Water Resour Manage DOI 10.1007/s11269-017-1642-5
Assessing the Impact of Climate Changes on Hydrological Drought Based on Reservoir Performance Indices (Case Study: ZayandehRud River Basin, Iran) S. Khajeh 1 & Sh. Paimozd 1 & M. Moghaddasi 1
Received: 28 November 2015 / Accepted: 20 March 2017 # Springer Science+Business Media Dordrecht 2017
Abstract We investigated the potential impacts of climate changes on hydrological drought of the Zayandehrud basin in Iran. The meteorological data was simulated using the LARS-WG model by output downscaling of HADCM3, INCM3 and NCCCSM from 2011 to 2040 under A1B, A2 and B1 scenarios. In order to estimate the runoff of the basin, the IHACRES model was calibrated and validated by data obtained from the Koohrang station during 1987–2007. The runoff was calculated for 2011–2040 period based on the mean of downscaling values in the IHACRES model. Then, the impact of climate changes on hydrological drought was studied by a probabilistic approach and MSUI index. The results revealed a drastic reduction in normal levels and also a significant increase in moderate, severe, and very severe drought levels in the future compared to the base period. The MSUI index represents a 42.32% reduction in the normal level based on the HADCM3 model under A1B scenario as well as an 86.8%increasein the severe drought. Furthermore, the HADCM3 model using the probabilistic approach showed a 73% reduction under the A2 scenario for normal condition and an increase of 20% for severe drought. Keywords Climate change . Hydrological drought . MSUI index . Probability approach . Reservoir
1 Introduction The drought is one of the major environmental challenges for human being in recent years; so, numerous researches have been concentrated on the field. According to IPCC, droughts will become more intense, frequent and severe in the future due to the impact of climate change. Blenkinsop and Fowler (Blenkinsop and Fowler 2007) assessed the probability of drought occurrence in the future years in England using GCM models. The results of the study
* M. Moghaddasi
[email protected];
[email protected]
1
College of Agriculture, Arak University, Arak, Iran
Khajeh S. et al.
suggested a reduction in the severity and duration of droughts. Sheffield and Wood (Sheffield and Wood 2008) scrutinized the change trends in the drought occurrence using soil moisture data for three future climate scenarios under eight GCM models. The models indicated a decrease in soil moisture at the global scale for the future scenarios with a corresponding doubling of the spatial extent of severe soil moisture deficits and frequency of short-term droughts from the mid-twentieth century to the end of the twenty-first. Loukas et al. (Loukas et al. 2008) estimated the severity of the drought affected by climate change using SPI index in Thessaly, Greece. In this research, the output of CGCM2 model was used under two A2 and B2 scenarios in the near future and in the far future. The results presented a significant increase in the severity of drought at the end of the century for both scenarios. Fujihara et al. (Fujihara et al. 2008) studied the effects of climate change on hydrology and water resources of Seyhan river basin in Turkey. In the research, the data of MRI-CGCM2 and CCSR/NIES/ FRCECMIROC models was employed under the A2 scenario. Evaluation of the average annual temperature changes revealed that compared to the present time, this parameter would undergo +2 °C and +2.7 °C changes in the future given the MRI and CCSR models. The results also indicated that the annual evapotranspiration decrease would be 36 mm (9%) under the MRI model and 39 mm (10%) under the CCSR model. The basin runoff evaluation was indicative of 118 mm (52%) and 139 mm (61%) runoff reduction respectively using the MRI model and the CCSR model. (Mishra and Cherkauer 2009) studied the impact of climate change on drought in the twenty-first century (near, far further) in west part of the United States. The results showed that in the region, precipitation and minimum temperature had followed an increasing trend while a decreasing trend had been recorded for maximum temperature. Considering the changes, they observed that drought events could occur as the previous period with a minor rise in drought at the end of the century. Wang et al. (Wang et al. 2011) have investigated the impact of climate change on meteorological, agricultural, and hydrological drought in Illinois. In the research, the output of three models was employed including PCM, CCSM and HADCM3 under A2 and B2 scenarios. It can be concluded that agricultural and hydrological drought would have greater intensity in the future because of the interference of temperature effect. In a study conducted in 2012 by Yang et al., the impact of climate change on drought was investigated in southern Taiwan by utilizing data from nine GCM models under A1B scenario. The results showed the increase probability of frequency and severity of droughts in the future compared to the base period. Vrochidou et al. (Vrochidou et al. 2013) used the GCM models output under two A2 and B1 scenarios to study the climate change impacts on meteorological drought. They calibrated the HMS-HBV model for Plat is basin during 1974–1999. The results obtained in the research showed that the number of drought events could rise 98%, 109%, and 81% (56%, 92%, and 34%) under the influences of water flow, soil moisture, and groundwater by 2100 given the A2 scenario (B1). Lespinas et al. (Lespinas et al. 2014) studied the impact of climate changes on the water resources of Mediterranean coastal rivers. Primarily, they run the GR2M model in order to produce the observational hydrological regime of the river, then they examined the impacts of climate changes under different climate models regarding A2 and B1 scenarios for 2100–2071 time period. The results revealed that under the A2 (B1) scenario, the average annual temperature would increase 4.3–4.5 °C (3.1–3.2) for 2071–2100 compared to the base period (1961–1990). Precipitation changes also showed a reduction between −10% to −15.6% under the A2 scenario and −6.1% to −11.6% under the B1 scenario. Furthermore, the output of GR2M model indicated that the reduction in water discharge under the A2 (B1) scenario would be −26% (−14%) to −54% (−41%) compared to the base period.
