Search NASA⌕ Search

SEARCH · Search NASA

Results for “error models”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 487 records · Page 27

On the Statistical Calibration of Physical Models

ABSTRACT We introduce a novel statistical calibration framework for physical models, relying on probabilistic embedding of model discrepancy error within the model. For clarity of illustration, we take the measurement errors out of consideration, calibrating a chemical model of interest with respect to a more detailed model, considered as “truth” for the present purpose. We employ Bayesian statistical methods for such model‐to‐model calibration and demonstrate their capabilities on simple synthetic models, leading to a well‐defined parameter estimation problem that employs approximate Bayesian computation. The method is then demonstrated on two case studies for calibration of kinetic rate parameters for methane air chemistry, where ignition time information from a detailed elementary‐step kinetic model is used to estimate rate coefficients of a simple chemical mechanism. We show that the calibrated model predictions fit the data and that uncertainty in these predictions is consistent in a mean‐square sense with the discrepancy from the detailed model data.

Sargsyan, K.↗

Improving Lidar Turbulence Estimates for Wind Energy

Remote sensing devices (e.g., lidars) are quickly becoming a cost-effective and reliable alternative to meteorological towers for wind energy applications. Although lidars can measure mean wind speeds accurately, these devices measure different values of turbulence intensity (TI) than an instrument on a tower. In response to these issues, a lidar TI error reduction model was recently developed for commercially available lidars. The TI error model first applies physics-based corrections to the lidar measurements, then uses machine-learning techniques to further reduce errors in lidar TI estimates. The model was tested at two sites in the Southern Plains where vertically profiling lidars were collocated with meteorological towers. This presentation primarily focuses on the physics-based corrections, which include corrections for instrument noise, volume averaging, and variance contamination. As different factors affect TI under different stability conditions, the combination of physical corrections applied in L-TERRA changes depending on the atmospheric stability during each 10-minute time period. This stability-dependent version of L-TERRA performed well at both sites, reducing TI error and bringing lidar TI estimates closer to estimates from instruments on towers. However, there is still scatter evident in the lidar TI estimates, indicating that there are physics that are not being captured in the current version of L-TERRA. Two options are discussed for modeling the remainder of the TI error physics in L-TERRA: machine learning and lidar simulations. Lidar simulations appear to be a better approach, as they can help improve understanding of atmospheric effects on TI error and do not require a large training data set.

17 WIND ENERGY↗

COSP: Satellite simulation software for model assessment

Errors in the simulation of clouds in general circulation models (GCMs) remain a long-standing issue in climate projections, as discussed in the Intergovernmental Panel on Climate Change (IPCC) Fourth Assessment Report. This highlights the need for developing new analysis techniques to improve our knowledge of the physical processes at the root of these errors. The Cloud Feedback Model Intercomparison Project (CFMIP) pursues this objective, and under that framework the CFMIP Observation Simulator Package (COSP) has been developed. COSP is a flexible software tool that enables the simulation of several satellite-borne active and passive sensor observations from model variables. The flexibility of COSP and a common interface for all sensors facilitates its use in any type of numerical model, from high-resolution cloud-resolving models to the coarser-resolution GCMs assessed by the IPCC, and the scales in between used in weather forecast and regional models. The diversity of model parameterization techniques makes the comparison between model and observations difficult, as some parameterized variables (e.g., cloud fraction) do not have the same meaning in all models. The approach followed in COSP permits models to be evaluated against observations and compared against each other in a more consistent manner. This thus permits a more detailed diagnosis of the physical processes that govern the behavior of clouds and precipitation in numerical models. The World Climate Research Programme (WCRP) Working Group on Coupled Modelling has recommended the use of COSP in a subset of climate experiments that will be assessed by the next IPCC report. Here we describe COSP, present some results from its application to numerical models, and discuss future work that will expand its capabilities.

