of Swiss Canton Wallis plus the areas that are not part of Can- ton Wallis but belong to the river basin. The research focus on the regional level is directed towards natural hazards that concern the whole river basin

6 Constantin R. Gogu, Helen Freimark, Boris Stern, Lorenz Hurni

(e.g. floods). The resolution of the collected datasets is generally lower than at the local level.

The General (country) level covers the administrative boundaries of Switzerland and to some extent the neighbouring countries taking into ac- count that natural phenomena do not respect manmade administrative boundaries. Phenomena like earthquakes or other various tectonic phe- nomena are observed and analysed on such a scale.

2.1.2 Dataset Description

The information collected in the HazNETH project consists in geological, hydrological, geomorphologic, soil, climate, land use, and anthropogenic parameters and their geometry. Topological, photogrammetric, and geo- logical information acquired from other organisations (e.g. SwissTopo, Natural Hazard Office of Canton Wallis) complement the datasets used with HazTool. The database concept requires these datasets to be organ- ised thematically.

The datasets will be available to the project partners at different stages of processing: raw data, processed data, and project generated hazard data.

Raw data is obtained through scientific measurements; processed data is received when applying different calibration and modelling procedures (process based, statistical, or empirical) of phenomena analysis to raw data. This step will be chiefly performed by each project partner with their specific modelling software. The treated datasets will then be made avail- able to the other HazNETH partners in the database. The project generated hazard data regroups the results (maps or other output types) obtained through applying various hazard quantification methods. This group of datasets may be persistently stored and not generated by HazTool on-the- fly for reasons of system performance or lack of functionality integration into the system. At the current stage the raw data was structured and has been processed to some extent.

2.1.3 Data Models

In order to set up the database several data models were analysed. The task was to design a data model that takes into account the special features of natural hazards of Switzerland, while permitting to direct the research fo- cus at the beginning of the project on alpine valleys. This was a logical consequence of the completeness and detail of available datasets of the lo- cal research area (Vispertal including Saastal und Mattertal). It was also

Web Based Information System for Natural Hazard Analysis 7 decided to direct research attention during the first two years of project du- ration primarily on the phenomena of torrent streams and debris flow.

Experience gathered during a similar project carried out at the Institute of Cartography of ETH Zurich (the GEOWARN project, Hurni et al. 2004) influenced the design of the database concept. Two other data models con- stituted sources of inspiration for the design process: ArcHydro data model developed by ESRI (Maidment 2002) for managing surface water re- sources and HYGES (Gogu et al. 2001) developed by University of Liege for managing groundwater resources.

The ESRI ArcHydro Data Model is used to restructure the base surface hydrology data to study torrent streams and debris flow phenomena. Using the ArcHydro data model with ESRI ArcGIS, it is possible to extract dif- ferent themes from hydrological data (Maidment 2002): Network (showing pathways and waterflow), Drainage (drainage areas and stream lines), Channel (three dimensional representations of river and channel shapes) and Hydrography (hydrographic features as found on topographic maps).

The ESRI ArcHydro Data Model has to be customized to the needs of the HazNETH project. Significant modifications are the adoption of a river cross section database scheme of the Swiss Federal Office for Water and Geology, the addition of a Sedimentology layer representing sediment vol- umes showing a potential to be eroded, relevant characteristics of soil composition (describing the soil layers, the ground covering, and the grain size distribution), and the addition of the Ground covering layer storing parameters like land roughness and erosion threshold.

2.2 Development of Integrated Natural Hazard Assessment