A fundamental problem in geophysical modeling is related to the identification and approximation of causal structures among physical processes. However, resolving the bidirectional mappings between ...physical parameters and model state variables (i.e., solving the forward and inverse problems) is challenging, especially when parameter dimensionality is high. Deep learning has opened a new door toward knowledge representation and complex pattern identification. In particular, the recently introduced generative adversarial networks (GANs) hold strong promises in learning cross‐domain mappings for image translation. This study presents a state‐parameter identification GAN (SPID‐GAN) for simultaneously learning bidirectional mappings between a high‐dimensional parameter space and the corresponding model state space. SPID‐GAN is demonstrated using a series of representative problems from subsurface flow modeling. Results show that SPID‐GAN achieves satisfactory performance in identifying the bidirectional state‐parameter mappings, providing a new deep‐learning‐based, knowledge representation paradigm for a wide array of complex geophysical problems.
Plain Language Summary
Development of physically based models requires two steps, mathematical ion (forward modeling) and parameter estimation (inverse modeling). A high‐fidelity model requires high‐quality parameter support. The need for identifying forward and reverse mappings (i.e., a function that associates element of one set to another) is thus ubiquitous in geophysical research. A significant challenge in geosciences is that geoparameters are spatially heterogeneous and high dimensional and yet can only be observed at limited locations. The conventional workflow, built on minimizing the model‐observation mismatch at measurement locations, does not offer an efficient way for estimating the spatial structure of high‐dimensional parameter fields. This work presents a deep‐learning‐based framework for identifying the state‐parameter bidirectional mappings using the recently introduced generative adversarial networks (GANs). GANs have been shown to be adept at associating images from one domain to another. Its potential for discovering mappings in physically based models has not been demonstrated so far. This work shows that GAN can achieve high performance in learning bidirectional parameter‐to‐state mappings in physically based models, thus providing a new way of thinking and doing things in geosciences. The implication for additional applications in subsurface modeling is significant.
Key Points
The need for identifying forward and reverse mappings is ubiquitous in all geophysical research fields
This study shows that generative adversarial networks (GANs) can be used to learn the forward and inverse mappings at the same time
GAN provides a new way of thinking in combining physical‐based modeling with data‐driven modeling and has broad applications for geosciences
Big Data and machine learning (ML) technologies have the potential to impact many facets of environment and water management (EWM). Big Data are information assets characterized by high volume, ...velocity, variety, and veracity. Fast advances in high-resolution remote sensing techniques, smart information and communication technologies, and social media have contributed to the proliferation of Big Data in many EWM fields, such as weather forecasting, disaster management, smart water and energy management systems, and remote sensing. Big Data brings about new opportunities for data-driven discovery in EWM, but it also requires new forms of information processing, storage, retrieval, as well as analytics. ML, a subdomain of artificial intelligence (AI), refers broadly to computer algorithms that can automatically learn from data. ML may help unlock the power of Big Data if properly integrated with data analytics. Recent breakthroughs in AI and computing infrastructure have led to the fast development of powerful deep learning (DL) algorithms that can extract hierarchical features from data, with better predictive performance and less human intervention. Collectively Big Data and ML techniques have shown great potential for data-driven decision making, scientific discovery, and process optimization. These technological advances may greatly benefit EWM, especially because (1) many EWM applications (e.g. early flood warning) require the capability to extract useful information from a large amount of data in autonomous manner and in real time, (2) EWM researches have become highly multidisciplinary, and handling the ever increasing data volume/types using the traditional workflow is simply not an option, and last but not least, (3) the current theoretical knowledge about many EWM processes is still incomplete, but which may now be complemented through data-driven discovery. A large number of applications on Big Data and ML have already appeared in the EWM literature in recent years. The purposes of this survey are to (1) examine the potential and benefits of data-driven research in EWM, (2) give a synopsis of key concepts and approaches in Big Data and ML, (3) provide a systematic review of current applications, and finally (4) discuss major issues and challenges, and recommend future research directions. EWM includes a broad range of research topics. Instead of attempting to survey each individual area, this review focuses on areas of nexus in EWM, with an emphasis on elucidating the potential benefits of increased data availability and predictive analytics to improving the EWM research.
