The Extreme Rainfall Detection System (ERDS) provides near real time alerts related to potential exceptional rainfalls at river basin scale, with almost worldwide extent, and it can be used by humanitarian assistance organizations to evaluate the events and to understand the potentially floodable areas in places where their assistance is needed.
Stream line and drainage basin layers used for the analysis. The river basin data derive from HYDRO1k geographic database, produced by U.S.Geological Survey. Rainfalls conditions are monitored at basin scale
The system is based on a precipitation analysis carried out by means of global rainfall satellite data processing. Specifically, rainfall conditions are monitored using data from Tropical Rainfall Measuring Mission (TRMM) Multisatellite Precipitation Analysis (TMPA).
The TMPA rainfall estimation chosen is performed at a spatial resolution of 0.25x0.25 degrees, available in a 'historical' version covering from 1998 to the delayed present (3B42 v6) and in a real time version from 2002 to present (3B42RT v6).
Depth Duration Frequency (DDF) curves with different return periods are represented in green, cumulated rainfalls are represented in red and critical rainfall, detected through the comparison with historical values, is indicated with a black dot.
The historical TMPA series (product 3B42 v6), is analyzed using a hydrological method based on Depth Duration Frequency (DDF) curves. The outputs are:
- the creation of a complete historical database of heavy rainfalls with nearly-global coverage and single river basin resolution;
- the estimation of thresholds for each basin, to be used for detecting extreme rainfall events in near real time, classified by levels of criticality.
The near real time data (3B42RT product, available every 3 hours, about 6 hours after acquisition) are compared to the output of the historical analysis, in order to monitor current rainfall conditions and to provide alerts in near real time.
Oracle 10g has been used to load and elaborate the whole dataset of rainfall that was structured in a data warehouse architecture. The capability of monitoring the whole set of basins at the same time is achieved by a grid computing system, consisting of a combination of computer resources used to parallelize the computational process.
Example of the output map: the alerted basins are shown with 3 different possible colours, according to the severity of rainfalls.
The web interface, which was developed by extending OpenLayers functionalities, aims to offer an easy access to the data and also to map the basins that are potentially affected by flood. A tool for zooming at basin level and for activating informative layers extracted from WFP SDI (Spatial Data Infrastructure) is accessible to users.
A fully operational beta version is already available for UN WFP and other humanitarian actors. Future developments include the implementation of an automated procedure for producing flood extent scenarios; the methodology aims at identifying correlations between rainfall events parameters and their effects on the ground, derived with remote sensing techniques. A statistical analysis on past events would allow to link current warnings with a suitable and reliable scenario.
