{"id":39306,"date":"2026-08-12T01:24:07","date_gmt":"2026-08-12T00:24:07","guid":{"rendered":"https:\/\/www.vtei.cz\/?p=39306"},"modified":"2026-08-12T09:57:56","modified_gmt":"2026-08-12T08:57:56","slug":"linking-critical-points-to-the-flash-flood-indicator","status":"publish","type":"post","link":"https:\/\/www.vtei.cz\/en\/2026\/08\/linking-critical-points-to-the-flash-flood-indicator\/","title":{"rendered":"Linking critical points to the flash flood indicator"},"content":{"rendered":"

ABSTRACT<\/h2>\n
This paper addresses the issue of refining the estimation of flash flood hazard by modifying the Flash Flood Indicator (FFI) application, which is used by the flood forecasting service of the Czech Hydrometeorological Institute (CHMI). The existing FFI approach was based primarily on hazard assessment determined by comparing the runoff response with threshold values derived from theoretical discharge rates with a given return period, while exposure and vulnerability were considered only implicitly. The article proposes extending the methodology by explicit consideration of these factors by means of two main approaches: (i) adjusting the threshold values for hazardous runoff depending on the degree of urbanisation of the area, and (ii) incorporating critical points (locations where concentrated runoff may enter urban areas) and their contributing catchment areas in the FFI calculations.<\/div>\n
Based on analysis of selected rainfall events from 2023\u20132024, it was found that lowering the threshold values in urban areas leads to a more realistic estimate of the risk of localised flooding, particularly within built-up areas of municipalities. At the same time, it was shown that the inclusion of critical points \u2013 defined as locations of concentrated runoff at the interface between built-up and non-built-up areas \u2013 significantly improves the identification of locations with an increased risk that are not adequately captured by the 3 \u00d7 3 km polygon grid. The results also confirm the crucial influence of the quality of meteorological data input \u2013 particularly radar precipitation estimates \u2013 on the accuracy of the resulting risk assessment. The proposed modifications represent a shift from assessing the hazard itself to a more comprehensive expression of actual risk, which takes into account the spatial variability of exposure and vulnerability. The article also identifies the main limitations of the current approach and gives recommendations for further development of the FFI system, particularly in terms of refining meteorological input and adaptive setting of threshold values.<\/div>\n

INTRODUCTION<\/h2>\n
Flash floods can cause significant material damage and loss of life. They occur as a result of rapid water runoff triggered by intense rainfall, usually confined to a small area. Precipitation over an area typically up to a few km2, lasting a matter of minutes, does not usually trigger a flash flood, but merely leads to rapidly forming surface runoff from the surrounding land. This can cause localised flooding of low-lying areas and buildings. However, if torrential rainfall persists for a longer period and affects a larger area, surface runoff concentrates in the valleys of permanent watercourses; and if several catchment areas of smaller watercourses are affected simultaneously, this leads to flash flooding at their confluence, which may also affect settlements situated further downstream.<\/div>\n
The media has been focusing on the occurrence of flash floods in various parts of the world and their catastrophic consequences. In Europe, notable examples include the floods of July 2021 in Germany on the River Ahr (180 fatalities) and those of October 2024 near Valencia in Spain in the Turia river basin (232 fatalities). The high death toll of the flooding in Germany was due to local underestimation of the situation: despite the weather forecasts being more or less accurate, the hydrological response was not adequately assessed, which resulted in a lack of essential warnings of extreme flooding [1]. In Spain, the main problem arose from late warnings by local authorities and, more generally, the overall lack of preparedness of the emergency response system, infrastructure and the population [2].<\/div>\n
In the Czech Republic (CR), torrential flooding of significant geographical extent and severity has also recently occurred, for example on the evening of 7th June 2020 in the Uni\u010dov region and parts of the \u0160umperk region [3], and on the evening of 29th June 2022 in the Po\u0161umav\u00ed region in the Blanice catchment area on the Zlat\u00fd potok watercourse [4]. The peak flows of both floods reached a return period of approximately 50 years.<\/div>\n
Early warning of flash flooding is often very difficult. Neither the resolution of meteorological models, nor the complex physics of the atmosphere during convective storms allow for a precise determination of the location of rainfall of hazardous intensity. The probability of a successful forecast is higher for extreme rainfall over a larger geographical area. However, meteorological models predict levels for certain parameters (e.g. CAPE \u2013 convective available potential energy), from which it is possible to deduce a general risk of torrential rainfall occurring in a given (larger) area. Forecast uncertainty often persists even on the day of the expected event. The variability of individual out-puts from forecast models is high, which complicates spatial precision of meteorological or hydrological warnings. Soil saturation from previous rainfall is one of the key factors affecting runoff; i.e. higher levels of soil saturation lower the threshold for potentially hazardous intensity of rainfall.<\/div>\n
For a more accurate prediction of flash flooding, it is therefore necessary to use data from meteorological radars. Based on the current development of water-bearing clouds in time and space, the propagation of these clouds is calculated over a time frame of approximately 30 minutes to an hour. This short-term forecast, which calculates the subsequent position of these rain clouds within a given time based on their movement and upper-air flow over the most recent time interval, is known as nowcasting. The nowcasting method currently used by the CHMI projects the movement of water-bearing clouds forward in time without any change to their structure (rain intensity), which is currently the greatest limitation of this short-term forecasting method.<\/div>\n
Information on current soil moisture levels and the latest developments in water-bearing cloud cover is utilised by the Flash Flood Indicator (FFI), which is used for short-term prediction of the risk of localised flooding and the occurrence of flash floods. This tool has been in use within the CHMI forecasting service for\u00a0several years and has undergone significant development since 2017 [5]. Experience gained from the operation of the FFI in 2022, and to some extent in 2023, is analysed and discussed in this article [4].<\/div>\n
Until now, the hazard of flash floods and localised flooding has not been closely linked to \u2018risk\u2019 within the FFI system, which, in addition to hazard, also includes elements of exposure and vulnerability. One of the objectives of the research project, on which this article is based, was to incorporate elements of exposure and vulnerability in the FFI system, in particular by utilising the location of newly identified \u201ccritical points\u201d and their contributing areas throughout the Czech Republic. Critical points (CP), defined as intersections of concentrated runoff paths and boundaries of built-up areas, represent locations where the effects of hazard posed by torrential rainfall are concentrated. At the same time, we can speak of risk (damage, loss of life and harm to population health) within urban areas.<\/div>\n
This article demonstrates the current capabilities of the FFI system in predicting torrential flooding, using selected flood events as examples. In the case of localised flooding, the article examines options for adjusting parameters to determine critical thresholds for hazardous runoff and the use of critical points to emphasize factors of vulnerability and exposure within FFI calculations. Through re-simulation of selected events, the inclusion of adjusted critical runoff thresholds and a layer of critical points is evaluated.<\/div>\n

