Hydrogeophysical Characterization of Nonstationary DNAPL Source Zones by Integrating a Convolutional Variational Autoencoder and Ensemble Smoother

Water Resources Research - Tập 57 Số 2 - 2021
Xueyuan Kang1,2, A. Kokkinaki3, Peter K. Kitanidis1, Xiaoqing Shi2, Jonghyun Lee4, Shaoxing Mo2, Jichun Wu2
1Department of Civil and Environmental Engineering, Stanford University, Stanford, CA, USA
2Key Laboratory of Surficial Geochemistry of Ministry of Education, School of Earth Sciences and Engineering, Nanjing University, Nanjing, China
3Department of Environmental Science, University of San Francisco, San Francisco, CA, USA
4Department of Civil and Environmental Engineering and Water Resources Research Center, University of Hawaii at Manoa, Honolulu, HI, USA

Tóm tắt

Abstract

Detailed characterization of dense nonaqueous phase liquid (DNAPL) source zone architecture (SZA) is essential for designing efficient remediation strategies. However, it is difficult to characterize a highly irregular and localized SZA, because traditional drilling investigations provide limited information. With limited data, the estimation accuracy of traditional geostatistical methods is strongly affected by the parameterization of the prior description of the SZA. To improve characterization performance, we parameterized the DNAPL saturation field using a physics‐based approach. We trained a convolutional variational autoencoder (CVAE) using data from multiphase modeling that captures the physics of DNAPL infiltration. The trained CVAE network was used in SZA inversion to obtain an improved prior DNAPL saturation field, instead of the typical stationary prior covariances. We then integrated the CVAE network into an iterative ensemble smoother (ES), to formulate a joint inversion framework. To overcome difficulties from limited/sparse data, we incorporated hydrogeological and geophysical datasets in the proposed inversion framework. To evaluate the performance of our method, we conducted numerical experiments in a hypothetical heterogeneous aquifer with an intricate SZA. The results show that the CVAE was an effective and efficient parameterization method which can capture the DNAPL infiltration patterns better than a Gaussian prior. The improved prior, combined with multisource datasets, can result in better resolution, and overall improved SZA characterization. In contrast to the standard ES method, the proposed framework reconstructed the SZA more accurately. We also demonstrated that DNAPL depletion behavior and dissolved concentration profiles can be predicted accurately using the estimated SZA.

Từ khóa


Tài liệu tham khảo

10.2307/3430652

Abriola L. M. Christ J. A. Pennell K. D. &Ramsburg C. A.(2012).Source Remediation Challenges. InP. K.Kitanidis &P. L.McCarty(Eds.) Delivery and mixing in the subsurface: Processes and design principles for in situ remediation(pp.239–276).New York NY:Springer.

10.1088/0266-5611/29/11/115014

Archie G. E., 1942, The electrical resistivity log as an aid in determining some reservoir characteristics, Transactions of the American Institute of Mining and Metallurgical Engineers, 146, 54

10.1029/2019WR026481

Bear J., 1972, Dynamics of fluids in porous media

10.1190/1.2432261

Brooks A. N. &Corey A. T.(1964).Hydraulic properties of porous media. InA. T.Corey R. E.Dils &V. M.Yevdjevich(Eds.) Hydrology papers(pp.1–25).Fort Collins CO:Colorado State University.

10.1002/2014WR016017

Canchumun S. W. A., 2020, Recent developments combining ensemble smoother and deep generative networks for facies history matching

10.1016/j.petrol.2019.02.037

10.1016/j.cageo.2019.04.006

10.1016/S0169-7722(03)00142-6

10.1016/j.jconhyd.2010.07.001

Chan S., 2020, Parametrization of stochastic inputs using generative adversarial networks with application in geology, Frontiers in Water, 2, 10.3389/frwa.2020.00005

10.1029/2006WR004886

10.1016/j.jconhyd.2010.02.005

Cohen R. M. Mercer J. W. &Matthews J.(1993).Properties of Fluid and Media. InC. K.Smoley(Eds.) DNAPL site evaluation.Boca Raton FL:CRC Press.