Numerical simulation of flow and transport in heterogeneous formations has long been studied, especially for uncertainty quantification and risk assessment. The high computational cost associated ...with running large‐scale numerical simulations in a Monte Carlo sense has motivated the development of surrogate models, which aim to capture the important input‐output relations of physics‐based models but require only a fraction of the cost of full model runs. In this work, we formulate a conditional deep convolutional generative adversarial network (cDC‐GAN) surrogate model to learn the dynamic functional mappings in multiphase models. The cDC‐GAN belongs to a class of semisupervised learning methods that can be used to learn the data generation processes. Like the original GAN, a main strength of the cDC‐GAN is that it includes a self‐training scheme for improving the quality of generative modeling in a game theoretic framework, without requiring extensive statistical knowledge and assumptions on input data distributions. In particular, our cDC‐GAN model is designed to learn cross‐domain mappings between high‐dimensional input (e.g., permeability) and output (e.g., phase saturations) pairs, with the ability to incorporate conditioning information (e.g., prediction time). As a use case, we demonstrate the performance of cDC‐GAN for predicting the migration of carbon dioxide (CO2) plume in heterogeneous carbon storage reservoirs, which has both numerical and practical significance because of the safe storage requirements now mandated in many countries. Results show that cDC‐GAN achieves high accuracy in predicting the spatial and temporal evolution patterns of the injected CO2 plume, as compared to the original results obtained using a compositional reservoir simulator. The performance of cDC‐GAN models, trained using the same number of training samples, stays relatively robust when the level of spatial heterogeneity is increased. Our cDC‐GAN is pattern based and is not limited by the underlying physics. Thus, it provides a general framework for developing surrogate models, and for conducting uncertainty analyses for a wide range of physics‐based models used in both groundwater and subsurface energy exploration applications.
Key Points
A machine learning‐based surrogate model is proposed for predicting CO2 plume migration
The method is based on conditional deep convolutional generative adversarial network (cDC‐GAN)
cDC‐GAN can facilitate the high‐dimensional cross‐domain learning and predict the CO2 saturation with high accuracy at any time instance
The purpose of this work is to investigate the feasibility of downscaling Gravity Recovery and Climate Experiment (GRACE) satellite data for predicting groundwater level changes and, thus, enhancing ...current capability for sustainable water resources management. In many parts of the world, water management decisions are traditionally informed by in situ observation networks which, unfortunately, have seen a decline in coverage in recent years. Since its launch, GRACE has provided terrestrial water storage change (ΔTWS) data at global and regional scales. The application of GRACE data for local‐scale groundwater resources management has been limited because of uncertainties inherent in GRACE data and difficulties in disaggregating various TWS components. In this work, artificial neural network (ANN) models are developed to predict groundwater level changes directly by using a gridded GRACE product and other publicly available hydrometeorological data sets. As a feasibility study, ensemble ANN models are used to predict monthly and seasonal water level changes for several wells located in different regions across the US. Results indicate that GRACE data play a modest but significantly role in the performance of ANN ensembles, especially when the cyclic pattern of groundwater hydrograph is disrupted by extreme climate events, such as the recent Midwest droughts. The statistical downscaling approach taken here may be readily integrated into local water resources planning activities.
Key Points
GRACE data are downscaled to predict in situ water level changes
A neural network statistical downscaling approach was taken
Modest gain in prediction accuracy was observed
MicroRNA (miRNA)-dependent regulation of gene expression confers robustness to cellular phenotypes and controls responses to extracellular stimuli. Although a single miRNA can regulate expression of ...hundreds of target genes, it is unclear whether any of its distinct biological functions can be due to the regulation of a single target. To explore in vivo the function of a single miRNA-mRNA interaction, we mutated the 3′ UTR of a major miR-155 target (SOCS1) to specifically disrupt its regulation by miR-155. We found that under physiologic conditions and during autoimmune inflammation or viral infection, some immunological functions of miR-155 were fully or largely attributable to the regulation of SOCS1, whereas others could be accounted only partially or not at all by this interaction. Our data suggest that the role of a single miRNA-mRNA interaction is dependent on cell type and biological context.
