APPLICATION OF RAINFALL-RUNOFF MODEL: CLIMATE CHANGE IMPACTS ON RESERVOIR INFLOW

Kateřina Knoppová1,2, Daniel Marton2, Petr Štěpánek1,3 

1. Czech Hydrometeorological Institute, Czech Republic
2. Institute of Landscape Water Management, Brno University of Technology, Czech Republic
3. Global Change Research Institute, Czech Academy of Sciences, Czech Republic

Corresponding author: Kateřina Knoppová, Czech Hydrometeorological Institute, Kroftova 2578/43, 616 67 Brno,

Czech Republic, katerina.knoppova@chmi.cz

ABSTRACT

The impacts of climate change are beginning to be felt in the Czech Republic. In recent years, we were challenging a dry period, which threatens to continue affecting Czech economy, agriculture and personal comfort of local people. The need to adapt to climate change is obvious. The groundwater resources are in continuous decline, consequently, the surface water supplies are increasing in importance. How would the quantity of available water change in the future? How much water would we be able to store within the year to manage it during the dry seasons?

Rainfall-runoff models enable us to simulate future changes in hydrological conditions based on climate projections. One of such tools is Runoff Prophet, the conceptual lumped model being developed at the Institute of Landscape Water Management at Brno University of Technology. It is used to simulate time series of monthly river flow in a catchment outlet without the need to describe the morphological characteristics of the catchment. Runoff Prophet produced good results of calibration and proved its suitability for conceptual hydrological modelling in variable hydrological conditions of the Czech Republic.

The aim of the paper was to assess the possible impact of climate change on future inflow into Vír I. Reservoir, one of the drinking water resources for Brno, a city of 380 000 inhabitants. The recently developed software Runoff Prophet was used to simulate future river flow time series. The model was calibrated on the catchment of gauging station Dalečín on Svratka River as the reservoir inflow. Prognoses of future river flow were performed using climate scenarios prepared by Global Change Research Institute of Czech Academy of Sciences. These scenarios (RCP types) are based on the outcomes from different regional climate models of Euro-CORDEX initiative. Characteristics of possible future air temperature and precipitation in the basin were evaluated in terms of its impact on reservoir management. The results of hydrological modelling gave the perspective of expected changes in Vír I. inflow yield. The options of using Vír I. Reservoir as a drinking water supply for Brno in coming decades were assessed.

Keywords: Runoff Prophet, climate change, rainfall-runoff model, hydrology, water resources

INTRODUCTION

The impacts of climate change are beginning to be felt in the Czech Republic. In future, the remarkable changes in temperature and precipitation are expected to continue (Štěpánek et al. 2016). According to the Czech Hydrometeorological Institute evaluation (Daňhelka et al. 2019), in 2018, our country was still suffering from a dry period, which is lasting since 2014. One of the indicators of drought is a low value of mean annual rainfall in comparison to its long-term average, which is 686 mm in the Czech Republic for the reference period 1981–2010. Year 2018 was evaluated as the most critical one from the dry period with 164 mm mean annual rainfall deficit (24 %). Year 2015 was the second driest with 154 mm deficit (22 %). Concerning the hydrological impact in the form of low river flows, the situation is getting worse due to long-term accumulation of water deficit from previous years. In year 2019, the drought is expected to continue. For the moment (07 2019), we have 9 % deficit with 370 mm of rainfall from beginning of the year, which is not enough to restore balance to the hydrological cycle.

The dry period in the Czech Republic has visible impact on water resources, both surface and underground. Water levels are declining in not only house wells, a boreholes and springs, but in large groundwater resources of drinking water. As the groundwater yield declines, the importance of surface water resources is increasing. Brno, the second biggest city of the Czech Republic with 380 000 inhabitants, has two main sources of drinking water - Březová nad Svitavou spring area (treated groundwater) and Vír I. Reservoir (treated surface water). In 2018 annual report of Brno water and sewage works (Brněnské vodárny a kanalizace, a.s. 2019), we can see that the ratio of drinking water taken from Vír I. increased from 6 % to almost 13 % since 2014. Therefore, it is appropriate to ask how the quantity of available surface water would change in the future.

