A web-framework for rapid flood event analysis in Germany – Kai Schröter & Stefan Lüdtke
A web-framework for rapid flood event analysis in Germany – Kai Schröter & Stefan Lüdtke
0 0
gisday
On Github sluedtke / gisday
A web-framework for rapid flood event analysis in Germany
Kai Schröter & Stefan Lüdtke
19 November 2014
Background
- CEDIM Research focus on Forensic Disaster Analysis in Near Real Time
- Rapid flood event analysis in Germany
- large-scale floods that affect multiple river catchments
- e.g. in June 2013
- In depth analysis of hazard and impact
Capture, evaluate and assess current flood situation
Focus today: Tool to capture and evaluate flood hazard: 'Flood Event Explorer'
Challenge
- capture and store data from online gauges for current flood situation
- heterogeneous online sources (16 federal states + 1 federal agency)
- evaluate current observations in long term context
- flood frequency analysis for times series 50 years of records
- visualisation
- comparison to similar? historic events
- in short: taming bulky space time data
Requirements
- data changes on a regular basis
- large data sets where only some observations are needed
- share huge data set among working groug (CEDIM task force)
- rapid queries
- web interfaces to data, especially dynamic data
Implementation
PostgreSQL
- PostgreSQL is a powerful, open source object-relational database system
- > 15 years of active development and a proven architecture
- runs on all major operating systems, including Linux, UNIX and many more ..
- It also supports storage of binary large objects, including pictures, sounds, or video
- diverse native programming interfaces (C/C++, Java, .Net, Perl, Python, ... )
- comprehensive documentation
PostGIS
- spatial extension to PostgresSQL
- offers full GIS functionality
- support for all common spatial data types (vectors, rasters, ...)
Key figures for CEDIM flood data-base
- 488 gauges
- mean daily discharge values
- mean daily water levels
- for available observation period (mostly longer than 50 years)
- > 17 Mio rows
- summary statistics for time series
- extreme value statistics for different reference periods using 9 different distributions
- 9 different return periods pre-processed
- sub-basin boundaries
- river network
Daily workflow
Application
example period
8/14/2014 - 09/05/2014 for return periods of 2, 5 and 10 years
Outlook
... or things that we want to implement next ...
- automatically identify similar historic events (region, magnitude, impact,...)
- inundation mapping
- loss estimation
for the curious among you
the most recent version is available under
within GFZ