•miR-155-mediated SOCS1 repression contributes to Treg cell competitive fitness•Th17 cell generation is independent of SOCS1 regulation by miR-155•CD8+ T cells during chronic but not acute infection need SOCS1 repression by miR-155•NK cell response to MCMV infection requires miR-155-mediated SOCS1 repression
A single microRNA (miRNA) can regulate expression of hundreds of target genes, but the biological consequences of individual miRNA-mRNA interactions remain unclear. Rudensky and colleagues show that miR-155-dependent regulation of SOCS1 in different immune cell subsets has cell type- and biological context-dependent in vivo relevance.
It has been suggested that radio telescopes may be sensitive to axion dark matter that resonantly converts to radio photons in the magnetospheres surrounding neutron stars (NSs). In this work, we ...closely examine this possibility by calculating the radiated power from and projected sensitivity to axion dark matter conversion in ensembles of NSs within astrophysical systems like galaxies and globular clusters. We use population synthesis and evolution models to describe the spatial distributions of NSs within these systems and the distributions of NS properties. Focusing on three specific targets for illustration, the Galactic Center of the Milky Way, the globular cluster M54 in the Sagittarius dwarf galaxy, and the Andromeda galaxy, we show that narrow band radio observations with telescopes such as the Green Bank Telescope and the future Square Kilometer Array may be able to probe the quantum chromodynamics axion over roughly 2 orders of magnitude in mass, starting at a fraction of a micro-electron-volt.
Assessing reliability of global models is critical because of increasing reliance on these models to address past and projected future climate and human stresses on global water resources. Here, we ...evaluate model reliability based on a comprehensive comparison of decadal trends (2002–2014) in land water storage from seven global models (WGHM, PCR-GLOBWB, GLDAS NOAH,MOSAIC, VIC, CLM, and CLSM) to trends from three Gravity Recovery and Climate Experiment (GRACE) satellite solutions in 186 river basins (∼60% of global land area). Medians of modeled basin water storage trends greatly underestimate GRACE-derived large decreasing (≤−0.5 km³/y) and increasing (≥0.5 km³/y) trends. Decreasing trends from GRACE are mostly related to human use (irrigation) and climate variations, whereas increasing trends reflect climate variations. For example, in the Amazon, GRACE estimates a large increasing trend of ∼43 km³/y, whereas most models estimate decreasing trends (−71 to 11 km³/y). Land water storage trends, summed over all basins, are positive for GRACE (∼71–82 km³/y) but negative for models (−450 to −12 km³/y), contributing opposing trends to global mean sea level change. Impacts of climate forcing on decadal land water storage trends exceed those of modeled human intervention by about a factor of 2. The model-GRACE comparison highlights potential areas of future model development, particularly simulated water storage. The inability of models to capture large decadal water storage trends based on GRACE indicates that model projections of climate and humaninduced water storage changes may be underestimated.
The Gravity Recovery and Climate Experiment (GRACE) satellite mission and its follow‐on, GRACE‐FO, have provided unprecedented opportunities to quantify the impact of climate extremes and human ...activities on total water storage at large scales. The ∼1‐year data gap between the two GRACE missions needs to be filled to maintain data continuity and maximize mission benefits. In this study, we applied an automated machine learning (AutoML) workflow to perform gridwise GRACE‐like data reconstruction. AutoML represents a new paradigm for optimal algorithm selection, model structure selection, and hyperparameter tuning, addressing some of the most challenging issues in machine learning applications. We demonstrated the workflow over the conterminous U.S. (CONUS) using six types of machine learning models and multiple groups of meteorological and climatic variables as predictors. Results indicate that the AutoML‐assisted gap filling achieved satisfactory performance over the CONUS. On the testing data, the mean gridwise Nash‐Sutcliffe efficiency is around 0.85, the mean correlation coefficient is around 0.95, and the mean normalized root‐mean‐square‐error is about 0.09. Trained models maintain good performance when extrapolating to the mission gap and to GRACE‐FO periods (after June 2017). Results further suggest that no single algorithm provides the best predictive performance over the entire CONUS, stressing the importance of using an end‐to‐end workflow to train, optimize, and combine multiple machine learning models to deliver robust performance, especially when building large‐scale hydrological prediction systems and when predictor importance exhibiting strong spatial variability.
