Early warning system for flash floods in the Krkonoše Mts

TESAŘ M. & ŠÍR M. 2013: Early warning system for flash floods in the Krkonoše Mts. Opera Corcontica 50/S: 107–112.

Early warning system for flash floods in the Krkonoše Mts Systém včasné výstrahy před bleskovými povodněmi v Krkonoších MIROSLAV TESAŘ & MILOSLAV ŠÍR Academy of Sciences of the Czech Republic, Institute of Hydrodynamics, Pod Paťankou 30/5, 166 12 Praha 6, CZ, [email protected], [email protected]

Abstract An early warning system for flash flood detection has been built in the Giant Mts (Krkonoše in Czech, Karkonosze in Polish). This system is located in the upper part of the Úpa River basin. The closing profile of this basin is situated in Horní Maršov (568–1,602 m a.s.l., 100 km2). Forecasting of flood risk is based on SAC-SMA software, which uses measured hydrological and hydrometeorological quantities (rain intensity and total, discharge, soil moisture, tensiometric pressure of the soil water, air and soil temperature) as input data. Keywords: early warning system, flash flood, surface and soil water monitoring

Abstrakt V Krkonoších se buduje systém včasné výstrahy před bleskovými povodněmi. Systém je umístěn v horní části povodí Úpy. Uzávěrový profil povodí se nachází v Horním Maršově 568–1 602 m n. m., 100 km2). Předpověd rizika povodně je založena na programu SAC-SMA. Tento program užívá jako vstupní data hydrologické a meteorologické údaje, měřené na povodí (intenzita a úhrn deště, tenzometrický tlak půdní vody, teplota vzduchu a půdy). Klíčová slova: systém včasné výstrahy, blesková povodeň, monitoring povrchové a půdní vody

Introduction A number of municipalities in the Czech Republic are situated in areas with a significant flood risk (MINISTERSTVO ŽIVOTNÍHO PROSTŘEDÍ 2011). Municipalities in the foothills and mountain areas are especially affected by flash floods. Flash floods affect very small catchments with a basin area of tens of km2. Every year there are 60–100 flash floods in the Czech Republic, some of them with disastrous consequences. They occur exclusively during the summer, usually when low pressure systems are followed by cold fronts (ŠTEKL et al. 2001) and precipitation of high intensity, typically more than 30 mm/hour, can be encountered (ČEKAL et al. 2012). Flash floods cause sudden surface and subsurface runoff (KAKOS 1978), sometimes accompanied by soil water leaching. Flash floods represent a challenge for forecasting and detection because they are not always caused simply by meteorological phenomena. Flash floods

result when specific meteorological and hydrological conditions exist together (NOAA 2010). Flash flood generation is influenced by these factors: • hydrometeorological influences: rain intensity and total (most important), streamflow; • hydrological influences: soil moisture (very important), soil cover hydrophobicity (which also affects the rate of infiltration), and soil profile – actual retention capacity (very important); • basin influences: shape and size, slope, roughness, stream density. In accordance with these findings, the flood forecasting service in the Czech Republic is based on: • continuous monitoring of hydrometeorological influences: among others, precipitation and discharge (KOCMAN et al. 2011); • evaluation of hydrological influences at daily intervals: assessment of the “indicator of flash floods” (ČEKAL et al. 2012), which also reflects water retention in the landscape;

