groundwater for over 20 years. Consequently the area suffered the highest aquifer-related subsidence rate detected in Europe: >10 cm/yr (Boni et al. 2015).
GEOSTATISTICAL MERGING OF INSAR AND GPS DATA FOR SUBSIDENCE MAPPING Carolina Guardiola-Albert1, Marta Béjar Pizarro1, Ramón P. García-Cárdenas2, Gerardo Herrera García1, Serena Tessitore3 Instituto Geológico y Minero de España, C/ Ríos Rosas, 23. 28003, Madrid, Spain. 2 Departamento de Ingeniería Civil. Universidad Católica San Antonio de Murcia, Campus de los Jerónimos. 30107, Murcia, Spain. 3 University of Naples Federico II, Vía Claudio, 21. 80125, Naples, Italy. 1
Subsidence data
Introduction Agricultural activities in
Several surveys have been conducted
the Alto Guadalentín Basin
to monitor land subsidence in the region,
(south-eastern Spain) led to
such as GPS (Global Positioning System)
an
campaigns
intensive
pumping
of
and
InSAR
(Interferometry
20
Synthetic Aperture Radar) studies. GPS
the
provides the vertical displacement and
area suffered the highest
moderate spatial sampling. Alternatively,
aquifer-related
InSAR data has a high spatial resolution
groundwater years.
for
over
Consequently
rate
detected
>10
cm/yr
subsidence in
but
Europe:
(Boni
et
al.
2015).
Geological setting of the Guadalentín Basin (Boni et al. 2015).
measurements
represent
the
projection of 3D deformation onto the sensor line-of-sight (Tomás et al. 2014).
Objective
Deformation velocity from InSAR data for the period 2003-2010 (color scale) and GPS data for the period 2004-2013 (star symbols). Negative values indicate land subsidence.
Integration of InSAR and GPS is crucial to obtain high accuracy and wide spatial information of land subsidence.
Kriging with external drift The standard methods for combining InSAR and GPS data are based on the correction of the InSAR interferograms to minimize the errors of highresolution three-dimensional surface velocity maps. The high correlation between both sets of data allows applying kriging with external drift interpolation (Goovaerts 1997). This version of kriging takes
into
account
discrepancies
between
GPS
and
InSAR
derived
measurements and predicts subsidence values at unsampled locations. To the author’s knowledge, this is the first application of kriging with external drift to merge InSAR and GPS data.
The geostatistical study was performed using the Isatis geostatistics software (Geovariances 2011). Correlation between deformation rates estimated from InSAR and GPS data.
Interpolated GPS velocity map (cm/yr) applying kriging with external drift.
Cross-validation To validate the interpolated maps the cross-validation method (Isaaks and Srivastava 1989) is applied. The standardized error ((Z-Z*)/σ*), which should follow a normal distribution, characterizes the ability of the interpolation to correctly re-estimate the data values. Outside the interval [-2.5; 2.5] are the 1% extreme values of a normal distribution. To check for conditional bias we plot the errors as a function of the estimates values. Ideally, this scatter should result in no correlation and no increase in estimation variance.
The correction introduced to the InSAR data based on the GPS values peformed with the geostatistical merging method can be seen as the differences between the InSAR data interpolated image and the kriging interpolated map.
In general GPS deformation
rates are greater than InSAR
deformation rates. This different deformation rate could be explained by a deformation phenomenon affecting the region during the period 2011-2013, only covered by the GPS data, or by the GPS errors.
Conclusions
Correlation between GPS velocities estimated with cross validation and GPS data.
GPS velocities estimated in the cross-validation analysis (in color) GPS measured data (with crosses).
Histrogram of the standarized error.
Standarized error versus estimated GPS rates in teh 39 GPS locations.
Based on combination of GPS and InSAR, areas of maximum land subsidence reached in Alto Guadalentín Basin from 2003 until 2013 can be mapped. This map is a valuable supporting tool to understand
the relation of these zones with agricultural borehole fields or thickness soft soils. This geostatistical merging provides another procedure for mapping subsidence based on GPS and InSAR data, as well as shows good application prospects in subsidence monitoring and disaster warning.
Acknowledgments This work is supported by the EC-GMES-FP7 initiative under the DORIS project and by the Spanish Ministry of Economy and Competitiveness under project ESP2013-47780-C2-2-R. ENVISAT data was provided through European Space Agency’s (ESA) Category-1 Research Grant C1P9044.
References Boni, R., Herrera, G., Meisina, C., Notti, D., Béjar-Pizarro, M., Zucca, F., González, P.J., Palano, M., Tomás, R., Fernández, J., Fernández-Merodo, J.A., Mulas, J., Aragón, R., Guardiola-Albert, C. and Mora, O., 2015. Twenty-year advanced DInSAR analysis of severe land subsidence: The Alto Guadalentín Basin (Spain) case study. Engineering Geology, 198, 40-52. Geovariances, 2010. Isatis technical references, version 10.0. Goovaerts, P., 1997. Geostatistics for natural resources evaluation. Oxford University Press on Demand. 489p. Isaaks, E.H. and Srivastava, R.M., 1989. An Introduction to Applied Geostatistics. Oxford University Press. New York. 561p. Tomás, R., Romero, R., Mulas, J., Marturià, J.J., Mallorquí, J.J., Lopez-Sanchez, J.M., Herrera, G., Gutiérrez, F., González, P.J., Fernández, J., Duque, S., Concha-Dimas, A., Cocksley, G., Castañeda, C., Carrasco, D. and Blanco, P., 2014. Radar interferometry techniques for the study of ground subsidence phenomena: a review of practical issues through cases in Spain. Environmental Earth Sciences 71, 163-181.