Assessing the impact of climate changes on hydrological drought
The present research aims at assessing the effect of climate change on hydrological drought based on reservoir performance indices capable of multi-aspect description of water shortage (including duration, number, and severity) and also evaluating the impact of climate change on reservoir inflow, storage and water supply. For the purpose, MSUI index and a probabilistic approach as well as data obtained from three GCM models (HADCM3, INCM3 and NCCCSM) under three A1B, A2 and B1 scenarios for the period of 2011–2040 are used.
2 Methodology 2.1 Case Study The ZayandehRud basin with 41,500 km2area is located in the central part of Iran (Fig. 1). The ZayandehRud reservoir with its 1470 Km2volume is the largest surface reservoir in the ZayandehRud River. The average annual total inflow to the ZayandehRud reservoir is about 1600 million cubic meters from which an average annual flow of 600 million cubic meters is transferred from its adjacent Karoon River basin. The ZayandehRud reservoir is the main source of water for supplying irrigation demands of the basin as well as domestic and industrial demands of Isfahan metropolitan area (Table 1). More than 70% of water demand from the ZayandehRud basin is related to agriculture sector. The monthly hydro-metrological data of Koohrang station including precipitation, temperature, and stream flow (inflow to reservoir) are used continuously from 1987 to 2007. Also, the reservoir data including reservoir storage, release, and demands are collected continuously from 1987 to 2007 (Ministry of Energy).
2.2 Research Flowchart To better follow the methodology of the research, its general flowchart is presented in this section (Fig. 2) that applied methods will be explained in the following.
Fig. 1 Zayandehrud river basin and its related demands
Khajeh S. et al. Table 1 Domestic, industrial, agricultural, and environmental monthly demands (MCM) Month
Sep.
Oct.
Nov.
Dec.
Jan.
Feb.
Mar.
Apr.
May.
Jun.
Jul.
Aug.
And domestic industrial Agricultural Environmental
12 56 3
21 98 3
18 84 3
9 42 3
9 42 3
15 70 3
30 140 3
42 196 3
45 210 3
36 168 3
36 168 3
27 126 3
2.3 Hydrological Drought To consider the hydrological indicators in addition to meteorological ones can be important in improving the efficiency of drought assessments. A variety of indicators has been developed for hydrological drought evaluation; in this study, a probabilistic approach and the MSUI index are used.
2.3.1 Probabilistic Approach According to the methodology, thresholds are defined as available storages in the system, S, requiring to satisfy a fraction, f, of demand in a time horizon, h, with a given probability, p. These parameters depend on several factors including type of demand in the system (urban, irrigation, hydropower, etc.), reliability of the current water supply system, vulnerability of demand to deficits of a certain magnitude, etc. They should be delimited in negotiations with stakeholders. The proposed approach can be used with any water resources simulation model. In this study, a
Producing climate data (LARS-WG model)
Data collection (Precipitation, Temperature, inflow,…)
Calculation of Probability approach
Calculation of MUSI, SUI, DRI
Calibration of IHACRAS model
Calculation runoff for further
Selection of suitable Index (MSUI) Calculation of Probability approach Drought Classification of Probability approach
Drought Classification of MSUI
Comparison of MUSI and Probability approach
Comparison of MUSI and Probability approach
Comparison of baseline and futher
Fig. 2 Research flowchart
Calculation of MUSI
Assessing the impact of climate changes on hydrological drought
simplified water balance model is used in which a low evaporation loss is assumed. The model was initially employed to estimate the minimum storage volume required every month to satisfy f percent of the demand during h time horizon. For each month in the simulation period, the minimum storage volume of reservoir for satisfying f percent of the demand during h time horizon is determined as drought warning threshold. Then, an estimate of the cumulative probability distribution of required reservoir storage is obtained (to satisfy the f percent of the demand during h time horizon) by sorting the values of determined storage for each month.