108

OPERA CORCONTICA 50/S 2013

• preliminary flood risk assessment based on the evaluation of basin influences (MINISTERSTVO ŽIVOTNÍHO PROSTŘEDÍ 2011). In some cases, flash floods can be predicted from the synoptic situation, but is it impossible to predict the exact time and place of these events (ČEKAL et al. 2012). Therefore the rain forecast has a lead time of only tens of minutes before the flood. Such a short time does not allow enough time to save residential and industrial buildings but is usually sufficient for the rapid evacuation of people from the affected area. This is significantly different from the floods which are caused by regional rain (VAŠKŮ 2009). In this case, the rain forecast is quite reliable in terms of time and place. It has a lead time of two to four days (ČEKAL et al. 2012), which is sufficient to implement large-scale protective measures (construction of flood walls, etc.) and proper evacuation of people (MINISTERSTVO ŽIVOTNÍHO PROSTŘEDÍ 2011). Efforts to achieve greater lead times for storm flood prediction have led to the analysis of lightning during storms (http://flash-eu.tau.ac.il/index.php). This should achieve a better predictor of dangerous rain storms. Similarly, the study of BASHA et al. (2008) is aimed at better forecasting of torrential rain. It can therefore be concluded that research activities are mainly focused on prolonging the lead time of rain storm forecasts. However, it must also be noted that at the present time, even with the most robust of forecasting schemes employing dense rain gauge networks, radar coverage, satellite algorithms, high resolution computer models of atmospheric processes, and distributed hydrologic models, it is beyond the state of the science to accurately forecast with an effective lead time where flash flooding will occur from convective storms in some situations. Flash flood events are still missed by even these most sophisticated warning systems due to science’s inability to pinpoint the location and timing of small-scale heavy rain (NOAA 2010). In the Czech Republic, it is necessary to create and enhance effective mechanisms to protect people and property from weather extremes and floods (BRÁZDIL 2002). Therefore, the construction of local early warning systems (EWSs) is widely supported by the state; it is the only effective way to improve the forecasting of flash floods (KOCMAN et al. 2011). The basic type of EWS is based on the monitoring of hydrometeorological influences using rain gauges

and stream flow gauges connected to the control unit, which issues a warning in the event of high rainfall and/or exceeding the dangerous water level in the water course (NOAA 2010). This type of EWS is a typical application of flood nowcasting to realtime warning of flood hazard. Nowcasting is a form of very short-range forecasting covering only a very specific geographic area. A nowcast is loosely defined as a forecast for the coming 12-hour period based on very detailed observational data. In the Czech Republic, the pilot system of this type of forecast was built for the community of Olešnice (OBRUSNÍK 2002, KOCMAN et al. 2011). Nowadays the precipitation estimates can be derived using the radar-based methods. These methods are further used for so-called precipitation nowcasting which enables detailed analysis of current weather and its forecast for the next few hours (nowcasting). Application shows measured data (radar echoes, lightning, ground station measurements) as well as analysis and forecasts derived from these data using several nowcasting system, e.g. INCA_ CZ, COTREC_CZ, MERGE (http://www.chmi.cz/ files/portal/docs/meteo/rad/inca-cz/index.html). The purpose of sophisticated EWSs is to increase the lead time of the flood forecast and to reduce the number of false alarms compared to the basic type of EWS. These EWSs take into account some of the hydrological effects, especially the current water retention in the soil cover, which was identified as a very important factor affecting flash flood generation (NOAA 2010). At the scale of a small plot, water retention in the soil can be derived from measurements of soil moisture content or tensiometric pressure of soil water. At a landscape scale, it is possible to estimate the water retention in the soil cover from satellite data. Retention of water in the landscape can also be estimated from the balance of rainfall, runoff, and evapotranspiration. In this way, in the Czech Republic the water retention in the landscape is estimated as an “indicator of flash floods” (ČEKAL et al. 2012). This article describes the sophisticated EWS for flash floods built in the upper part of the Úpa River basin in the Krkonoše Mts. During June 2013, the upper part of Úpa River was impacted by a flash flood created by heavy rainfall. This flash flood caused severe economic losses in the Úpa River basin, especially in the watersheds of Albeřický Potok and Lysečinský Potok creeks, where many houses were damaged or

