Integrating petrophysical uncertainty in salinity estimation from ERT and well data using indicator kriging
Cong-Thi, D.; Van De Vijver, E.; Dieu, L.P.; Huu, H.H.; Nguyen, F.; Hermans, T. (2026). Integrating petrophysical uncertainty in salinity estimation from ERT and well data using indicator kriging. J. Hydrol. (Amst.) 677: 135822. https://dx.doi.org/10.1016/j.jhydrol.2026.135822
In: Journal of Hydrology. Elsevier: Tokyo; Oxford; New York; Lausanne; Shannon; Amsterdam. ISSN 0022-1694; e-ISSN 1879-2707, more
|  |
| Author keywords |
Saltwater intrusion;Coastal aquifers;Indicator kriging;Electrical resistivity tomography;Probabilistic mapping;subsurface heterogeneity;geostatistics;probability;fresh |
| Authors | | Top |
- Cong-Thi, D.
- Van De Vijver, E.
- Dieu, L.P., more
|
- Huu, H.H.
- Nguyen, F., more
- Hermans, T., more
|
|
| Abstract |
Understanding and mapping saltwater intrusion (SI) in coastal aquifers is crucial for safeguarding freshwater resources, yet its spatial characterization remains highly uncertain. Conventional well sampling delivers accurate salinity measurements but is spatially limited, whereas geophysical imaging captures broader subsurface patterns but provides only indirect information constrained by uncertain petrophysical models. To address this, we integrate hydrochemical and geophysical datasets within a probabilistic geostatistical framework. Specifically, total dissolved solids (TDS) concentrations from 70 shallow wells (5-10 m) are combined with electrical resistivity tomography (ERT) data collected along 28 transects in the Luy River catchment, Vietnam. Laboratory experiments established two distinct resistivity-salinity relationships for clay-free and clay-bearing sediments. Uncertainty arises from the unknown clay content at specific locations, representing a key source of ambiguity in petrophysical relationships. Accordingly, each resistivity value was expressed as a probability of exceeding predefined salinity thresholds derived from the TDS distribution. Both a traditional binary indicator transformation and an extended encoding with continuous probabilities were tested. As highlighted by a crossvalidation analysis, the combined dataset produces a more consistent and spatially detailed delineation of SI than TDS-only kriging. Moreover, the probabilistic framework explicitly accounts for alternative petrophysical scenarios and quantifies the likelihood of crossing critical salinity limits. This study illustrates the effectiveness of integrating hydrochemical observations with geophysical measurements using indicator kriging to improve the reliability of SI mapping in heterogeneous coastal aquifers. |
|