2.3.2 Composite Indices These indices are defined on the basis of the performance of water sources system. They are used to evaluate reservoirs performance. Generally, failures in the operation of a reservoir have many aspects: extent, number and severity (Jain 2010) that indicators such as Reliability, Resiliency and Vulnerability (single indices) show these aspects that will be explained in the following (Hashimoto et al. 1982):
Reliability Water supply reliability is the probability that the available water supply meets the water demand during the period of simulation. For each time period t, deficit Dt is positive when the water demand XDt is more than the water supply XSt; if the water supply is equal to water demand (XDt = XSt), deficit is zero (Dt = 0) (Loucks 1997). Dt ¼
X Dt −X S t 0
if if
X Dt > X Dt X Dt ¼ X Dt
ð1Þ
The most widely accepted and applied definition for water resources systems is occurrence reliability, which is the portion of time that the water demand is fully supplied (i.e. non-failure state, NF) and can be estimated as: Rel ¼ 1−
No:ofdaysDt > 0 n
ð2Þ
WhereDt is water deficit on day t and n is the total number of time intervals (days).
Resilience Resilience (Res) is a measure of how fast a system is likely to return to a satisfactory state (i.e. NF state) once the system has entered an unsatisfactory state (i.e. F state). Hashimoto et al. (Hashimoto et al. 1982) define resilience as a conditional probability: Res ¼
PfS t ∈NF; S t−1 ∈F g PfS t ∈F g
ð3Þ
Where St is the system state variable under consideration. Resilience is the probability that a successful period follows a failure period (the number of times Dt = 0follows Dt > 0) for all failure periods (the number of times Dt > 0toccurred). This statistic assesses the recovery of the system once it has failed: Res ¼
No:of days Dt follows Dt > 0 No:of days Dt > 0 occurred
Where Dt is water deficit on day t.
ð4Þ
Khajeh S. et al.
Vulnerability Vulnerability expresses the severity of failures. Vulnerability can be expressed as (Blenkinsop and Fowler 2007) the average failure ((Fujihara et al. 2008; Loucks and Van Beek 2005; Sandoval-Solis et al. 2011) the average of maximum shortfalls over all continuous failure periods (McMahon et al. 2006); and (Hashimoto et al. 1982) the probability of exceeding a certain deficit threshold. This paper uses the first approach, the expected value of deficits, which is the sum of the deficits, Dt, divided by the deficit period, the number of times (days) Dt > 0 occurred. Dimensionless vulnerability is calculated by dividing the average daily deficit by the average daily water demand (WD): Vul ¼
∑ Dt =No:of days Dt > 0 occurred
t¼n t¼0
ð5Þ
WD
where Dt is water deficit on day t and n is the total number of time intervals (days); WD is the average daily water demand. In the recent past, some attempts (Loucks 1997; Zongxue et al. 1998) have been made to quantitatively represent sustainability of water resources managements by using the composite indices which are composed of the three single indices. Loucks (Loucks 1997),provided a combined index called Sustainability index (SUI). The index has been used by other researchers, as well.()Loucks 1997; McMahon et al. 2006; Ray et al. 2010; Sandoval-Solis et al. 2011)). 1
SUI ¼ ðRel Res ð1−VulÞÞ 3
ð6Þ
In this regard Rel is Reliability, Res is Resiliency and Vul is Vulnerability. SUI’s values vary from 0–1and the value closer to 1 means the condition of water shortage is less serious. In another study, an integrated risk index, drought risk index (DRI), was proposed as a linear weighted function of Reliability, Resiliency and Vulnerability (Zongxue et al. 1998): DRI ¼
1 1 1 ð1−RelÞ þ ð1−ResÞ þ ðVul Þ 3 3 3
ð7Þ
The values of this index are between 0 and 1, too, while a value closer to 1means the condition of water shortage is more serious. Another indicator called Modified the Sustainability index (MSUI) was created (Yang et al. 2012): MSUI ¼
13 1−Rel ð1−ResÞ Vul
ð8Þ
As DRI, the MSUI’s value closer to 1 means the condition of water shortage is more serious.