TESAŘ & ŠÍR: EARLY WARNING SYSTEM

even destroyed. At the request of the local municipality at Horní Maršov, the basins of the Albeřický Potok creek and Lysečinský Potok creek were involved in the EWS though the original proposal of the year 2011 did not include these basins. The EWS is based on knowledge of runoff generation gained through long-term experimental research in small catchments in the Giant Mts and Bohemian Forest (TESAŘ et al. 2008, BÍL et al. 2010, TESAŘ et al. 2011, VONDRKA et al. 2013).The theoretical background and case study from the Úpa headwater region were described by ŠÍR & TESAŘ (2013). In this case the EWS takes into account the so-called fillspill effect which can be formulated as follows: there is a threshold value of the soil moisture content/soil water potential. If the soil moisture content/soil water potential is smaller than the threshold value, water fills the soil pores during the fill phase without leaking into the underlying bedrock while if the soil moisture content is greater, water is released from the soil and the spill phase arises. From this reason it is necessary to monitor also either soil moisture (using soil moisture meters) or soil water potential (using water tensiometers). The best solution is to involve into the EWS both the monitoring of soil physics properties and send the warning message after the reaching the threshold value of at least one of them.

Early warning system for flash floods in the Úpa River basin The EWS is located in the upper part of the Úpa River basin with a closure profile at Horní Maršov below Albeřický Potok creek (Fig. 1). The catchment area is 100 km2, the altitude is 568–1,602 m a.s.l., and the mean elevation is 985.8 m a.s.l. The EWS requires a variety of hardware, software (including computer applications and programs), and communication capabilities to support and maintain the flash flood detection and prediction capacity. Maintenance programs and backup capacity are also needed at a centre (NOAA 2010). In accordance with these requirements the EWS involves: • instrumentation: (1) water-stage recorders (M4016-G – Fiedler-Mágr, Czech Republic), (2) basic stations equipped with rain gauges with a collecting area of 200 or 500 cm2 (Fiedler-

109

Mágr, Czech Republic), air and soil thermometers (Pt100 – Fiedler-Mágr, Czech Republic), soil water tensiometers (T4e – UMS, Germany), and soil moisture meters (Virrib – Amet, Czech Republic); • software: (1) internal software of the water-stage recorders and monitoring stations for the measurement, storage, and transfer of the data to the forecast centre, (2) software of a central data repository, (3) software for data transmission to the internet browser, (4) software for the forecasting of flood risk and sending flood warnings; • forecast centre: (1) operating the data repository, (2) forecasting the water level for each of the connected stream flow gauges, (3) sending flood warnings; • handbooks and user guides: (1) hydrological and soil surveying, (2) EWS scheme, (3) operation and maintenance of the EWS; • complete project documentation of the EWS. The core of the EWS is the M4016-G3 recording unit equipped with a GSM/GPRS modem, solar panel or power supply voltage unit, and reserve charging accumulator (Fig. 2). All necessary sensors can be easily connected to this unit by a wired connection. Sensors can be placed at a distance of 100 m from the M4016-G3 unit. Presently the EWS includes six stream flow gauges with water-stage recorders created by the M4016-G3 unit with connected pressure or ultrasonic water level sensors. The stream gauges are placed in key profiles on the network of watercourses in the basin (Fig. 1). These key profiles were localised with the help of hydrological surveying. The pressure sensor of the water-stage recorder is placed in a metallic protective tube embedded into the riverbank or attached to a bridge pier or to another fixed buttress in the water. An ultrasonic sensor is usually installed on a bridge construction. Nowadays the EWS includes 12 basic stations composed of the M4016-G3 unit with connected sensors for measurement of precipitation intensity and amount, soil moisture, tensiometric pressure, and air and soil temperatures (Fig. 1). Basic stations are located in the vicinity of stream flow gauges and in stands with representative vegetation and soil cover. These stands were localised with the help of soil and vegetation surveying. The meteorological station close to the Výrovka chalet, built on the MS unit (METEOSERVIS – Czech

110

OPERA CORCONTICA 50/S 2013

Republic), provides year-round measurement of the solar radiation, wind direction and velocity, precipitation intensity and amount, air and soil temperatures, and soil moisture. Every M4016-G3 unit has a large data memory for storing measured values. These data are transferred via GSM/GPRS modem to the central data repository. Access to this repository is possible through a standard Internet browser. Every M4016-G3 unit can itself be used for flood nowcasting and sending real-time warnings of flood hazards. When the alarm level is reached (usually

the first stage of flood activity according to the Water Act), a warning SMS will be automatically sent from the station. Up to 30 different hazardous situations can be distinguished (water level, growth of water level, rain intensity, rain total, etc.). In this way, every M4016-G3 unit included in the EWS can work independently. This increases the operational safety of the EWS in the event of a large disturbance in data repository (e.g. caused by failure of several M4016-G3 units in the EWS network). The EWS is therefore resistant to failure of network connection.

