AZ EGER-PATAK HIDROLÓGIAI VIZSGÁLATA, A FELSZÍNI VÍZKÉSZLETEK VÁRHATÓ VÁLTOZÁSÁBÓL ADÓDÓ MÓDOSULÁSOK AZ ÉGHAJLATVÁLTOZÁS HATÁSÁRA GÁBOR KEVE 1, GÉZA HAJNAL 2, KATALIN BENE 3, PÉTER TORMA 4 EXTRAPOLATING FLOOD HAZARD POTENTIAL OF EGER CREEK DUE TO CLIMATE CHANGE Abstarct The impacts of climate change on water resources are a new agenda for modeling and management. Not only changes in availability, but flood risks due to extreme weather conditions cause proplems for hydraulic engineers (BÁLINT, 2009). In this research, we develop a rainfall-runoff model for the Eger Creek watershed, and a one-dimensional hydrodynamic model for downtown Eger. The models were based on daily field measurements between 1991-2010 as well as site observations. Using a lumped parameter approach we compared available water resources of the calibrated model to a future model incorporating projections from climate researchers. Finally, we evaluated the change in flood risk in downtown Eger and the Eger Creek watershed using a one dimensional dynamic model. Keywords: Climate change, rainfall-runoff model, calibrated, future model, flood risk, hydrodynamic model Study site and relevant data This study was initiated by the Eszterházy Károly College (EKF) to evaluate the impact of climate change on the surface water resources and flood risk in the Eger Creek watershed. The Eger watershed (approximately 296.93 km 2 in area) is located in the SE-E-SW-n side of the Bükk Mountain. The watershed can be divided into two subwatersheds; the Upper Eger-Creek subwatershed (AEP450) from the origination to the joining of the Tárkány Creek tributary, and the Eger- Creek subwatershed (AEP449) from the Tárkány Creek tributary to the Nagytályai dam (shown in Figure 1.). The watershed is located in a transition area between the mountainous and plain region of the country. Nearly 50% of the watershed is flat, elevation less than 200 m above sea level and in the northern area, 25-30% hilly rural area with elevation above 400. Geological structure and soil conditions of the area are varied. Overall, the infiltration capacity of the soil surface is low to poor, except in the southern part of the watershed where the infiltration capacity of the surface soils are higher. (VITUKI, 1961) 1 Assistant Lecturer, Eötvös József College, keve.gabor@ejf.hu; 2 Associate Professor, Technical and Economic University of Budapest, hajnal@vit.bme.hu; 3 Associate Professor, Széchenyi István University, benekati@sze.hu; 4 PhD student, Technical and Economic University of Budapest, torma@vit.bme.hu
Eger Creek is connected through the karst system to the rest of the Bükk streams. A large part of the watershed is karst. The eastern part of the watershed is highly branched and wide and receives almost all of the runoff volume from the eastern side of the Bükk Mountain. From the eastern side the only major tributary is the Villői Creek watershed. Tárkány Creek joins the Eger Creek at Felnémet, doubling the flow in the creek. After the Eger creek flows though the town of Eger, it reaches the Nagytálya dam. (VIZITERV, 2003) There are several small reservoirs on the watershed. The reservoirs, along with the karst system, are lowering and delaying the peak of the surface runoff. (VGI, 1984) The shape and the volume of the runoff hydrograph is impacted as well; the peak is lower with a long tail. As a result, on Eger Creek, flood waves occur with less frequency but higher volume. The following data (Table 1.) were obtained for modeling 1. Table: Hydro-meteorological data for modeling Source Data Location Recorded Time Recorded time increment ÉMVIZIG. Flow depth Runoff Almár, Felnémet hydrological stations OMSZ Rainfall Eger 2014.05.24-27. CARPATCLIM EKF Temperature Rainfall Global radiation Rainfall Temperature Eger-creek watershed Eger-creek watershed (ALADIN/REMO difference from recorded base time ) 1992-2014 Daily or hourly data 1961-2010. daily 10 minutes 2021-2050. daily