2.4 Climate Change This section consists of two stages: 1) producing climate data and 2) rainfall-runoff model, which are briefly described below.
Assessing the impact of climate changes on hydrological drought
2.4.1 Producing climate Data (LARS-WG model) In this study, in order to down scale the data of GCM, LARS-WG software was used. LARS-WG is a stochastic weather generator used to generate long-term weather data series to study the assessment of climate change. The model produces synthetic daily time series of maximum and minimum temperatures, rainfall and solar radiation. The weather generator uses observed daily weather for a given site to determine parameters specifying probability distribution for weather variables as well as correlations between the variables. The generation procedure to produce synthetic weather data is then based on selecting values from the appropriate distributions using a pseudo-random number generator. The weather generator distinguishes dry and wet days depending on rainfall values. Rainfall is modeled using semi-empirical probability distributions for each month for the lengths of series of wet and dry days and for the amount of rainfall on a wet day. A semi-empirical distribution Emp = {e0, ei, hi, i = 1,…, 10} is a histogram with 10 intervals, (ei–1, ei), where ei–1 < ei and hi denotes the number of events from the observed data in the ith interval. The histogram has the effect of slightly smoothing the exact distribution of the empirical values. Minimum temperature, maximum temperature and radiation are related to the amount of cloud cover band. LARS-WG uses separate wet and dry day distributions for each of these variables. The normal distribution is used for the temperature variables with the mean and standard deviation taking into account the daily variation estimated by finite Fourier series of order 3. Time auto-correlations used for minimum and maximum temperature are constant throughout the year for the particular site. The cross-correlation of the standardized residuals from the daily mean is pre-set for all sites in the model. Semi-empirical distributions with equal interval size are used for estimating solar radiation for a given set of parameters. LARS-WG produces synthetic data on a daily time step by first determining the rainfall status of the day. In this research, 5th edition of LARS-WG model is used that contains in formation of various general circulation models. According to the observations data of the past and irrespective of any climate change for the basic period (1987–2007) precipitation and temperature data were simulated. Afterwards, the outputs of the model, including the minimum and maximum daily temperature, precipitation and their standard deviation were compared with basic data. After reviewing the results of the assessment and ensuring the ability of LARS-WG model to simulate meteorological data, as a consequent this model was implemented for down scaling statistical data ofHADCM3, INCM3andNCCCSM models for the 2011–2040 periodusingA1B, A2 andB1 scenarios, and finally the daily values of the parameters were generated for future periods.
2.4.2 Rainfall-Run off model (IHACRES Model) Development of mathematical models, relating the regional precipitation to the runoff, has been a major focus of surface water hydrology for many decades. There are different hydrological models used for rainfall- runoff modeling with different characteristics and limitations. The IHACRES model is developed by Jakeman and Hornberger (Jakeman and Hornberger 1993). This model requires limited input data including basin size (m2), a time series of rainfall, stream flow data for model calibration and a surrogate variable representing evaporation. In this study the monthly mean of air temperature is used as a representative of evaporation. The first rainfall (rk) is converted into effective rainfall (uk) using a non-linear loss
Khajeh S. et al.