Fig. 1. Scheme of the monitoring network in the upper part of the Úpa River basin. Yellow circle – meteorological station close to the Výrovka chalet (year-round measurement of the solar radiation, wind direction and velocity, precipitation intensity and amount, air and soil temperatures, soil moisture). Blue circles – basic stations (precipitation intensity and amount, soil moisture, tensiometric pressure, air and soil temperatures). Blue triangles – stream flow gauge stations. Obr. 1. Schema monitorovací sítě v horní části povodí Úpy. Žlutý kroužek – meteorologická stanice u chaty Výrovka (celoroční měření sluneční radiace, směru a rychlosti větru, intenzity a úhrnu srážek, teploty vzduchu a půdy, vlhkosti půdy). Modré kroužky – základní stanice (intenzita a úhrn srážek, vlhkost půdy, tenzometrický tlak půdní vody, teploty vzduchu a půdy). Modré trojúhelníky – limnigrafické stanice.

TESAŘ & ŠÍR: EARLY WARNING SYSTEM

111

Fig. 2. Scheme of the early warning system. Obr. 2. Schéma systému včasné výstrahy.

Forecasting of the water level for each of the stream flow gauges is based on the SAC-SMA software, which uses hydrological and hydrometeorological quantities stored in a data repository (BUCHTELE et al. 2009). This software takes into account, among other things, the actual retention capacity of the soil cover. In the event that the predicted water level in one of the stream flow gauges exceeds a dangerous level, the forecast centre decides whether to issue a warning. The forecast centre is connected to the central data repository containing data from the entire EWS network, and therefore a robust forecasting scheme can be used for creating warnings. Thus, this sophisticated EWS is able to increase the lead time of flood forecasts and to reduce the number of false warnings compared to the basic type of EWS.

Conclusions The sophisticated early warning system for flash flood detection built in the upper part of the Úpa River basin in the Krkonoše Mts is based on knowledge of runoff

generation gained through long-term experimental research in small catchments in the Giant Mts and the Bohemian Forest. Presently the EWS consists of a meteorological station, six stream flow gauge stations, and 12 stations for measurement of precipitation intensity and amount, soil moisture, tensiometric pressure, and air and soil temperatures. Every station has a large data memory for storing measured values. These data are transferred via GSM/GPRS modem to the central data repository. Every station can independently send real-time warnings of flood hazards. Forecasting of the flood risk is based on the continuous monitoring of hydrometeorological and hydrological quantities in all stations: rain intensity and total, discharge, soil moisture, tensiometric pressure of the soil water, and air and soil temperatures. Forecasting of the water level for each of the stream flow gauges is based on SAC-SMA software, which uses data stored in a central repository. A robust forecasting scheme is used for creating warnings. Thus, this sophisticated EWS should be able to increase the lead time of flood forecasts and to reduce the number of false warnings compared to the basic type of EWS.

112

OPERA CORCONTICA 50/S 2013

Souhrn

KAKOS V. 1978: Hydrometeorologická charakteristika povodní na území ČSR. VTEI 4: 127–131.

