Search

The fidelity of climate projections is often undermined by biases in climate models due to their simplification or misrepresentation of unresolved climate processes. While various bias correction methods have been developed to post‐process model outputs to match observations, existing approaches usually focus on limited, low‐order statistics, or break either the spatiotemporal consistency of the target variable, or its dependency upon model resolved dynamics. We develop a Regularized Adversarial Domain Adaptation (RADA) methodology to overcome these deficiencies, and enhance efficient identification and correction of climate model biases. Instead of pre‐assuming the spatiotemporal characteristics of model biases, we apply discriminative neural networks to distinguish historical climate simulation samples and observation samples. The evidences based on which the discriminative neural networks make distinctions are applied to train the domain adaptation neural networks to bias correct climate simulations. We regularize the domain adaptation neural networks using cycle‐consistent statistical and dynamical constraints. An application to daily precipitation projection over the contiguous United States shows that our methodology can correct all the considered moments of daily precipitation at approximately 1° resolution, ensures spatiotemporal consistency and inter‐field correlations, and can discriminate between different dynamical conditions. Our methodology offers a powerful tool for disentangling model parameterization biases from their interactions with the chaotic evolution of climate dynamics, opening a novel avenue toward big‐data enhanced climate predictions.

We adopt a technique, known in the machine learning community as transfer learning, to reduce the bias of computer simulation using very sparse experimental data. Unlike the Bayesian calibration, which is commonly used to estimate the simulation bias, the transfer learning approach discussed in this article involves calculating an artificial neural network surrogate model of the simulations. Assuming that the simulation code correctly predicts the trends in the experimental data but it is subject to unknown biases, we then partially retrain, or transfer learn, the initial surrogate model to match the experimental data. This process eliminates the bias while still taking advantage of the physics relations learned from the simulation. Transfer learning can be easily adapted to a wide range of problems in science and engineering. In this article, we carry out numerical tests to investigate the applicability of this technique to predict the observable outcomes of inertial confinement fusion (ICF) experiments under new conditions. Using our synthetic validation data set, we demonstrate that an accurate predictive model can be built by retraining an initial surrogate model with experimental data volumes so small that they are relevant to the ICF problem. This opens up new opportunities for knowledge transfer and building predictive models in physics. After implementing transfer learning in a standard neural network, we successfully extended the method to a more complex, generative adversarial network architecture, which will be needed for predicting not only scalars but also diagnostic images in our future work.

The path toward realizing the potential of seasonal forecasting and its socioeconomic benefits depends heavily on improving general circulation model based dynamical forecasting systems. To improve dynamical seasonal forecast, it is crucial to set up forecast benchmarks, and clarify forecast limitations posed by model initialization errors, formulation deficiencies, and internal climate variability. With huge cost in generating large forecast ensembles, and limited observations for forecast verification, the seasonal forecast benchmarking and diagnosing task proves challenging. In this study, we develop a probabilistic deep neural network model, drawing on a wealth of existing climate simulations to enhance seasonal forecast capability and forecast diagnosis. By leveraging complex physical relationships encoded in climate simulations, our probabilistic forecast model demonstrates favorable deterministic and probabilistic skill compared to state-of-the-art dynamical forecast systems in quasi-global seasonal forecast of precipitation and near-surface temperature. We apply this probabilistic forecast methodology to quantify the impacts of initialization errors and model formulation deficiencies in a dynamical seasonal forecasting system. We introduce the saliency analysis approach to efficiently identify the key predictors that influence seasonal variability. Furthermore, by explicitly modeling uncertainty using variational Bayes, we give a more definitive answer to how the El Nino/Southern Oscillation, the dominant mode of seasonal variability, modulates global seasonal predictability.