METHODOLOGY<\/h2>\n

Risk estimation in the Flash Flood Indicator<\/h3>\n
Risk is most commonly expressed as the probability of an undesirable event occurring, and of adverse impact on life, health, property or the environment. Generally, such risk arises from the interaction of hazard, vulnerability and exposure, i.e. the duration for which the hazard is present.<\/div>\n
<\/div>\n
<\/div>\n
\"\"<\/a><\/div>\n
<\/div>\n
The longer the duration of exposure \u2013 that is, the longer the period during which an object is exposed to a hazard and the greater its vulnerability \u2013 the greater the overall risk.<\/div>\n
From a perspective of probability, risk is defined as an n<\/em>-tuple of vectors [6]:<\/div>\n
<\/div>\n
<\/div>\n
\"\"<\/a><\/div>\n
<\/div>\n
where:<\/div>\n
<\/div>\n
Sc\u00a0 \u00a0 \u00a0 \u00a0 is\u00a0 \u00a0 \u00a0 \u00a0hazard scenario<\/div>\n
P\u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 the probability of the hazard scenario occurring<\/div>\n
D\u00a0 \u00a0 \u00a0 \u00a0 \u00a0are\u00a0 \u00a0 \u00a0the consequences (damage, loss), expressed in appropriate units<\/div>\n
<\/div>\n
All these quantities are time-dependent, as the hazard scenario, the probability of its occurrence and the damage may all change over time. In accordance with the general definition given above, risk can be defined as a continuous function according to the following equation:<\/div>\n
<\/div>\n
<\/div>\n
\"\"<\/a><\/div>\n
<\/div>\n
where:<\/div>\n
<\/div>\n
D(u|O)\u00a0 \u00a0 is\u00a0 \u00a0 \u00a0the risk function<\/div>\n
u\u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0an event quantifier dependent on the vector of measures and decisions<\/div>\n
O\u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 the height of the protective embankment along the watercourse, the capacity of the spillway, timely evacuation, etc., which influence the consequences of an extreme event.<\/div>\n
<\/div>\n
The function g(u) represents the probability density of the occurrence of an event (hazard scenario) with consequences.<\/div>\n
A simple conceptual model based on the equation described above (1) was applied to express risk within the tool \u2013 the Flash Flood Indicator. In the context of flash floods and to enhance the clarity of the text, individual components of risk are defined as follows:<\/div>\n
<\/div>\n