module. The underlying conceptualization of this module, in converting rainfall to effective rainfall, is that the basin wetness varies with recent rainfall and temperature. uk ¼ s k * r k
ð9Þ
where sk (basin wetness index) is computed at each time step k on the basis of recent rainfall and temperature as follows: 1 ð10Þ sk ¼ C r k þ 1 þ sk−1 s0 ¼ 0 τw ðtk Þ τw ðtk Þ ¼ τw e0:062 f ðR−tk Þ τw ðtk Þ > 1
ð11Þ
R is the reference temperature and C is determined based on the mass balance between effective rainfall and runoff in the calibration period. Two major parameters in this model are τw and f. Parameter τw (the river basin drying time constant) is the value of τw (tk) at a reference temperature tk that controls the rate in which the basin wetness index (sk) decays in the absence of rainfall. Parameter f (the temperature modulation factor) controls the sensitivity of τw (tk) to temperature. In the second module a linear unit hydrograph (UH) module converts effective rainfall to stream flow xk. The linear module allows the application of the well-known unit hydrograph theory which conceptualizes the river basin as a configuration of linear storages acting in series and/or parallel. The configuration of linear storage in the UH module which is allowed in IHACRES includes a single storage or two storage units, in series or parallel. The optimal pair of (τw,f) are identified by trial and error for a given configuration of simple UH’s and a given value of the pure time delay between rainfall and runoff occurrence. Then the model automatically estimates the relevant parameters for a subsequent simulation. In IHACRES model, monthly data of temperature and precipitation of Koohrang station were used. After checking out the performance of the model for different periods, the period from1987 to 1993 and the period from 1994 to 1999 were selected for calibration and validation, respectively. Also for data analysis and model evaluation, criteria such as the coefficient of determination (R2),root mean square error (RMSE) and the mean absolute error (MAE) were used. Their relationships are presented below:
R2 ¼
1 n
2 ∑nm¼1 Xp −μ0 Xp −μ0 σXp σX0
RMSE ¼
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u n u ∑ Xp −X0 2 t m¼1
n
ð12Þ
ð13Þ
n ∑ Xp −X0
MAE ¼
m¼1
n
ð14Þ
Where X is the simulated data, μ average of data, σ the standard deviation and n is the number of data. Subscripts p and or e presents the simulated data and observed data,
Assessing the impact of climate changes on hydrological drought
respectively. R2 represents the linear relationship between simulated and observed data which is between 0 and 1. The much closer to 1, R2 represents a stronger linear relationship between simulated data and observed data.
3 Results and Discussion As a matter of fact, a probabilistic approach and MSUI index are used in this section ford rought assessment. Then, the methods’ performance is evaluated given the impact of climate changes.
3.1 Drought classification using probabilistic approach In this study, f values for the first, second, and third scenarios are 80, 60and30,respectively (Hashemi 2011). The values of p and h in the set three scenarios are considered as 100% (the best probability)and6months(according to the reservoir operation in this basin), respectively. In each year, the minimum storage volume for satisfying 80, 60,and30% of demand for the next 6 months is determined. Then, the cumulative probability distribution for stratifying 80, 60, and 30 percent of demand for each month is calculated (Fig. 3). Now, the required minimum storage volume for stratifying 80, 60, and 30% of demand for each month (Fig. 4) can be calculated using these distributions.
3.2 Drought Classification using MSUI Based on the research purposes, the behavior of combined indicators i e. DRI, SUI, and MSUI were estimated for selecting the appropriate indicator. It is noteworthy that these indicators should behave consistent with the variables involved in their calculations (Jain 2010; Kjeldsen and Rosbjerg 2004). The behavior of these three indicators was analyzed by changing the parameters with some influences on them: (Blenkinsop and Fowler 2007) Inflow to the reservoir, (Fujihara et al. 2008) water demand, and (Hashimoto et al. 1982) storage capacity of the reservoir (Fig. 5). According to the above mentioned results, the MSUI index was selected as an appropriate indicator for analyzing the considered region. In order to classify the levels of drought by 1.2
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
1
Probability
0.8 0.6 0.4 0.2 0 150
250
350
450
550
650
Storage(mcm)
Fig. 3 The cumulative probability distribution for stratifying 80% of demand for 6 months
750
Khajeh S. et al. 800 f=80%, p=1
Level 1
700
f=60%, p=1
f=30%, p=1
Storage (mcm)
600
Level 2
500 400
Level 3
300 200
Level 4
100 0 Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Fig. 4 Probabilistic threshold for different levels of drought based on satisfying 100%of demand
Fig. 5 Analysis results of each index monotonic behavior
Assessing the impact of climate changes on hydrological drought
MSUI index, determination of the indicators’ thresholds is necessary. To define the thresholds, different amounts of water deficiency < 20, 20–40, 40-70and > 70%wereconsidered for levels1 to4of drought (based on 80, 60, and 30% for demand satisfaction), respectively. A six-month period was chosen for demand satisfaction. In order to be coordinated with the operation of water resources in the region, the MSUI values and lack of water for anytime point in the six month period were calculated from 1973 to 2003.Finally,corresponding to the shortage of water supply for each level of drought, a range was selected for MSUI which is presented in the below table (Table 2).