From the perspective of climate, we have scenarios of possible future development from different climate models at our disposal. Global Change Research Institute of Czech Academy of Sciences is one of organizations specialized in climate projections for the Czech Republic. For this study, they provided bias corrected scenarios (RCP types) based on outcomes from different regional climate models of Euro-CORDEX initiative.

Based on these climate data, the projections of future hydrological situation can be made.   Nevertheless, we have to be able sufficiently describe the rainfall-runoff process in the particular catchment. Rainfall-runoff models are tools enabling simulation of this natural cycle. There are some very complex ones with high demands on quantity of input data and description of catchment characteristics. However, there are also very simple models with empirical or conceptual approach, where the relationship between cause and consequence is searched using calibration datasets of input elements. One of such models is also Runoff Prophet described in Knoppová (2018) and Knoppová & Marton (2019) which was developed at the Institute of Landscape Water Management at Brno University of Technology.

The aim of the paper is to assess the possible impact of climate change on the quantity of water available in Vír I. Reservoir. One of the key factors is the future water yield of the reservoir inflow, which is the Svratka River above the gauging station Dalečín. Software Runoff Prophet was used to simulate its future river flow time series. The model was calibrated on the catchment of gauging station Dalečín and the hydrological projections were made using ensemble of climate scenarios.

We have evaluated the characteristics of projected air temperature and precipitation in terms of its impact on reservoir management. The results of hydrological modelling have given the perspective of expected changes in Vír I. inflow yield. We have assessed the options of using Vír I. Reservoir as a drinking water supply for Brno in coming decades.

METHODS AND DATA

Climatological and Hydrological data

The Czech Hydrometeorological Institute provided processed historical climate and hydrological time series for model calibration. We have been using time series of monthly mean air temperature [°C], monthly precipitation sum [mm] and mean monthly discharge [m3/s].

The projected climatological data have been processed at the Global Change Research Institute of Czech Academy of Sciences. The input data have been prepared according to Štěpánek et al. (2009, 2011) and the output scenarios bias corrected by quantile mapping method (Štěpánek et al. 2016). Based on projections of future temperature and precipitation for the Czech Republic, an ensemble of 11 simulation have been prepared. It consists of three RCP scenarios from five climate models (Table 1). The ensemble covers range of possible future climate development. We have been using simulations of daily mean air temperature [°C] and precipitation sum [mm] time series recalculated to monthly time step.  

Table 1. Description of used climate models/scenarios

Historical data (climate and hydrological) have been divided into tree datasets - one for calibration (1970–2005) and two for validation of calibrated model (1961–1969 and 2006–2018). Simulated climatological data consisted of control run of used climate models (1970–2005) and projected scenarios (2006–2099).

KEY:

HISTORICAL DATA MODEL CONTROL RUN

MODEL PROJECTION

Fig. 1. Scheme of used historical and projected data 

To evaluate the results, we have recalculated the data into a difference to its long-term average (reference period 1981–2010). We have used a double moving average MA(5x30) to show the trends in projected climatological and hydrological data. Thus, we have calculated a moving average of order 30, and then applied another moving average of order 5 to the results. We have also used boxplots to show basic statistical characteristics of the ensemble of scenarios. It shows minimum, first quartile Q25, median, mean (cross mark), third quartile Q75, and maximum.

Runoff Prophet

Runoff Prophet is a hydrological model for simulations of river flow in the catchment outlet. It is based on parametrized rainfall-runoff equations described in Wang et al. (2013). For chosen catchment, 37 parameters in equations have to be optimized. Nash-Sutcliffe model efficiency coefficient (NS) is used as the optimization criterion and it also express the model calibration success.  

The calculation requires no geographical characteristics of the catchment but its area. Firstly, the model is calibrated and validated using historical climatological and hydrological data. After that, future discharge can be simulated based on projected climatological data.  

