application of groundwater modeling to predict … · application of groundwater modeling to...

The International Conference of “Multidisciplinary Approaches on UN Sustainable Development Goals” (UNSDGs) 28th - 29th December 2016

Application of Groundwater Modeling to predict Drought Impacts on Groundwater Resources in the

Garbaygan Plain, Iran

Manfred Koch and Majid Taie Semiromi

Department of Geohydraulics and Engineering HydrologyUniversity of Kassel, Germany

[email protected], [email protected]


Table of contents

2

1. Introduction1.1. Drought and classification of drought1.2. Groundwater drought1.3. Forecasting of groundwater drought

2. Material and methods2.1. Study area2.2. Geological situation and boundary condition2.3. Rainfall and groundwater level data2.4. Groundwater flow model2.5. Groundwater drought prediction2.6. Drought scenarios2.7. Rainfall and recharge relationship

3. Results3.1. Observed rainfall and groundwater level changes3.2. Model calibration3.3. Model validation3.4. Groundwater budget variation as a function of the Z-Score index

4. Conclusions5. References


1. Introduction

3

1.1. Drought and classification of drought

Drought is a complex phenomenon, which involves different human and natural factors that determine the risk and vulnerability to drought. Although the definition of drought is very complex, it is usually related to a long and sustained period during which water availability becomes scarce.

Droughts are calssified as fallow:

� Meteorological drought - lack of precipitation� Agricultural drought - lack of soil moisture� Hydrologic drought -reduced streamflow or groundwater levels� Socio-economic drought- migration of the people due to the impacts of drought

1.2. Groundwater droughtIt is a kind of hydrological drought which is resulterd from less recharge and accordingly drawdown in groundwater level

1.3. Forecasting of groundwater droughtGroundwater drought can be predicted using statistical models (e.g. autoregressive–moving-average (ARMA)) and groundwater flow models (e.g. MODFLOW)


Table of contents

4






2. Material and methods2.1. Study area

5

The Garbaygan plain is situated in thesoutheast of Iran and covers an area ofabout 83 km2. Geographically, the areaextends from 28° 35´ E to 28° 41´ Elatitude and from 53 ° 53´ N to 53° 57´ Nlongitude (see Fig. 1). Geologically, thisplain extends on an alluvial fan, wherethe major Bishe Zard and two otherephemeral streams enter the Garbayganplain. Because of this ephemeral-streamsystem, surface- groundwaterinteractions play a considerable role inthe area (Fatehi Marj, 2000). Thethickness of the alluvium varies from 19to 58 meters, so that the transmissivity,- though the hydraulic conductivity ofthe alluvium may be more or lessconstant – varies spatially throughout theaquifer.

Figure 1. Location of the Garbaygan plain, Iran,with aquifer area in green (Ghahari andPakparvar, 2007).


2. Material and methods2.2. Geological situation and boundary condition

6

Figure 2. Left panel: Google image with boundary lines of the aquifer (red:Neumann (no-flow)- condition; blue: Cauchy BC) and location of the waterspreading project. Right panel: Geology.


2. Material and methods2.3. Rainfall and groundwater level data

6

� Groundwater level hydrographs are often the most important source of information about the hydrogeological conditions of aquifers

� A propagating meteorological drought through a groundwater system, known as groundwater drought, is also detectable by means of such a groundwater hydrograph, even though it is difficult to distinguish whether a groundwater level drop is not also due to over- abstraction (Shahid and Hazarika, 2009).

� To understand how groundwater levels fluctuate in response to the precipitation, the monthly groundwater table, based on a geospatial average of four piezometers across the region, has been plotted, together with the annual rainfall for 16 hydrological years beginning in October, 1992 and ending in September, 2008 in Fig. 4.

.


2. Material and methods2.4. Groundwater flow model

8

1-The MODFLOW groundwater model

� In this study, the groundwater system is simulated numerically by means of the well-known MODFLOW groundwater flow model (McDonald and Harbaugh, 1988).