54 ENVIRONMENTAL SCIENCES↗

Detecting and Testing for Structural Error in Computer Models with Application to the Community Atmospheric Model

Bayesian calibration/inversion methods for uncertain input parameters to simulation codes have generated an increasing amount of interest in the climate community in the recent years. This approach uses multiple disparate observational datasets related to the output of the simulator of interest to yield a probabilistic constraint on the uncertain inputs (i.e. probability distribution). A major challenge in applying Bayesian calibration methods to simulation codes is related to the possible presence of a large structural error in the output fields of interest; that is, the presence of discrepancy between the observed and the simulated output that is neither explained by the uncertainty in the inputs nor the observations. We present here methods to test for and visualize structural error in an ensemble of input-perturbed simulations. The methods are based on bootstrap (resmpling) test statistics and well-known metrics to quantify the difference between two probability distributions. The proposed methods are demonstrated using selected precipitation output fields of the Community Atmospheric Model (CAM), which are compared to observations provided by the Global Precipitation Climatology Project (GPCP). The methods are found to be useful in identifying regions and seasons with a large structural error. The proposed structural error detection methodology can be used to screen multiple observational datasets of interest for structural error, making it possible to retain only the datasets with a small structural error for use in Bayesian calibration. In addition, information about the amount of structural error can be valuable to model developers.

58 GEOSCIENCES↗

Balancing aggregation and smoothing errors in inverse models

Inverse models use observations of a system (observation vector) to quantify the variables driving that system (state vector) by statistical optimization. When the observation vector is large, such as with satellite data, selecting a suitable dimension for the state vector is a challenge. A state vector that is too large cannot be effectively constrained by the observations, leading to smoothing error. However, reducing the dimension of the state vector leads to aggregation error as prior relationships between state vector elements are imposed rather than optimized. Here we present a method for quantifying aggregation and smoothing errors as a function of state vector dimension, so that a suitable dimension can be selected by minimizing the combined error. Reducing the state vector within the aggregation error constraints can have the added advantage of enabling analytical solution to the inverse problem with full error characterization. We compare three methods for reducing the dimension of the state vector from its native resolution: (1) merging adjacent elements (grid coarsening), (2) clustering with principal component analysis (PCA), and (3) applying a Gaussian mixture model (GMM) with Gaussian pdfs as state vector elements on which the native-resolution state vector elements are projected using radial basis functions (RBFs). The GMM method leads to somewhat lower aggregation error than the other methods, but more importantly it retains resolution of major local features in the state vector while smoothing weak and broad features.

Turner, A. J.↗

LIKELIHOOD-FREE COSMOLOGICAL INFERENCE WITH TYPE Ia SUPERNOVAE: APPROXIMATE BAYESIAN COMPUTATION FOR A COMPLETE TREATMENT OF UNCERTAINTY

Cosmological inference becomes increasingly difficult when complex data-generating processes cannot be modeled by simple probability distributions. With the ever-increasing size of data sets in cosmology, there is an increasing burden placed on adequate modeling; systematic errors in the model will dominate where previously these were swamped by statistical errors. For example, Gaussian distributions are an insufficient representation for errors in quantities like photometric redshifts. Likewise, it can be difficult to quantify analytically the distribution of errors that are introduced in complex fitting codes. Without a simple form for these distributions, it becomes difficult to accurately construct a likelihood function for the data as a function of parameters of interest. Approximate Bayesian computation (ABC) provides a means of probing the posterior distribution when direct calculation of a sufficiently accurate likelihood is intractable. ABC allows one to bypass direct calculation of the likelihood but instead relies upon the ability to simulate the forward process that generated the data. These simulations can naturally incorporate priors placed on nuisance parameters, and hence these can be marginalized in a natural way. We present and discuss ABC methods in the context of supernova cosmology using data from the SDSS-II Supernova Survey. Assuming a flat cosmology and constant dark energy equation of state, we demonstrate that ABC can recover an accurate posterior distribution. Finally, we show that ABC can still produce an accurate posterior distribution when we contaminate the sample with Type IIP supernovae.

