A fully Automated Multi-Scale Flood Monitoring System based on MODIS and TerraSAR-X Data
Martinis, Sandro1; Twele, André1; Kersten, Jens1; Eberle, Jonas2; Strobl, Christian1; Stein, Enrico1

Flooding is the most frequent and widespread natural hazard in the world. The near-real time provision of detailed information about the inundation extent and its spatio-temporal evolution is essential to support crisis management activities. Satellite remote sensing based Earth observation is the most effective technology for monitoring disasters in a temporally repetitive fashion at various spatial scales. Especially in regions with missing or only sparsely available in-situ measurements, which are relevant for hydrologic modeling purposes, the utilization of remote sensing is most worthwhile.
In this contribution, an automatic system for a multi-scale monitoring of flood situations is presented. This is a result of combining two fully automated flood services, which are currently under development at DLR's (German Aerospace Center) Center for Satellite based Crisis Information (ZKI).
The continuous monitoring of the Earth to detect inundations at a high revisit interval can be accomplished by using medium resolution satellite data. Within this context a flood service based on data of the Moderate Resolution Imaging Spectroradiometer (MODIS) on NASA's Terra and Aqua satellites is being developed which provides information about the disaster extent in even large-scale watersheds on a national to continental scale (spatial resolution 250m). Based on a flood alert derived from the MODIS flood service or on other information sources a fully automated flood service based on DLR's Synthetic Aperture Radar satellite TerraSAR-X (spatial resolution 1-16 m) can be triggered on demand to derive more details about the flooding on local to regional scales and at an improved spatial resolution. The alert service monitors the database with flooded areas derived from the MODIS flood service. When data from a new MODIS scene is inserted, the size of every flooded area is being checked against a defined value. In case the flooded area is larger than this value, an alert is triggered for the acquisition of TerraSAR-X data. The delivery of newly acquired data then invokes a further web processing service to accomplish the TerraSAR-X processing and flood mask derivation.
The fully automatic processing chains of both services including preprocessing of the optical and radar satellite data, computation and adaption of global auxiliary data (digital elevation models, topographic slope information, reference water masks), unsupervised initialization of the classification as well as post-classification refinement are activated after an automatic transfer of the delivered data to a local directory using an ftp pull (see fig. 1). The further processing is based on a framework of Web Processing Services standard-compliant to the Open Geospatial Consortium (OGC). These services are integrated in processing chains for pre-processing and the designation of inundated areas. The final flood masks are exported in raster format and are visualized on a web-based user interface.
Examples drawn from various flood situations all over the world are presented. Additionally, the robustness and effectiveness of the proposed flood mapping services and their adaption to upcoming satellite systems such as the Sentinel missions are discussed.
Figure. 1: Workflow of combining fully automatic processing chains for a multi-scale flood monitoring using MODIS and TerraSAR-X data