��

��=

�

��

��

��+

�

� �

��

�+

�

��

��

�+� (1)

2- MODFLOW- model setup� Cell size 100 *100 m, 92 rows and 116 columns� One layer of an unconfined aquifer � No flow and Cauchy boundary conditions (Fig. 2)� April 1978 as the initial condition

3- Model calibration and validation

� Steady state calibration in April 1978� Stress periods: Wet period (December to May) and dry period (from June to

November)� Transient calibration from May1992 to April 2003� Model validation from May 2003 to November 2008


2. Material and methods2.5. Groundwater drought prediction

9

� Z- score index to define Meteoroligical drought

�� =�

��

��

�� + 1

�

�−

�

��+

��

�(2)

where Cs is the skewness coefficient:

� =∑ ��̅��

�

∗"(3)

�� =��̅

"(4)

is the standardized deviate of the precipitation xi (mm) in month i (i=1,..,n).

From Eq. (4), a real deviate for xi is computed through

#� = �� ∗ $ + #̅ (5)

Table 1. Drought classification scale by the Z- Score values Edwards and McKee (1997).

Category ExtremelyWet

Verywet

Moderatelywet

Near normal Moderately dry Severelydry

Extremelydry

Z-Score > 2.00 1.50 to1.99

1.00 to 1.49 −0.99 to0.99

−1.00 to −1.49 −1.50 to−1.99

< −2.00

Value used …… 1.745 …… 0 -1.245 -1.745 ……


2. Material and methods2.6. Drought scenarios

10

Table 1 lists the 7 classes of the standard drought classification (Asadi Zarch and et al.

2014) starting from “extremely wet” and ending at “extremely dry”. As these two extreme

class members, occur only very infrequently, i.e. have very long return periods, they are

not considered further in the subsequent analysis. Furthermore, a “moderately wet” class is

left out, in order to better define distinguished separate classes with the available rainfall

data and, similarly, of the MODFLOW-simulated groundwater levels and storage.

Consequently only 4 drought classes are considered in the Z-score- analysis. To calculate

the monthly precipitation corresponding to each scenario, firstly the mean of each drought

class is computed which leads to a specific value making it possible to compare and to

contrast all the other scenarios in the same situation and to define the upper and lower

truncation levels of each scenario class in a clear way. Afterwards, using (Eq. 5), with �

now denoting the Z-score- value, the corresponding real precipitation limits are estimated

for each drought scenario category of Table 1.


Table of contents

11






3. Results

12

3.1. Observed rainfall and groundwater level changes

Figure 3. Empirical density histogram of the observed rainfall with fitted density kernel.

Figure 4. Fluctuations of the groundwater levels along with the annual rainfall variability.


2. Material and methods2.7. Rainfall and recharge relationship

13

As the study aquifer is recharged by (1) the diffuse recharge as a direct natural recharge by the rainfall and (2) artificial recharge induced by the floods entering the water spreading recharge project, two separate linear regression models are developed between the rainfall (explanatory variable) and the named two recharge processes (response variable) in the transient calibration and validation simulations steps:

% &�= −6.293 ∗ 10�- + 2.763 ∗ 10�/∗ 0(6)

%&1� = 0.0643 + 1.524 ∗ 10�4 ∗ 0 (7)

where Rnat and Rart denote the natural and artificial recharge rate (mm/year), respectively, and P (mm/year) is the amount of the rainfall. Subsequently, the precipitations calculated by Eq. 5 for each of the drought condition scenarios are imported into these regression models to compute the corresponding recharge rates which are then used further as inputs into MODFLOW, to predict the ensuing groundwater levels and storage under these scenarios.


Table of contents

14


2. Material and methods2.1. Case study2.2. Geological situation and boundary condition2.3. Rainfall and groundwater level data2.4. Groundwater flow model2.5. Groundwater drought prediction2.6. Drought scenarios2.7. Rainfall and recharge relationship




3. Results3.2. Model calibration

15

Figure 5. Comparison of observed and simulated groundwater levels in the transient calibration step.


3. Results3.3. Model validation

16

Figure 6. Comparison of observed and simulated groundwater level inthe transient validation step.