79 ASTRONOMY AND ASTROPHYSICS↗

Machine Learning Models for GPU Error Prediction in a Large Scale HPC System

GPUs are widely deployed on large-scale HPC systems to provide powerful computational capability for scientific applications from various domains. As those applications are normally long-running, investigating the characteristics of GPU errors becomes imperative for reliability. In this paper, we first study the system conditions that trigger GPU errors using six-month trace data collected from a large-scale, operational HPC system. Then, we use machine learning to predict the occurrence of GPU errors, by taking advantage of temporal and spatial dependencies of the trace data. The resulting machine learning prediction framework is robust and accurate under different workloads.

Nie, Bin↗

Lattice Element Error Solver for Modeling of Accelerators, Storage Ring, Transport Systems, and Insertion Devices

The goals of CRADA 350 were achieved in this Phase I SBIR work. Specifically, a software environment was developed which included a nonlinear solver, OASIS. During the Phase I work, we applied the algorithm to a number of nonlinear-solver benchmark problems, including SLAC accelerator problems. By searching over only two optimization-control parameters, this Phase I work has enabled the development of a systematic procedure to facilitate the finding of a “best-guessed” global minimum solution for various types of non-linear problems with different size and complexity. This set of rules was codified and led to a versatile nonlinear program, IMIGO.

43 PARTICLE ACCELERATORS↗

High-Resolution Regional Wave Hindcast for the U.S. Alaska Coast

Regional long-term wave hindcast with state-of-art third-generation models is essential for characterization of regional wave resources. This report summarizes modeling efforts for the simulation of the wave climate along southern Alaska, which used an unstructured, nested-grid modeling approach that incorporates a global-regional nested grid using WAVEWATCH III ® and the high-resolution SWAN (Simulating Waves Nearshore) model. Wave resource and wave bulk parameters were simulated for a 32-year period from 1979 to 2010 and were subsequently validated with wave buoy data in the model domain. Error statistics of model skills were calculated at all buoy stations. Overall, model results, both resource and bulk parameters, match observations well. The model hindcast was able to reproduce the seasonal variation of the sea state with large waves that occur in the winter and early spring months when wind forcing is strong and the calm sea state during the summer when wind forcing is weak. The hindcast also considers the effect of sea ice on wave growth, which affects the southern Bering Sea. The hindcast results, including resource and bulk parameters, were archived at every model grid point for the entire model domain at three-hour intervals. In addition, spectral results were saved at National Data Buoy Center buoy stations and selected virtual buoy locations at hourly intervals. The nested-grid modeling framework employed in this study provides a powerful and efficient modeling approach for accurately simulating wave climate at regional and long-term temporal scales with sufficiently fine resolutions in the nearshore region.

54 ENVIRONMENTAL SCIENCES↗

Robust Bayesian Calibration of a k-ε Model for Compressible Jet-in-Crossflow Simulations

Compressible jet-in-crossflow interactions are difficult to simulate accurately using Reynolds-averaged Navier–Stokes (RANS) models. This could be due to simplifications inherent in RANS or the use of inappropriate RANS constants estimated by fitting to experiments of simple or canonical flows. Our previous work on Bayesian calibration of a k-ε model to experimental data had led to a weak hypothesis that inaccurate simulations could be due to inappropriate constants more than model-form inadequacies of RANS. In this work, Bayesian calibration of k-ε constants to a set of experiments that span a range of Mach numbers and jet strengths has been performed. The variation of the calibrated constants has been checked to assess the degree to which parametric estimates compensate for RANS’s model-form errors. An analytical model of jet-in-crossflow interactions has also been developed, and estimates of k-ε constants that are free of any conflation of parametric and RANS’s model-form uncertainties have been obtained. It has been found that the analytical k-ε constants provide mean-flow predictions that are similar to those provided by the calibrated constants. Further, both of them provide predictions that are far closer to experimental measurements than those computed using “nominal” values of these constants simply obtained from the literature. It can be concluded that the lack of predictive skill of RANS jet-in-crossflow simulations is mostly due to parametric inadequacies, and our analytical estimates may provide a simple way of obtaining predictive compressible jet-in-crossflow simulations.

42 ENGINEERING↗