Information technologies for floods monitoring and short-term forecasting

CHALLENGE DESCRIPTION Creating and wide utilizing flood monitoring and forecasting systems is a very important issue when making decisions on loss prevention activities or planning the development of territories. The frequency and the severity of the effects of such emergencies are still high. For example, in Russia, 40–70 large floods take place every year. Roshydromet (http://www.meteorf.ru/) data show that this type of natural emergency affects about 500000 sq km; and those with catastrophic effects, 150000 sq km. These areas contain about 300 populated towns and cities, tens of thousands populated localities, many economic facilities, and more than 7 million ha of agricultural fields.
Flood monitoring and forecasting systems are based on models and methods that can be conditionally combined into several large groups.
Firstly, satellite monitoring has been one of the most important methods so far. Although the benefits of this approach are widely known, it is far from being flawless from the decision-making point of view when planning territory development or loss prevention activities. Most notably, it provides post-event mapping and defining boundaries of an inundation areas as well as only post-event damage analysis.
A second group of well-known methods are medium-term and long-term forecasting for a period of several weeks to several months. These methods are based on hydrological modelling and require large amount of initial data, such as characteristics of the snow cover, soil properties, meteorological parameters, etc. Naturally, medium-term and long-term forecasting gives valuable information which allows analysing possible scenarios of probable flooding for following few weeks or months. Nevertheless, the low reliability of the initial data and the large time interval of forecasting do not provide accurate estimates of the flooding time at each specific point of the river valley.
In order to provide high accuracy of floods forecasting the special kind of information systems should be developed and utilized – short-term forecasting systems. Recently, a number of new systems and services have been developed, for example:

• Copernicus Emergency Management Service, which includes Mapping service (Copernicus Emergency Management Service, Mapping) and Global Flood Awareness System (GloFAS);
• Thematic Exploitation Platform – Hydrology (TEP Hydrology) (European Space Agency), which includes the Flood Monitoring Service.
  However, the existing services often do not involve the use of mathematical models that most adequately describe features of rivers with rare network of gauge station and are intended mainly for the flow monitoring and forecasting tasks, and not for short-term forecasting of river floods.
  

Information systems using mathematical models include: Flood Early Warning Systems (FEWS) (Deltares), North American National Water Model (NWM) (NOAA National Water Center), European Flood Awareness System (EFAS). These systems are focused on the territories with well-developed network of gauge stations and observation posts – sources of hydrological and meteorological data. To forecast situations on remote rivers in rural areas, it is necessary to take into account such features as sparseness of hydrometeorological observations network, the occurrence of ice jams, the absence of highly detailed digital elevation models, etc.
Besides, there exists a lack of information systems adapted for a user who doesn’t have special skills in information technologies, in hydrology and data processing.

DESCRIPTION OF THE SOLUTION Suggested approach includes utilizing a set of information technologies and software modules, which provide creation of fully automatic and use-friendly systems for river floods short-term forecasting. Main features of the approach are as follows:

• utilization of a complex of hydrological and hydrodynamic models with adjustment of model parameters to provide high accuracy of forecasting (different models can be used: LISFLOOD, MIKE, STREAM-2D, ECOMAG, etc – most suitable for a particular river and catchment);
• gathering and combining diverse remote sensing and other data (both spatial and non-spatial) including data on water stream and runoff parameters from gauging stations and meteo information; data on territory and infrastructure, crowdsourcing data, satellite images, etc.;
• maximum automation up to fully automatic operation of the system – starting from gathering initial data, through modelling to interpreting results, visualizing, damages assessment, reports preparation, and alerting interested parties;
• easiness for any user who should not have special knowledge and skills in information technologies, hydrology and data processing.
• Information system that implements the proposed approach contains the following main components:
• Input Data Collection and Pre-processing components to provide input data collection (e.g., satellite images, on-line hydrological data from meteorological stations; a digital terrain model) and pre-processing (i.e., primarily image processing, information filtering and information fusion).
• Automatic modelling components to forecast of water levels, calculation and forecasting of the water discharge, depths of flooding, and also automatic modelling of the water spread using hydrodynamic models.
• Post-Processing, publishing and visualization components to provide post-processing of modelling results, output data storage in a database, flood map vectorization and visualization of inundation areas.
• Distribution components to provide dissemination of the modelling results to external systems, publishing short-term forecasts at the public GeoPortal, and automatic notification of the local citizens and organizations using web services and mobile personal devices;
• User-friendly interface to provide using the system not only by specialists with a high level of knowledge in the field of GIS and information technology, but also by all other users interested in the results of flood forecasting (emergency services, executive authorities, commercial organizations and citizens).

The system operates automatically and provides flood forecasts for the following 12–48 hours with hourly outlines of the potential flooded zones, objects, and a water depth map. The interface screen also displays the current hydrological situation, data from gauging stations, as well as satellite images. The flood forecasting results are provided as a remote web service. Moreover, the users are not required to have specific knowledge in modelling and simulation or to have programming skills.
Two additional useful functions are implemented in the system. First, every user can move the timeline slider and see the inundated area 12 hours in advance, all the complexity is hidden from the user. The second additional function also allows the user to see the flood area for every possible water level. Finally, automatic generation and analysis of flooding scenarios allow studying of the river floods dynamics and to evaluate their potential effects in the near future to support preventive actions in order to mitigate the floods impacts.
Software of the system is based on service-oriented architecture. All system components are implemented as web services and can be geographically distributed and localized in various organizations, cities and countries.
The considered system, built using the proposed information technologies, can operate in three main modes: continuous hydrological monitoring, short-term floods forecasting during dangerous periods, and a scenario mode for modeling possible hazardous situations in advance.

POSSIBILITY OF ADAPTING THE SOLUTION The proposed set of information technologies and software was used and adapted to create operational flood forecasting systems on the Daugava (Latvia) in 2013 and Northern Dvina (Russia) rivers in 2016 when there were intensive floods. The systems were also used for hydrological monitoring and scenario modeling.
The systems operated in automatic mode, and in both cases the discrepancy between the forecast data on flooded areas and current data obtained from satellite images was not more than 7%. The forecast accuracy, assessed by composition of infrastructure in the flooded zone, was at least 90%.
Case studies results were actively used by local governments, hydrometeorological and emergency services in scenario modeling and operational flood forecasting, as well as for the current analysis of the hydrological situation.
In general, the kit of proposed information technologies may be considered as a quite versatile constructor to create and apply information systems for monitoring and short-term forecasting river floods with detailed consideration of the features of a specific catchment area and riverbed.
For the proposed solution to be successfully adapted, it is advisable to involve hydrologists to select and preset models for the analyzed catchment area.