1. Figure: Eger Creek Watershed The SRTM (Shuttle Radar Topography Mission) 9x9 m resolution, multipurpose digital surface model was used for watershed delineation, in harmony with the river network system. The land use and soil types were determined using Google Maps (maps.google.com). Watershed model and calibration The HEC-HMS (USACE, 2013) numerical model was selected to evaluate the rainfall-runoff processes in the Eger Creek watershed. In this model, the hydrologic elements can be connected in a network imitating watershed hydrologic structure. The watershed processes are organized into six main components. The meteorological component is the first computational component of the continuous model, by means of rainfall input. In the next step, rain falls on either a pervious or an impervious surface. Rainfall from the previous surface is subject to losses (interception, infiltration, evaporation and transpiration) modeled by the rainfall loss component; in our case the SMA model. The effective rainfall from the loss component contributes to direct runoff and to groundwater flow in aquifers. Rainfall from the impervious surface is not subject to losses and instantly enters the direct runoff component, where is it transformed to overland flow For direct runoff, we used the Clark unit hydrograph method. The movement of water in aquifers is modeled by the
baseflow component. We used a linear reservoir baseflow, which is used in conjunction with the SMA model. Both overland flow and baseflow enter creek channels (HEC, 2000). The translation and attenuation of streamflow in the creek is simulated by the river routing component; in our case the Muskingum- Cunge method. 2. Figure: Continuous model representation The watershed was divided into 13 subwatersheds with the SRTM model, and watershed characteristics, such as slope, land used, time of concentration, etc., were determined (Figure 2). The continuous model uses 19 stream reaches and input parameters for the model were determined from map measurements and on-site observations. The model has 20 junctions and among these are two locations where daily streamflow measurements were conducted; JQ Almár, marked with a black box, JQ Felnémet. We evaluated the simulation results at
these points and at J164 where the two streams join, and at the outfall point (Figure 2). Meteorological components included daily rainfall and temperature between 1961 and 2010. Data was downloaded from the CARPATCLIM (2014) public model. We overlaid the available grid points on the watersheds, and selected 5 points that were in the watershed. The numerical averages of the components were calculated, and given as the input into the model. Daily evapo-transpiration was calculated with the Priesley-Taylor model. For forecasting the REMO and Aladdin model was applied with a slight modification in the case of rainfall data. The 5-layer soil-moisture accounting (SMA) model was used to calculate precipitation loss. The SMA model has four different storages; canopyinterception storage, surface-depression storage, soil storage, and groundwater storage. The Clark unit hydrograph method was used in the continuous model to transform excess rainfall into direct runoff. The water that exceeds the infiltration rate and overflows the surface storage in the SMA model is the input to the direct runoff component. The parameters of this method, the time of concentration and the surface storage coefficient, were estimated with calibration and measurements. The SMA is designed to be used in conjunction with the linear reservoir baseflow model, and outflows from SMA groundwater layers are inflows to baseflow linear reservoirs. We used only the upper linear reservoir model. The systematic manual calibration relied on the initial estimates of the input parameters. The output from the manual calibration was assessed by flow comparison graphs, and the statistical goodness-of-fit measures. For calibration the 2004, 2006, 2008-2010 years were selected. Table 2 shows the statistical performance measures selected for the model evaluation at the Almár junction. 2. Table:Statistical measures of the calibration results at the Almár junction Statisztikal measure Calibrated value Measured alue Average Q (m 3 /s) 0,496 0,506 Standard deviation Q (m 3 /s) 0,930 0,848 RPWRMSE (%) 202 CORR 0,67 RRMSE (%) 242 PEV (%) 1,90 NS 0,27 The 2010 year was selected for showing graphical comparison between the calibrated HMS model results and the measured flows (Figure 3)