3.3 Climate Data Generation (LARS-WG Model) Comparison test (T-test) was used in order to assess the ability of the model for simulating meteorological data. The results showed that there was not any significant difference between the simulated and observed values at the 95% confidence level in each month of the year. To downscale the data of GCM (HADCM3, INCM3, NCCCSM) statistically, the A1B, A2, andB1scenarios were used. Consequently, daily values of the parameters were generated. In order to evaluate the ability of the model to produce the meteorological data of stations in 2011–2040 period, the plots were drawn and the results were analyzed (Figs. 6, 7). The results showed an increase in all three models as well as all months; the HADCM3model indicated1.18 °C maximum increase in temperature. Under the A1 B scenario, the least temperature rise was0.24 °C occurred in Julyand January (Fig. 6). Unlike temperature with its always increasing trend, diagrams such as Fig. 4 indicate that in some months during2011–2040, the basin would witness a decline in rainfall, and in other cases, the precipitation would be increased in all three models under all three scenarios. It is obvious that in June, July, August, and September, there is not any significant increase or decrease trend. In general, the greatest increase in rainfall was 55.36mmintheINCM3 model underB1 Scenario and its maximum reduction was 50mminthismodelunder theA2scenario (Fig. 7).
3.4 Rainfall-runoff Simulation (IHACRES Model) In order to calibrate and validate the IHACRES model, at first the monthly runoff was estimated for the up stream basin of Zayandeh Rudmaking up the main part of runoff of the basin. Notably, the flows of Koohrangtunnels (tunnels no. 1 and 2) and Cheshme Langan tunnel were subtracted from the total in put to the dam of the river during 1987– 2007.Afterward,the best performance was obtained from values of f = 2.25, τw = 7 and a single hydrograph considering various combinations of calibration and verification periods. Table 3 and Fig. 8 show the model in the calibration and verification periods. According to Fig. 8,althoughthemodelcould not simulate some maximum (peak) months, but with respect to the correlation coefficient (R2) value as 0.769 during the calibration and 0.6
Table 2 Classification of MSUI values for different classes of drought
Drought
MSUI classifications
Level 1 Level 2 Level 3 Level 4
0 ≤ MSUI < 0.4 0.4 ≤ MSUI < 0.65 0.65 ≤ MSUI < 0.75 0.75 ≤ MSUI ≤ 1
Khajeh S. et al.
1.4
B1
Dec
Oct
A2
Nov
(B)
Month
Sep
(A)
Jan
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
0.00
Jul
0.20
Aug
0.40
Jun
0.60
Apr
0.80
May
1.00
A1B
1.4 1.2 1 0.8 0.6 0.4 0.2 0
B1
Mar
A2
Feb
A1B
1.20
Temperature (oc)
Temperature (oc)
1.40
Month
A1B
A2
B1
Temperature (oc)
1.2 1 0.8 0.6 0.4 0.2
Dec
Nov
Oct
Sep
Jul
Aug
Jun
Apr
May
(C)
Mar
Jan
Feb
0
Month
Fig. 6 Changes of monthly mean temperature in the next period (2011–2040) under a HADCM3,b INCM3 and c NCCCSM Models compared to the baseline
60
A1B
-40
Month
A1B
Dec
Nov
Oct
Sep
Aug
Jul
Jun
-40
Month
A2
B1
40 20
Dec
Oct
Nov
Sep
Jul
May
Apr
Mar
-20
Feb
0
Jan
Precipitaon change (mm)
60
(C)
-20
(B) -60
Aug
(A)
Jun
-60
0
May
Dec
Oct
Nov
Sep
Jul
Aug
Jun
Apr
May
Mar
-20
Jan
0
B1
20
Apr
20
A2
40
Mar
40
Feb
B1
Jan
A2
Precipitaon change (mm)
A1B
Feb
Precipitaon change (mm)
60
-40 -60
Month
Fig. 7 Changes of monthly precipitation in the next period (2011–2040) under a HADCM3, b INCM3, and c) NCCCSM Models compared to the baseline
Assessing the impact of climate changes on hydrological drought Table 3 Statistical evaluation criteria for the observed and simulated values of flow by IHACRES model MAE
RMSE
R2
3.71 9.74
1.21 3.89
0.769 0.6
Calibration Validation
during the validation periods, as well as the values of other parameters presented in Table 3, the model shows an acceptable performance level for simulating the flow of the river basin. After calibrating the model, the time series of monthly run off of the basin were simulated for the generated temperature and precipitation by means of LARS-WG for each model underA1B, A2 andB1 scenario during 2011–2040usingthe IHACRES hydrology model. To calculate the total input to the dam of the river, the discharge values obtained from the IHACRES model were collected along with the discharges of Koohrang tunnels (tunnels no. 1 and 2) and Cheshme Langan tunnel. The long-term mean annual runoff(30 years interval) in 2011–2040 periodfortheHADCM3 model was reduced12.51m 3/sin theA1B scenario, 10.03m3/s in theA2 scenario and 8.62 m3/s in the B1 scenario compared to 1973-2003 period. On the other hand, these values are 5.2, 8.08, and 7.09 for the INCM3model and 9.67, 11.26, and 10.49 s for the NCCCSM model, respectively (Table 4).
