Modelling Dynamic Products of Natural Disasters with the OGC Observations & Measurements Format
Peyrega, Charles; Ruiloba, Rosario
CS Information Systems, FRANCE

The latest Earth Observation missions will produce a huge amount of data, which will help in watching our planet carefully. In the framework of the current European spatial projects handling natural disasters such as GMES (Global Monitoring for Environment and Security) and SAFER (Services and Applications For Emergency Response), these big data must be properly formatted and processed in order to extract added value information necessary to define protection programs for the population. In this context, our work focused on the definition of a data format to represent dynamic products of disasters.

A product of disaster is an heterogeneous added value data set gathering several pieces of information about a disaster, like its location, its date, the concerned geographical zones. There are several ways to represent such features, for example, it is possible to generate a geolocated satellite image as background on which several layers (rasters or vectors) are superimposed to illustrate the concerned zones. Moreover, the metadata like the date of the disaster, the methods used for the processing, the responsible agencies, or the satellite used for the observation, can be represented too. Finally, the term "dynamic" for a product of disaster stands for two concepts in our work, first it characterizes the temporal variations of the added value data, and secondly, their variations with the different parameter sets used during the processing of raw data to generate them.

Since the latest spatial missions such as Sentinel and Pleiades are highly driven by interoperability between agencies, the data generated should follow these rules as well. For example, both European projects HMA (Heterogeneous Mission Accessibility) and INSPIRE (Infrastructure for Spatial Information in Europe) share a common objective: harmonizing and promoting interoperability of data for ground segments and for data users of heterogenous Earth Observation missions. For that purpose, the idea consists in providing standards and formats to manipulate these data. Our work focused on the standards of the OGC (Open Geospatial Consortium), which are widely used in the community of Earth Observation. Among these formats, four of them are especially suitable to represent features of interest of dynamic products of disasters: Observations & Measurements (O&M) (ISO 19156), which can embed other OGC formats like SensorML, GML (Geographic Markup Language) and KML (Keyhole Markup Language).

Based on the abstract concept of "observation", the OGC O&M 2.0 standard is a very complete format to represent dynamic products of disaster. Thus, each O&M product is handled as an observation, to which it is possible to associate heterogeneous added value data, such as the date and the location of the disaster, the procedure used to extract the features of interest from raw data (i.e. the geolocated concerned zones). The metadata associated to the product can be directly embedded into it according to the ISO 19115 framework and with respect to the INSPIRE recommandations. The procedure used to generate the geolocated concerned zones can be described with the sensorML format, which can be embedded in the O&M observation object too. The sensorML standard is part of the OGC SWE (Sensor Web Enablement) framework and is used to represent chains of geolocated sensors or chains of processing units, and to precisely characterize their inputs, outputs, parameters and connections. The geolocated concerned zones can be modelled as GML geometries or as external KML files. Several reasons let O&M be an excellent standard to generate dynamic products: first, it is based on OGC standards and it is an OGC and ISO standard itself, which lets it be suitable for interoperability, then, it is autodescriptive through the use of references to internal and external objects (thanks to both gml:id and xlink:href mechanisms). Finally, O&M is very complete and compatible with other OGC standards, which makes it very useful to represent all the added value data and metadata of dynamic products of disasters. However, O&M is such a complete and multipurpose format, that the data embedded in it can not be directly visualized. On the contrary, a specific solution should be developped depending on the application. Thus, O&M is more adapted for data exchange.

The last step of our work focused on global methods to generate such O&M dynamic products of disasters, and to extract added value data, in order to help countries in preventing and reducing the impact of future damages on hazardous areas. For that purpose, we propose a solution based on OGC web services, especially based on WPS servers (Web Processing Service) which works in two steps. The first one consists in generating O&M databases of dynamic products from raw data (images, vector data, other O&M products), and the second one extracts added value information from O&M dynamic products for visualization, exploitation and concrete decision making.