3. Results3.4. Groundwater budget variation as a function of the Z-Score index

17

To understand how the groundwater budget responds to the annual precipitation

variability, meteorological droughts defined by the Z-Score index are plotted parallel

with the MODFLOW simulated groundwater budget variation (MMC) (Fig 7). From this

figure one may notice that the groundwater budget follows well the drought index and

this finding is supported by a rather good R2 = 0.65 obtained for the following cubic

polynomial regression model between the Z-score index and the water budget WB.

�5 = −0.986 − 0.075 ∗ �789:; + 0.036 ∗ �789:;� +0.272 ∗ �789:;< (8)


3. Results3.4. Groundwater budget variation as a function of the Z-Score index (Cont.)

18

Figure 7. Variations of the Z-score index and of the simulated groundwater budget over time, with the four colored bars on the right of the vertical dashed line representing water budgets (deficits) predicted for year 2009, assuming one of the four drought scenarios of Table 1 (as indicated by the corresponding Z-score bars), and denoted in the axis label as D1, D2 D3 and D4, for that year.


Table of contents

19


2. Material and methods2.1. Case study2.2. Geological situation and boundary condition2.3. Rainfall and groundwater level data2.4. Groundwater flow model2.5. Groundwater drought prediction2.6. Drought scenarios2.7. Rainfall and recharge relationship




4. Conclusions

20

� The analysis of the groundwater levels shows that after the installation of thegroundwater recharge spreading system, these have been rising until October 1996,after which they have been dropping more or less monotonously.

� The MODFLOW-simulations predict that regardless of four consideredprecipitation/drought scenarios, the groundwater budget is still always negative andthis holds even for very benevolent “very wet” year conditions.

� In addition to the hydrological drought situation, the groundwater exploitation playsalso a key role in the groundwater level drops. Thus, an appropriate groundwatermanagement plan should be provided to effectively limit the groundwater utilization inthe Garbaygan aquifer (Fatehi Marj and Taie Semiromi, 2013).

� To alleviate the drought impacts, two practical and efficient methods arerecommended. One would be to change the traditional irrigation system which isknown to be poorly efficient as large parts of the pumped water are evaporated in thishot and arid region, and the other would be to set up other potential artificial rechargeprojects which, in turn, could be an economical method to also alleviate the negativeimpacts of the frequently occurring floods in this watershed.


5. References

21

Asadi Zarch MA, Sivakumar B, Sharma A. Droughts in a warming climate: A global assessment ofStandardized Precipitation Index (SPI) and reconnaissance drought index (RDI). Journal ofHydrology [Online], Available from:http://dx.doi.org/10.1016/j.jhydrol.2014.09.071. 2014.

Edwards DC, McKee TB. Characteristics of 20th century drought in the United States at multiple timescales. Atmospheric Science Paper. 1997; 634: 1–30.

Fatehi Marj A. Investigation and determination of flood spreading effects using mathematical modelin Fasa Garbaygan plain (Fars Province). Final report of research plan, SoilConversation andWatershed Management Research Institute; 2000.

Fatehi Marj A, and Taie Semiromi M. Forecasting of groundwater table and water budget underdifferent drought scenarios using MODFLOWmodel (Case Study: Garbaygan Plain, Fars Province,Iran). Journal of Water Sciences Research. 2013; 5 (1): 41-54.

Ghahari GR, Pakparvar M. Evaluation of flood water spreading effects on groundwater resources inGarbaygan plain. Scientific-Research Journal on Range and Desert. 2007; 14(3): 368-390.

McDonald MG, Harbaugh AW. A modular three-dimensional finite difference ground-water flowmodel. In: U.S. Geological Survey Techniques of Water-Resources Investigations; 1988.

Shahid S, Hazarika MK. Groundwater drought in the northwestern districts of Bangladesh. WaterResources Management [Online], Available from: DOI 10. 1007//s 11269-009-9534-y; 2009.


Thank you for your attention!Any Question?

application of groundwater modeling to predict … · application of groundwater modeling to...

Documents