Via

Invo

egen

|

Kopt

ekst

en

Voett

ekst

invoe

gen

Suba

fdelin

g<2s

patie

s>|<

2spat

ies>

Titel

van

de

prese

ntati

e

1

High Resolution Global Scale Groundwater Modelling

Department of Physical Geography – Faculty of Geosciences

Rens van Beek

Inge de Graaf, Edwin Sutanudjaja, Yoshi Wada & Marc Bierkens

Limits to global groundwater consumption: effects on low flows and groundwater levels

2

De Graaf et al., 2014

3

Simulated water stress index for 2000 (Wada et al., 2011)

Towards a high resolution global hydrological model

4

Global-scale simulations at 5 arc minutes (~10 x 10 km at equator)

River discharge Domestic water use

What has been the effect of abstractions on groundwater levels?

5

• Simulation of groundwater head dynamics;

• Lateral flow (exchange between cells) cannot be ignored at finer spatial resolutions.

Challenges to the construction of a physically based global-scale groundwater model: • Quality of available global datasets:

- Surficial hydro-lithology; - Aquifer thickness estimates.

6

Model lay-out

• 5’ resolution; • Steady-state; • Offline coupling to MODFLOW; • Coupling to iMOD under construction.

Sutanudjaja et al., 2011

7

Land-surface model

Available global datasets

Gleeson et al. 2010

Hartmann and Moosdorf 2012

Hydro-lithology

Conductivity

Model Input

Groundwater recharge

Surface water levels • Imposed as average long-term levels for

lakes, reservoirs and lakes; • Based on discharge for rivers and imposed

by means of the RIVER package using uniform conductivity;

• DRAIN package is used where no main river channel is available.

8

range Land surface

Sediment basin

50 m

Floodplain elevation

range

1) 𝐹′ 𝑥 = 1 − 𝐹 𝑥 − 𝐹𝑚𝑖𝑛

𝐹𝑚𝑎𝑥 − 𝐹𝑚𝑖𝑛

F’(x) is spatial frequency distribution of elevation above the floodplain 2) Associated Z-score

𝑍 𝑥 = 𝐺−1(𝐹′ 𝑥 )

Where G-1 is the inverse of the standard normal distribution.

Sediment basin aquifers: delineation and depth (1)

9

3) Using case studies in the US

• range of aquifer thickness

• average coefficent of variation

• Aquifer thickness is assumed to be log-normally distributed (positive skew)

4) 𝑙𝑛𝐷 = 𝑈(𝑚𝑖𝑛;𝑚𝑎𝑥)

𝑌 𝑥 = 𝑙𝑛𝐷 × (1 + 𝐶𝑣𝑙𝑛𝐷Z x )

𝐷 𝑥 = 𝑒𝑌(𝑥)

dmax

dmin

Sediment basin aquifers: delineation and depth (2)

10

Cumulative probability of aquifer depth

Average simulated aquifer thickness

12

Transmissivity (m2d-1)

0.5 5 15 40 >100

𝑇 𝑥 = 𝑘0𝑒−𝑧//λ

𝐷(𝑥)

0

𝑑𝑧

Aquifer-scale transmissivities

Steady-state water table depth

13

14

Validation on groundwater wells

0.25-2.5 320- 640 > 640

Observed GW heads

Sediment basins

All wells

15

Validation on groundwater wells

0.25-2.5 320- 640 > 640

Observed GW heads

Sediment basins

All wells

Relative residuals

𝑅𝑟𝑒𝑙= 𝐻𝑠𝑖𝑚−𝐻𝑜𝑏𝑠

𝐻𝑜𝑏𝑠

16

0.01 0.1 1 10 100 1000 Years - Months years decades centuries millennia

Flow paths and travel times

• Suitable method to develop aquifer schematization and properties for data poor environments;

• The large scale-distribution of groundwater levels is captured; starting point to assess groundwater level fluctuations;

• Confirms the relevance of including lateral flow in global scale hydrological models at finer resolutions.

17

Conclusions

What is the effect of past and future abstractions on groundwater levels?

Major limitations, currently being addressed: • Steady-state;

• Single, unconfined layer;

• Coupling of surface and groundwater.

18

Thank you for your attention

http://www.hydrol-earth-syst-sci-discuss.net/11/5217/2014/hessd-11-5217-2014.pdf

19

20

Europe USA

Validation groundwater depths