Key Points
A hybrid physics‐based and data‐driven approach was used to fill in the ∼1‐year data gap between GRACE and GRACE‐FO missions
An automated machine learning workflow was introduced to automate the algorithm selection, model structure, and hyperparameter tuning process
The hybrid machine learning approach, which is demonstrated over the CONUS, helped to correct model biases during the gap reconstruction
•Gaussian Process Regression (GPR) is applied to monthly streamflow forecasting.•The efficacy of GPR is demonstrated for over 400 stations in the continental U.S.•Budyko framework is used to examine ...dependence of prediction skill on aridity index.
Streamflow forecasting plays a critical role in nearly all aspects of water resources planning and management. In this work, Gaussian Process Regression (GPR), an effective kernel-based machine learning algorithm, is applied to probabilistic streamflow forecasting. GPR is built on Gaussian process, which is a stochastic process that generalizes multivariate Gaussian distribution to infinite-dimensional space such that distributions over function values can be defined. The GPR algorithm provides a tractable and flexible hierarchical Bayesian framework for inferring the posterior distribution of streamflows. The prediction skill of the algorithm is tested for one-month-ahead prediction using the MOPEX database, which includes long-term hydrometeorological time series collected from 438 basins across the U.S. from 1948 to 2003. Comparisons with linear regression and artificial neural network models indicate that GPR outperforms both regression methods in most cases. The GPR prediction of MOPEX basins is further examined using the Budyko framework, which helps to reveal the close relationships among water-energy partitions, hydrologic similarity, and predictability. Flow regime modification and the resulting loss of predictability have been a major concern in recent years because of climate change and anthropogenic activities. The persistence of streamflow predictability is thus examined by extending the original MOPEX data records to 2012. Results indicate relatively strong persistence of streamflow predictability in the extended period, although the low-predictability basins tend to show more variations. Because many low-predictability basins are located in regions experiencing fast growth of human activities, the significance of sustainable development and water resources management can be even greater for those regions.
Patients with centrally located early-stage non-small-cell lung cancer (NSCLC) are at a higher risk of toxicity from high-dose ablative radiotherapy. NRG Oncology/RTOG 0813 was a phase I/II study ...designed to determine the maximum tolerated dose (MTD), efficacy, and toxicity of stereotactic body radiotherapy (SBRT) for centrally located NSCLC.
Medically inoperable patients with biopsy-proven, positron emission tomography-staged T1 to 2 (≤ 5 cm) N0M0 centrally located NSCLC were accrued into a dose-escalating, five-fraction SBRT schedule that ranged from 10 to 12 Gy/fraction (fx) delivered over 1.5 to 2 weeks. Dose-limiting toxicity (DLT) was defined as any treatment-related grade 3 or worse predefined toxicity that occurred within the first year. MTD was defined as the SBRT dose at which the probability of DLT was closest to 20% without exceeding it.
One hundred twenty patients were accrued between February 2009 and September 2013. Patients were elderly, there were slightly more females, and the majority had a performance status of 0 to 1. Most cancers were T1 (65%) and squamous cell (45%). Organs closest to planning target volume/most at risk were the main bronchus and large vessels. Median follow-up was 37.9 months. Five patients experienced DLTs; MTD was 12.0 Gy/fx, which had a probability of a DLT of 7.2% (95% CI, 2.8% to 14.5%). Two-year rates for the 71 evaluable patients in the 11.5 and 12.0 Gy/fx cohorts were local control, 89.4% (90% CI, 81.6% to 97.4%) and 87.9% (90% CI, 78.8% to 97.0%); overall survival, 67.9% (95% CI, 50.4% to 80.3%) and 72.7% (95% CI, 54.1% to 84.8%); and progression-free survival, 52.2% (95% CI, 35.3% to 66.6%) and 54.5% (95% CI, 36.3% to 69.6%), respectively.
The MTD for this study was 12.0 Gy/fx; it was associated with 7.2% DLTs and high rates of tumor control. Outcomes in this medically inoperable group of mostly elderly patients with comorbidities were comparable with that of patients with peripheral early-stage tumors.
PDF