C. O. M. S. O. L Multiphysics., 1994, COMSOL Multiphysics

10.1016/j.jhydrol.2016.11.036

10.1029/2008WR007009

10.1029/93WR01070

10.1029/WR001i004p00563

10.1016/j.cageo.2012.03.011

10.1109/MCS.2009.932223

10.1002/2014WR016384

10.1016/0926-9851(96)00028-6

10.1201/9781482296426

Gerhard J. I., 2003, Relative permeability characteristics necessary for simulating DNAPL infiltration, redistribution, and immobilization in saturated porous media, Water Resources Research, 39, 1213

10.1111/j.1745-6584.2006.00269.x

10.1016/j.jhydrol.2020.125266

Goodfellow I., 2016, Deep learning

Goodfellow I. Pouget‐Abadie J. Mirza M. Xu B. Warde‐Farley D. Ozair S. et al. (2014).Generative adversarial nets. InConference on neural information processing systems (NeuIPS)(pp.2672–2680).

10.1016/j.advwatres.2006.10.005

10.1007/BF01520320

Higgins I., 2017, Beta‐vae: Learning basic visual concepts with a constrained variational framework, 6

Hodges R. A., 2001, Numerical modeling of free‐phase TCE/PCE emplacement and transport in the M‐area basin and evaluation of thermal remediation technologies: SCUREF task order west 041 final report, 126

10.1021/es101654j

10.1029/2000WR900004

10.1016/j.advwatres.2012.08.005

10.1016/j.jappgeo.2015.08.010

10.1029/2020WR027627

10.1016/j.jhydrol.2018.10.019

10.1016/j.jhydrol.2019.124092

Kingma D. P., 2014, Adam: A method for stochastic optimization

Kingma D. P., 2014, Auto‐encoding variational Bayes

10.1017/CBO9780511626166

10.1002/2013WR014630

10.1002/2014WR015478

10.1002/2015WR017894

10.1002/2013WR014663

10.1029/2009WR008947

10.1007/978-1-4614-6922-3

10.1007/s11263-006-7007-9

10.1002/2017WR022148

10.1016/j.advwatres.2017.09.029

LeCun Y., 1989, Generalization and network design strategies

10.1029/WR025i007p01727

10.1029/2006WR005427

10.1002/2016WR020168

Linde N. &Doetsch J.(2016).Joint inversion in hydrogeophysics and near‐surface geophysics. InM.Moorkamp P.Lelièvre N.Linde &A.Khan(Eds.) Integrated imaging of the Earth. (ch.7 pp.119–135). Hoboken New Jersey: John Wiley & Sons.https://doi.org/10.1002/9781118929063.ch7

Liu M. &Grana D.(2018).Ensemble‐based seismic history matching with data reparameterization using convolutional autoencoder. In2018 SEG International exposition and Annual Meeting.(5):3156–3160.Anaheim CA:Society of Exploration Geophysicists.

10.1007/s11004-019-09794-9

10.1111/j.1745-6584.2005.00194.x

Major D., 2012, Development of a protocol and a screening tool for selection of DNAPL source area remediation: Naval facilities engineering command port Hueneme CA

10.1029/2004WR003214

10.1007/s11004-019-09832-6

10.1029/2018WR024638

10.1029/2019WR026082

National Research Council 2005 National research Council Washington DC

National Research Council 2013 National research Council Washington DC

Pankow J. F., 1996, Dense chlorinated solvents and other DNAPLs in groundwater: History, behavior, and remediation

10.1029/WR023i012p02187

10.1029/2009WR008575

10.1016/j.jconhyd.2014.04.004

10.1190/geo2012-0395.1

10.1016/j.jappgeo.2014.10.022

10.1029/98WR02471

Pruess K., 2002, TMVOC, a numerical simulator for three‐phase non‐isothermal flows of multicomponent hydrocarbon mixtures in saturated‐unsaturated heterogeneous media

10.1029/2000WR900187

10.1111/j.1745-6584.2004.tb02667.x

10.1002/2017WR020655

Sneed M., 2001, Hydraulic and mechanical properties affecting groundwater flow and aquifer system compaction, San Joaquin Valley, California, 26

10.1016/j.scitotenv.2016.09.117

10.1002/2017WR020385

10.1029/WR025i009p01959

10.1016/j.jcp.2020.109456

10.1007/s11004-014-9541-2

10.1016/0169-7439(87)80084-9

10.1002/2016WR019111

10.1016/j.advwatres.2017.12.011

10.1029/2006WR004877

10.1016/j.advwatres.2011.09.011

10.1002/2016WR018598

10.1021/es061675q

10.1002/2014WR015740

10.1007/s10040-015-1314-6

10.1016/j.jconhyd.2017.03.005

10.1002/wrcr.20356

10.1126/science.1127647

Ioffe S., 2015, Batch normalization: Accelerating deep network training by reducing internal covariate shift, 448

10.1002/wrcr.20503