Predictive models that accurately emulate complex scientific processes can achieve speed-ups over numerical simulators or experiments and at the same time provide surrogates for improving the subsequent analysis. Consequently, there is a recent surge in utilizing modern machine learning methods to build data-driven emulators. In this work, we study an often overlooked, yet important, problem of choosing loss functions while designing such emulators. Popular choices such as the mean squared error or the mean absolute error are based on a symmetric noise assumption and can be unsuitable for heterogeneous data or asymmetric noise distributions. We propose Learn-by-Calibrating, a novel deep learning approach based on interval calibration for designing emulators that can effectively recover the inherent noise structure without any explicit priors. Using a large suite of use-cases, we demonstrate the efficacy of our approach in providing high-quality emulators, when compared to widely-adopted loss function choices, even in small-data regimes., The success of machine learning for scientific discovery normally depends on how well the inherent assumptions match the problem in hand. Here, Thiagarajan et al. alleviate this constraint by allowing the change of optimization criterion in a data-driven approach to emulate complex scientific processes.

The 2011–16 California drought illustrates that drought-prone areas do not always experience relief once a favorable phase of El Niño–Southern Oscillation (ENSO) returns. In the twenty-first century, such an expectation is unrealistic in regions where global warming induces an increase in terrestrial aridity larger than the changes in aridity driven by ENSO variability. This premise is also flawed in areas where precipitation supply cannot offset the global warming–induced increase in evaporative demand. Here, atmosphere-only experiments are analyzed to identify land regions where aridity is currently sensitive to ENSO and where projected future changes in mean aridity exceed the range caused by ENSO variability. Insights into the drivers of these changes in aridity are obtained using simulations with the incremental addition of three different factors to the current climate: ocean warming, vegetation response to elevated CO2levels, and intensified CO2radiative forcing. The effect of ocean warming overwhelms the range of ENSO-driven temperature variability worldwide, increasing potential evapotranspiration (PET) in most ENSO-sensitive regions. Additionally, about 39% of the regions currently sensitive to ENSO will likely receive less precipitation in the future, independent of the ENSO phase. Consequently aridity increases in 67%–72% of the ENSO-sensitive area. When both radiative and physiological effects are considered, the area affected by arid conditions rises to 75%–79% when using PET-derived measures of aridity, but declines to 41% when an aridity indicator for total soil moisture is employed. This reduction mainly occurs because plant stomatal resistance increases under enhanced CO2concentrations, resulting in improved plant water-use efficiency, and hence reduced evapotranspiration and soil desiccation. Imposing CO2-invariant stomatal resistance may overestimate future drying in PET-derived indices.

In this short paper, we present constraints on the Galileon inflationary model from the CMB bispectrum. We employ a principal-component analysis of the independent degrees of freedom constrained by data and apply this to the WMAP 9-year data to constrain the free parameters of the model. A simple Bayesian comparison establishes that support for the Galileon model from bispectrum data is at best weak., 11 pages. Matches published version

We use WMAP 9-year bispectrum data to constrain the free parameters of an 'effective field theory' describing fluctuations in single-field inflation. The Lagrangian of the theory contains a finite number of operators associated with unknown mass scales. Each operator produces a fixed bispectrum shape, which we decompose into partial waves in order to construct a likelihood function. Based on this likelihood we are able to constrain four linearly independent combinations of the mass scales. As an example of our framework we specialize our results to the case of 'Dirac-Born-Infeld' and 'ghost' inflation and obtain the posterior probability for each model, which in Bayesian schemes is a useful tool for model comparison. Our results suggest that DBI-like models with two or more free parameters are disfavoured by the data by comparison with single parameter models in the same class., 25 pages, plus references and appendix

We use transport techniques to calculate the trispectrum produced in multiple-field inflationary models with canonical kinetic terms. Our method allows the time evolution of the local trispectrum parameters, tauNL and gNL, to be tracked throughout the inflationary phase. We illustrate our approach using examples. We give a simplified method to calculate the superhorizon part of the relation between field fluctuations on spatially flat hypersurfaces and the curvature perturbation on uniform density slices, and obtain its third-order part for the first time. We clarify how the 'backwards' formalism of Yokoyama et al. relates to our analysis and other recent work. We supply explicit formulae which enable each inflationary observable to be computed in any canonical model of interest, using a suitable first-order ODE solver., Comment: 24 pages, plus references and appendix. v2: matches version published in JCAP; typo fixed in Eq. (54)