3.5 Evaluation of hydrological drought Using the values obtained from the inputs to the dam of Zayandeh Rudunder different climate change scenarios, the warning level changes were determined based on the probabilistic approach and MSUI index. Climate change has a significant effect on the reduction of level 1 (normal level) warnings and increase of other levels of droughts; the number of level 1 warnings for MSUI index reduced from189inthe baseperiodto109, 124, and 130in theHADCM3modelunder A1B, A2 andB1scenarios, respectively (Fig. 9a). However, 58 warnings of drought (level 4) in the base period increased to 59, 67, and61 in the INCM3modelunderA1B, A2 andB1 scenarios, respectively (Fig. 9b).
Fig. 8 Observed and modeled time series of run off by IHACRES in the calibration (1987–1993) and validation(1994–1999) periods
Khajeh S. et al.
Table 4 Long-term means annual run off for the base period (1973– 2003) and future period (2011– 2040) under different models.
Period 1973–2003 2011–2040
Base HADCM3 INCM3 NCCCSM
A1B
A2
B1
48.67 36.16 43.47 39
48.67 38.64 40.59 37.41
48.67 40.04 41.57 38.18
In the probabilistic approach, the reduction trend ofthelevel1in the models and scenarios is clearly visible. In particular, the normal level decreased from126inthebase conditionto29 in theHADCM3 model under the A1B scenario. It can be seen that the level4 of drought increases from60 cases in the base period to74 cases in the NCCCSMmodelunderA2 scenario, or the warnings of level2 of drought rise from 93 casesto163 cases in the same model under B1 scenario Fig. 10.
4 Conclusion This study aims at determining the impact of climate change on hydrological drought in the river basin of the ZayandehRud. Therefore, the data about climate change in HADCM3, INCM3 and NCCCSM models (A1B, A2, andB1 scenarios) were downscaled over 2011– 2040 periods using the LARS-WG model. The results of this research are provided as follows:
200
200
Observaon INCM3-AIB INCM3-A2 INCM3-B1
150
Number of warning
Number of warning
Observaon
100 50
150 100 50 0
0
(A)
Level 1
Level 2
Level 3
Level 4
Number of warning
200
Level 1
Level 2
Level 3 Level 4
(B) Observaon NCCCSM-AIB NCCCSM-A2 NCCCSM-B1
150 100 50 0
(C)
Level 1
Level 2
Level 3
Level 4
Fig. 9 a Comparison of warnings for different levels of drought in the base period(1973–2003) and the next period (2011–2040) under a HADCM3, b INCM3, and c NCCCSM models based on MSUI index
Assessing the impact of climate changes on hydrological drought
200
200
Observa on INCM3AIB
Number of warning
Number of warning
Observaon 150 100 50 0
(A)
150 100 50 0
Level 1 Level 2 Level 3 Level 4
Number of warning
200
(B)
Level 1
Level 2
Level 3
Level 4
Observaon NCCCSM-AIB NCCCSM-A2 NCCCSM-B1
150 100
(C)
50 0 Level 1
Level 2
Level 3
Level 4
Fig. 10 Comparison of warnings for different levels of drought in the base period (1973–2003) and the next period (2011–2040) under a HADCM3, b INCM3, and c NCCCSM Models based on the probabilistic approach
–
– – –
–
–
To determine an appropriate indicator for the under study region, three composite indicators of reservoir performance (DRI, SUI, and MSUI) were compared. The MSUI index was selected because of its more uniform behavior in relation to changes in (Blenkinsop and Fowler 2007) inflow to the reservoir, (Fujihara et al. 2008) water demand, and (Hashimoto et al. 1982) storage capacity of the reservoir. Generally, the results of climate change evaluation demonstrated1.5 °C rise in monthly temperature and fluctuation of rain fall overtime. To assess the impact of changes on runoff, the IHACRES model was calibrated using data from Koohrang station. Afterwards, the data generated from climate change were entered into the model. The results showed a general 12.5% decline of inflow to reservoir. In all of the considered models in both probabilistic approach and MSUI index method, the A2 scenario showed a more severe level of drought than other scenarios. With regard to the condition of the region in the past few years, it can be concluded that this scenario matches better to the region’s condition. The comparison of the indices regarding the levels of drought showed that the percent of increase in normal levels (mild, severe, and very severe) by using the MSUI index was considerably higher than the results of probabilistic approach. Given that the MSUI index accounts the number and severity of drought in addition to its quality, it can be concluded that the MSUI index is more accurate and sensitive than the probabilistic approach. Using the changes to these indices, a marked reduction of normal levels and dramatic increases of drought levels were obtained for future period. Therefore, it is time for researchers to focus their attention toward more severe cases of drought periods with greater intensity in the catchment area of the case study.