Fig. 2. Study area - Dalečín as the main Vír I. Reservoir inflow

STUDY AREA

Vír I. is an open water reservoir with 56 mil. m3 of total water storage capacity. Its´ purpose is mainly to secure ecological discharges in Svratka River under the reservoir; it serves as drinking water supply, for hydro power production and as a flood protection. The gauging station Dalečín on Svratka River represents the reservoir inflow. It´s basin occupies an area of 366.94 km2. After previous evaluation, we have chosen climate station Polička as a representative one for the catchment. Thus, the rainfall-runoff process in the catchment has been described by time series of temperature and precipitation in Polička and discharge in Dalečín.

RESULTS AND DISCUSSION

Model calibration and validation 

Success of model calibration and validation has been quantified by the value of NS and the difference in observed and simulated mean yearly discharge (MQ). We have calibrated the model on 36 years of historical data (1970–2005). Resulting NS has been 0.69 that signifies good efficiency of hydrological model according to Moriasi et al. (2007). Mean difference in MQ has been -2.79%. Calibrated model has been validated on two datasets: 1961–1969 (A) and 2006–2018 (B). Dataset A has had very pleasing results. Dataset B, nevertheless, has had worse result - satisfactory model efficiency with NS < 0.65 and MQ considerably undervalued (-8.17%).  

We have assumed that the rainfall-runoff process in the period B was affected by the drought. Therefore, we have split dataset B into two parts – B1 (2006–2013) and B2 (2014–2018). The results have confirmed our assumption. Model has had good results on B1. Within the dry period B2, Runoff Prophet has underestimated MQ by almost 19% and the value of NS was unsatisfactory.

Table 2. Results of model calibration and validation

Projected changes in temperature and precipitation

All climate scenarios from the ensemble show increasing trend in projected mean temperature (Fig. 3). When we look at the historical data (red line), we can see that it is in the upper part of the model ensemble. It could indicate that we would stay in the upper, i.e. warmer part also in the future.  The precipitation sum has slightly increasing trend in projected data. In late 50´s, all the ensemble members are above the reference long-term average. Until beginning of the dry period (2014), the historical data are in the upper part of the ensemble. Since then, the precipitation sum is on rapid decline.

In addition to the continuous trends, we have evaluated the modelled data in three consecutive 30-year periods: 1981–2010 (present), 2011–2040 (near future) and 2041–2070 (distant future). Fig. 5 shows basic statistical characteristics of temperature and precipitation ensemble in these periods. In the distant future, the median mean temperature is 1.6°C above the long-term average. If the temperature growth continued in current trend, it will more likely move towards the maximum projected value - increase of 2.8°C.  

Concerning the precipitation ensemble, we can see that in the distant future the variance of scenarios is markedly smaller than in the near future. The median precipitation sum in distant future is 5% higher than the long-term average. We have to consider that the simulations slightly underestimated the present mean sum of yearly precipitation (96% of long-term average). Thus, we can assume that it has underestimated precipitation also in the future and the resulting sum could be even higher.

Fig. 3. Temperature trends 

Fig. 4. Precipitation trends 

Fig. 5. Statistical characteristics of temperature and precipitation ensemble 

Projected changes in the reservoir inflow 

Most scenarios from the ensemble have slightly decreasing trend in projected mean discharge, but they are rather diverse. In historical data, we can see sharp decrease since 2014.

Fig. 6. Discharge trends 

As with climatological projections, we have evaluated the statistical characteristics of the ensemble in three consecutive 30-year periods. However, the results have shown that for the discharge evaluation, 30-year periods are insufficient. Therefore, we have examined also 20-year consecutive periods (Fig. 7).  

Fig. 7. Statistical characteristics of discharge ensemble – 20 and 30-year periods

In 30-year periods boxplot, we can see that the discharge modelled for presence slightly underestimate the reality (median = 99 % of the long-term average). Distant future has mean discharges from 89 to 103 % of the long-term average. However, more than 25 % of ensemble is still above the long-term average. Thus, it is hard to say if the future discharge would be higher or lower then today.  

