Advancing "Autonomous" sensing and prediction of the subsurface environment: a review and exploration of the challenges for soil and groundwater contamination

Abstract

Can we hope for autonomous (self-contained in situ) sensing of subsurface soil and groundwater pollutants to satisfy relevant regulatory criteria? Global advances in sensors, communications, digital technologies, and computational capacity offer this potential. Here we review past efforts to advance subsurface investigation techniques and technologies, and computational efforts to create a digital twin (representation) of subsurface processes. In the context of the potential to link measurement and sensing to a digital twin computation platform, we outline five criteria that might make it possible. Significant advances in sensors based on passive measurement devices are proposed. As an example of what might be achievable, using the five criteria, we describe the deployment of online real-time sensors and simulations for a case study of a petroleum site where natural source zone depletion (NSZD) is underway as a potential biodegradation management option, and where a high-quality conceptual site model is available. Multiple sensors targeting parameters (major gases and temperature influenced by soil moisture) relevant to the subsurface NSZD biodegradation processes are shown to offer the potential to map subsurface processes spatially and temporally and provide continuous estimates of degradation rates for management decisions, constrained by a computational platform of the key processes. Current limitations and gaps in technologies and knowledge are highlighted specific to the case study. More generally, additional key advances required to achieve autonomous sensing of subsurface soil and groundwater pollutants are outlined.

Keywords: Automation; Contamination; Digital twin; Groundwater; Sensor; Soil.

Fig. 1

Schematic showing an ambitious depiction of a linked future system of “autonomous” sensors (green and black diamonds) deployed into the subsurface to monitor pollution distributions and processes with data wirelessly telemetered (black radial arcs) and linked to the internet of things (IOT), an intelligent database and computational platforms to track and interpret subsurface conditions and allow real time management responses

Fig. 2

Depiction of important criteria for advancing autonomous sensing and prediction of subsurface pollution. Note that CSM is the conceptual site model

Fig. 3

Soil gas depth measurements (manual) of oxygen (O₂), carbon dioxide (CO₂), and methane (CH₄) at the aviation gasoline site on 26 October 2020, for the multilevel sampler location VZ03

Fig. 4

September to December 2020 time series of (a) oxygen concentrations from oxygen probes buried at 0.1, 0.5, 1.0, 1.5 m depths below ground, and (b) NSZD rates determined from the oxygen probe data from each of these depths. The vertical dashed line is the date of sampling for the data depicted in Fig. 3

Fig. 5

Depiction of possible gas depth profiles above (and within) a petroleum NAPL at steady-state where no layering or soil moisture variations are assumed, and no background natural soil carbon respiration processes are occurring. Note that the maximum scale for VOCs and methane would be different to oxygen and carbon dioxide. All show linear oxygen concentration decreases with depth and carbon dioxide increases with depth in the soil vadose zone due to NAPL biodegradation. The symbol denotes the approximate depth where heat would be produced due to aerobic biodegradation processes. The panels depict a) where NAPL has no VOCs and methane is not produced; b) where NAPL has VOC vapours moving into the vadose zone but methane is not produced c) where NAPL has no VOCs but methane is being produced and d) where NAPL is at a shallower depth of the vadose zone and VOCs and methane may be present in the vadose zone