Vízhozam (m 3 /s) 12.0 10.0 Mért HMS 8.0 6.0 4.0 2.0 0.0 1-Jan 1-Feb 1-Mar 1-Apr 1-May 1-Jun 1-Jul 1-Aug 1-Sep 1-Oct 1-Nov 1-Dec 2010 3. Figure:2010 calibration result between measured and modeled flows at the Almár junction. Impact of climate change on water resources After watershed calibration the runoff hydrograph between 1961 and 1990 was determined at the Nagytályai outflow point. The REMO program was used to predict rainfall and temperature data between 2021 and 2050. Applying the calibrated model, the outflow hydrographs were calculated from 2021 to 2050, at the Nagytálya dam. A frequency analysis was used to compare the two 30-year discharge values. On figure 4, the horizontal axis shows discharge, the vertical axis shows the non-exceedance probability. It appears that for equal probability, discharge between 0-2 m 3 /s is expected to be slightly higher in the future than previously seen. However, over the range of 2 m 3 /s, no significant differences can be observed. The results predicted by the REMO model indicates also that there will be fewer rain-free days, however the frequency of lower precipitation days will increase. From the hydrological point of view, this means that the catchment will be wet for longer periods of time. Therefore an already moderate rain will cause runoff, since the initial loss will be substantially less, and therefore the effective rainfall increases. The same test was performed on the winter and summer season, and it can be concluded that
the majority of runoff will occur in winter. For less than 4.5 m 3 /s discharge, at the same probability, the runoff will be higher as compared to previous times. In the summer season, the changes will be smaller compared to the previous 30- year period. The frequency of large runoff will be slightly smaller, while the frequency of low flows will be slightly larger. 4. Figure: Non-exceedence probability distribution of past and predicted flows.. 5. Figure:Flow duration curve of past and predicted minimum annual flows One of the most important question in our study was What can we expect in the low flow region? Since this flow level will serve as a design standard for water resources management decisions. We examined climate change impact on the minimum annual water discharge distribution. For both 30-year periods we collected the annual minimum daily discharges and then determined their
frequency distribution. Figure 5 shows in order to maintain the same magnitude of non-exceedance; one must have much higher discharge rates in the future. For example, 0,1 m 3 / s annual minimum flow or less in the future can be observed with a probability of 80 %, compared with 98 % of the previous period. The second major issue is concerned with flooding. We generated flow duration curves for the annual maximum flows. Figure 6 shows that the annual maximum will increase. In the future, for the same probability, the maximum annual flow will be higher than before. As an example, for 30 m 3 /s or less annual maximum flow, the non-exceedance probability is 86% (compared to 95% previously); and the flow will be greater than 30 m 3 /s 14% of the time. 6. Figure: Flow duration curve of past and predicted minimum annual flows Summary Based on the available data and information a rainfall-runoff model was built in HEC-HMS software to investigate the climate change impacts on Eger creek catchment. Our model behaved well during the Calibration and Verification periods (1990-2010). The errors between the measured and computed results were acceptable, they were were less than the uncertainty of climate change estimation. In this way our forecasted discharge series for the future (2021-2050) based on the predicted meteorological scenarios (ARMA/REMO) were ready to
analyze. Many hydrological methods were used to examine the changes in water regime and finally we got the next results: The frequency of large runoff will be slightly smaller, while the frequency of low flows will be slightly larger. In the future, for the same probability, the maximum annual flow will be higher than before. References BÁLINT, G. et al. 2009: Kisvízfolyások árvízi veszélytérképezése, XVII. MHT Országos vándorgyűlés CD-ROM kiadványa, Baja CARPATCLIM, 2014: http://www.carpatclim-eu.org/pages/home/, 2014.08.22. HEC 2000: Technical Reference Manual, Hydrologic Modelling System HEC-HMS, Hydrologic Engineering Center, US Army Corps of Engineering USECE 2013: User s Manual version 4, Hydrologic Modelling System HEC-HMS, Hydrologic Engineering Center, US Army Corps of Engineering VIZITERV CONSULT Kft, KERTAI, E. et. al. 2003: Eger-Rima-Laskó vízrendszer vízrendezési, vízkárelhárítása fejlesztése, megalapozó tanulmány VITUKI 1961: Az Eger-patak hidrológiai tanulmánya, Budapest VGI 1984: Magyarország vizeinek műszaki-hidrológiai jellemzése, A Felső-Tisza jobb parti vízrendszere, Budapest