Včasné varování před bleskovými povodněmi se v posledních letech stává velmi aktuálním tématem pro řadu sídel v podhorských a horských oblastech. Příspěvek se zabývá různými metodickými přístupy k zajištění včasných prognóz/predikcí vniku bleskových povodní a popisuje systém včasné výstrahy, realizovaný v horní části povodí Úpy v Krkonoších. Rozebírá jednotlivé složky použitého systému, který je založen na kontinuálním monitoringu rozsáhlého souboru hydrologických a meteorologických ukazatelů (množství a intenzita dešťových srážek, půdní vlhkost, tenzometrický tlak půdní vody, teplota půdy a vzduchu a další). Hodnocení dat využívá softwaru SAC-SMA, který, kromě jiného, hodnotí i aktuální retenční kapacitu půdního krytu.

KOCMAN T., KUBÁT J. & MUSIL P. 2011: Lokální výstražné a varovné systémy v ochraně před povodněmi. Ministerstvo životního prostředí, Praha. 70 pp.

Acknowledgements The EWS described is supported by the Technology Agency of the Czech Republic (Project TA02021451).

References BASHA E., RAVELA S. & RUS D. 2008: Model-based monitoring for early warning flood detection. In: Proceedings of the 6th ACM Conference on Embedded Networked Sensor Systems, Raleigh, North Carolina: 295–308. BÍL M., FÜRST T., VODÁK R., PRAŽÁK J., ŠÍR M. & TESAŘ M. 2010: Nový podezřelý ve vyšetřování příčiny bleskových povodní aneb jak jsme sto let řešili špatnou rovnici. In: HOCH T. & ŠUMBEROVÁ M. (eds), Geografie pro život ve 21. století, Ostravská univerzita v Ostravě: 235–239. BRÁZDIL R. 2002: Meteorologické extrémy a povodně v České republice – Přirozený trend nebo následek globálního oteplování? Geografie – Sborník České geografické společnosti 107, 4: 349–370. BUCHTELE J., TESAŘ M. & KRÁM P. 2009: Variability of the water regime in forested experimental catchments. Soil & Water Research 4 (Special Issue 2): 93–101. ČEKAL R. et al. 2012: Průvodce informacemi pro odbornou vodohospodářskou veřejnost. Český hydrometeorologický ústav, Praha. 51 pp.

Ministerstvo životního prostředí 2011: Předběžné vyhodnocení povodňových rizik v České republice. Ministerstvo životního prostředí. Praha. 31 pp. NOAA 2010: Flash Flood Early Warning System Reference Guide. University Corporation for Atmospheric Research. 204 pp. OBRUSNÍK I. 2002: Úloha předpovědní a výstražné služby Českého hydrometeorologického ústavu při snižování následků přírodních pohrom. In: ROŽNOVSKÝ J. & LITSCHMANN T. (eds), XIV. Česko-slovenská bioklimatologická konference: 330–338. ŠÍR M. & TESAŘ M. 2013: Water Retention and Runoff Formation in the Krkonoše Mts. The Supplementum of Opera Corcontica 50: 97–106. ŠTEKL J., BRÁZDIL R., KAKOS V., JEŽ J., TOLASZ R. & SOKOL Z. 2001: Extrémní denní srážkové úhrny na území ČR v období 1879 –2000 a jejich synoptické příčiny. Národní klimatický program České republiky 31. 140 pp. TESAŘ M., ŠÍR M., LICHNER Ľ. & FIŠÁK J. 2008: Extreme runoff formation in the Krkonoše Mts in August 2002. Soil & Water Research 3 (Special Issue 1): 147– 154. TESAŘ M., ŠÍR M. & POLÍVKA J. 2011: Návrh systému včasné výstrahy před bleskovými povodněmi v horských podmínkách. In: ŠÍR M. & TESAŘ M. (eds), Hydrologie malého povodí 2011, Ústav pro hydrodynamiku AVČR: 469–472. VAŠKŮ Z. 2009: Bleskové povodně. Vesmír 88: 619–622. VONDRKA A., TESAŘ M. & ŠÍR M. 2013: Vyplavování sodného a draselného iontu z malého horského povodí na Šumavě. Vodní hospodářství 7: 21–25.

Websites http://www.chmi.cz/files/portal/docs/meteo/rad/ inca-cz/index.html