 |
| Titel |
Using land subsidence observations for groundwater model calibration |
| VerfasserIn |
Nesrin Tufekci, Gerrit Schoups, Mohamed Al Faitouri, Pooja Mahapatra, Nick van de Giesen, Ramon Hanssen |
| Konferenz |
EGU General Assembly 2017
|
| Medientyp |
Artikel
|
| Sprache |
en
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 19 (2017) |
| Datensatznummer |
250142727
|
| Publikation (Nr.) |
 EGU/EGU2017-6380.pdf |
|
|
|
|
|
| Zusammenfassung |
| PS-InSAR derived subsidence and groundwater level time series are used to calibrate a
groundwater model of Tazerbo well field, Libya, by estimating spatially varying
elastic skeletal storage (Sske) and hydraulic conductivity (Hk) of the model area.
Tazerbo well field is a part of the Great Man-Made River Project (GMMRP) designed
with 108 wells and total pumping rate of 1 million m3/day. The water is pumped
from the deep sandstone aquifer (Nubian sandstone), which is overlaid by a thick
mudstone-siltstone aquitard. Pumping related deformation patterns around Tazerbo well field
are obtained by processing 20 descending Envisat scenes for the period between 2004
and 2010, which yield a concentrated deformation around the well field with the
maximum deformation rate around 4 mm/yr. The trends of time series of groundwater
head and subsidence are in good agreement for observation wells located in the
vicinity of the pumping wells and the pattern of subsidence correlates with the
locations of active wells. At the beginning of calibration, different pairs of Sske and Hk
are assigned at observation well locations by trial and error so that the simulation
results of the forward model would approximate heads and mean linear deformation
velocity at these locations. Accordingly, the estimated initial parameters suggest
relatively constant Hk(5 m/d) and increasing Sske from south to north (1x10−6
m−1 – 5x10−6 m−1). In order to refine their spatial distribution, representative
values of Sske and Hk are assigned at 25 equidistant points over the area, which are
restricted by the predetermined values. To calibrate the parameters at assigned locations
UCODE is used along with MATLAB. Once the convergence is achieved the estimated
parameter values at these locations are held constant and new “in between – equidistant”
locations are determined to estimate Sske and Hk in order to spatially refine their
distribution. This approach is followed until the relation between observed and
simulated results cannot be improved anymore. The final results, show good relation
in the areas close to well sites, while difference between simulated and observed
values increases towards the southeast boundary of PS-InSAR observations. This
divergence can be explained by i) the uncertainty in the parameters at the southeast
part of the model area, due to lack of deformation observations, ii) Sske may have
been underestimated at the locations with thin aquifer and thick aquitard material. |
|
|
|
|
|
|