In 20-year periods boxplot, from 2021–2040 the median discharge visibly decreases. From 2061, more than 75 % of the scenarios are under the long-term average. Based on these results, we can say that the mean yearly discharges in 2061–2099 will be more likely lower than the long-term average.  

CONCLUSIONS

As the results have demonstrated, mean temperature will increase in the basin of Dalečín. According to the current trends, until 2070 it could be between 1.6 to 2.8 °C above the long-term average. The yearly precipitation sum will have slightly increasing trend in the range of several percent. Based on used climate scenarios, mean yearly discharge in Svratka River should remain in current state until 2030´s. Since then, the yield of the Vír I. Reservoir inflow will decline. Until 2099, the ensemble shows median discharge 9 % down from the long-term average.

Simulated changes in the reservoir inflow are not very high. However, there are some facts that should not be overlooked. The current dry period (2014–now) do not fit very well into the modelled ensemble. In precipitation, we have recorded a sharp change in the long-term increasing trend. The current low values of mean discharge are even beyond the simulated ensemble. Thus, the presented results are valid unless the dry period is only a rare extremity. Unless the discharge decline could be much worse. Furthermore, the rising temperature would mean greater evaporation from open water bodies. It could also negatively influence the water quality and thus its treatability.

According to our results, the amount of drinking water in Vír I. Reservoir will decrease in the future. If the yield of Březová groundwater resource would continue to decline, Brno city should rethink its drinking water management.

REFERENCES 

Brněnské vodárny a kanalizace, a.s. Výroční zpráva 2018. 2019. 134 s. [Brněnské vodárny a kanalizace, a.s. (= Brno water and sewage works) Annual Report 2018. 2019. 134 s.]

Daňhelka et al., Vyhodnocení sucha na území České republiky v roce 2018. Český hydrometeorologický ústav. 2019. 90 s. [Daňhelka et al., Evaluation of Drought in the Czech Republic in 2018. Czech Hydrometeorological Institute. 2019. 90 p.]

Knoppová K., Marton D. Development of a rainfall-runoff model: Suitability Assessment in the Morava River Basin, Czech Republic. WIT Transactions on Ecology and the Environment. 2019. Vol 234. P. 143 – 154.

Knoppová, K. Srážko-odtokový proces v podmínkách klimatické změny (Diplomová práce). Vysoké učení technické v Brně. 2018. 60 s. [Knoppová, K. Rainfall-runoff process in time of climate change (Master's thesis). Brno University of Technology. 2018. 60 p.]

Moriasi, D. N., Arnold, J. G., Van Liew, M. W., Bingner ,R. L., Harmel, R. D. & Veith, T. L., Model evaluation guidelines for systematic quantification of accuracy in watershed simulations. Transactions of the ASABE. 50(3), pp. 885900, 2007.

Štěpánek, P., Zahradníček, P., Brázdil, R., Tolasz, R. Metodologie kontroly a homogenizace časových řad v klimatologii. 2011, ČHMÚ, 118 s. ISBN 978-80-86690-97-1 [Štěpánek, P., Zahradníček, P., Brázdil, R., Tolasz, R. Methodology of data quality kontrol and homogenization of time series in climatology. 2011, ČHMÚ, 118 s. ISBN 97880-86690-97-1)

Štěpánek P., Zahradníček P., Farda A., Skalák P., Trnka M., Meitner J., Rajdl K. Projection of drought-inducing climate conditions in the Czech Republic according to Euro-CORDEX models. Clim Res. 2016. Vol. 70. P. 179-193. https://doi.org/10.3354/cr01424

Štěpánek, P., Zahradníček, P. a Skalák, P. Data quality control and homogenization of air temperature and precipitation series in the area of the Czech Republic in the period 1961–2007.Advances in Science and Research, 2009, 3, pp. 23–26, https://doi.org/10.5194/asr-3-23-2009

Wang, G.Q., Zhang, J.Y., Xuan, Y.Q.  et al., Simulating the Impact of Climate Change on Runoff in a Typical River Catchment of the Loess Plateau, China. Journal of Hydrometeorology, 14, pp. 1553–1561, 2013.

Back to CONTENTS