Khajeh S. et al. Compliance with Ethical Standards Conflict of Interest No conflict of interest.
References Blenkinsop S, Fowler HJ (2007) Changes in drought frequency, severity and duration for the British isles projected by the PRUDENCE regional climate models. J Hydrol 342:50–71 Fujihara Y, Tanaka K, Watanabe T, Nagano T, Kogira T (2008) Assessing the impacts of climate change on the water resources of the Seyhan River basin in Turkey: use of dynamically downscaled data for hydrologic simulations. J Hydrol 353:33–48 Hashemi A (2011) Linking drought indicators to policy actions in drought management of Zarineh-Rud water resources. Tarbiat Modares University, Master's thesis Hashimoto T, Stedinger JR, Loucks DP (1982) Reliability, resiliency, and vulnerability criteria for water-resource system performance evaluation. Water Resour Res 18(1):14–20 Jain SK (2010) Investigating the behavior of statistical indices for performance assessment of a reservoir. J Hydrol 391:90–96 Jakeman AJ, Hornberger GM (1993) How much complexity is warranted in a rainfall-runoff model? Water Resour Res 29(8):2637–2649 Kjeldsen TR, Rosbjerg D (2004) Choice of reliability, resilience and vulnerability estimators for risk assessments of water resources systems. Hydrological Sciences–Journal–des Sciences Hydrologiques. 49(5):755–767 Lespinas F, Ludwig W, Heussner S (2014) Hydrological and climatic uncertainties associated with modeling the impact of climate change on water resources of small Mediterranean coastal rivers. J Hydrol 511:403–422 Loucks DP (1997) Quantifying trends in system sustainability. Hydrological Sciences–Journal–des Sciences Hydrologiques 42(4):513–530 Loucks DP, Van Beek E (2005) Water resources systems planning and management: an introduction to methods, models and applications. United Nations Educational, Scientific and Cultural Organization (UNESCO) Loukas A, Vasiliades L, Tzabiras J (2008) Climate change effects on drought severity. Adv Geosci 17:23–29 McMahon TA, Adeloye AJ, Sen-Lin Z (2006) Understanding performance measures of reservoirs. J Hydrol 324: 359–382 Mishra V, Cherkauer KA (2009) Assessment of drought due to historic climate variability and projected future climate change in the Midwestern United States. J Hydrometeorol 11:46–68 Ray PA, Vogel RM, Watkins DW (2010) Robust optimization using a variety of performance indices. In: world environmental and water resources congress, ASCE, providence, RI, 4710–4719, 16–20 may Sandoval-Solis S, McKinney D, Loucks D (2011) Sustainability index for water resources planning and management. J Water Resour Plan Manag 137(5):381–390 Sheffield E, Wood F (2008) Projected changes in drought occurrence under future global warming from multimodel, multi-scenario, IPCC AR4 simulations. Clim Dyn 31:79–105 Vrochidou AEK, Tsanis IK, Grillakis MG, Koutroulis AG (2013) The impact of climate change on hydrometeorological droughts at a basin scale. J Hydrol 476:290–301 Wang D, Hejazi M, Cai X, Valocchi AJ (2011) Climate change impact on meteorological, agricultural, and hydrological drought in central Illinois. Water Resour Res 47(9):1–13 Yang TC, Chen C, Kuo CM, Tseng HW, Yu PS (2012) Drought risk assessments of water resources systems under climate change: a case study in southern Taiwan. Hydrology and Earth System Science Disussions 9: 12395–12433 Zongxue X, Jinno K, Kawanura A, Takesaki S, Ito K (1998) Performance risk analysis for Fukuoka water supply system. Water Resour Manag 12(1):13–30