Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “target variability”

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 37 records · Page 2
All sourcesNASA NTRSDOE OSTI
DOE OSTI2022

Using neural network ensembles to separate ocean biogeochemical and physical drivers of phytoplankton biogeography in Earth system models

Abstract. Earth system models (ESMs) are useful tools for predicting and understanding past and future aspects of the climate system. However, the biological and physical parameters used in ESMs can have wide variations in their estimates. Even small changes in these parameters can yield unexpected results without a clear explanation of how a particular outcome was reached. The standard method for estimating ESM sensitivity is to compare spatiotemporal distributions of variables from different runs of a single ESM. However, a potential pitfall of this method is that ESM output could match observational patterns because of compensating errors. For example, if a model predicts overly weak upwelling and low nutrient concentrations, it might compensate for this by allowing phytoplankton to have a high sensitivity to nutrients. Recently, we demonstrated that neural network ensembles (NNEs) are capable of extracting relationships between predictor and target variables within ocean biogeochemical models. Being able to view the relationships between variables, along with spatiotemporal distributions, allows for a more mechanistically based examination of ESM outputs. Here, we investigated whether we could apply NNEs to help us determine why different ESMs produce different spatiotemporal distributions of phytoplankton biomass. We tested this using three cases. The first and second case used different runs of the same ESM, except that the physical circulations differed between them in the first case, while the biological equations differed between them in the second. Our results indicated that the NNEs were capable of extracting the relationships between variables for different runs of a single ESM, allowing us to distinguish between differences due to changes in circulation (which do not change relationships) from changes in biogeochemical formulation (which do change relationships). In the third case, we applied NNEs to two different ESMs. The results of the third case highlighted the capability of NNEs to contrast the apparent relationships of different ESMs and some of the challenges it presents. Although applied specifically to the ocean components of an ESM, our study demonstrates that Earth system modelers can use NNEs to separate the contributions of different components of ESMs. Specifically, this allows modelers to compare the apparent relationships across different ESMs and observational datasets.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2021

Learning to Correct Climate Projection Biases

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 $1^\circ$ 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.

58 GEOSCIENCES↗
DOE OSTI2025

Refining Jets for CMS Run 3 using Fast Simulation

As the LHC moves into its high-luminosity phase, the CMS experiment must handle more complex data collected at much higher rates. While the Geant4-based simulation application (FullSim) provides highly accurate simulation to complement real data, FullSim’s intensive consumption of computing resources becomes an increasing liability as the rates increase, while faster tools offer an advantage. The fast MC production application (FastSim) delivers a complete simulation with a factor of 10 speedup over FullSim, but introduces inaccuracies in some observables. A specialized refinement method, Fast Perfekt, employs machine learning to improve the accuracy of FastSim. An initial report of this work focused on the refinement of jet flavor tagging observables. This article presents an update on the refinement, focusing on PUPPI jets with Run 3 data-taking conditions. Refinement is extended to include jet transverse momentum as well as its propagation to missing transverse momentum. A gridbased framework and real-time monitoring system have been developed to facilitate optimization and scaling of the refinement to a large number of target variables.

Güngördü, Açelya Deniz [Istanbul Tech. U.]↗
DOE OSTI2023

Exploring the Effects of Population and Employment Characteristics on Truck Flows: An Analysis of NextGen NHTS Origin-Destination Data

Truck transportation remains the dominant mode of US freight transportation because of its advantages, such as the flexibility of accessing pickup and drop-off points and faster delivery. Because of the massive freight volume transported by trucks, understanding the effects of population and employment characteristics on truck flows is critical for better transportation planning and investment decisions. The US Federal Highway Administration published a truck travel origin-destination data set as part of the Next Generation National Household Travel Survey program. This data set contains the total number of truck trips in 2020 within and between 583 predefined zones encompassing metropolitan and nonmetropolitan statistical areas within each state and Washington, DC. In this study, origin-destination-level truck trip flow data was augmented to include zone-level population and employment characteristics from the US Census Bureau. Census population and County Business Patterns data were included. The final data set was used to train a machine learning algorithm-based model, Extreme Gradient Boosting (XGBoost), where the target variable is the number of total truck trips. Shapley Additive ExPlanation (SHAP) was adopted to explain the model results. Results showed that the distance between the zones was the most important variable and had a nonlinear relationship with truck flows.

Uddin, Majbah↗
DOE OSTI2020

Probabilistic neural networks for fluid flow surrogate modeling and data recovery

We consider the use of probabilistic neural networks for fluid flow surrogate modeling and data recovery. This framework is constructed by assuming that the target variables are sampled from a Gaussian distribution conditioned on the inputs. Consequently, the overall formulation sets up a procedure to predict the hyperparameters of this distribution which are then used to compute an objective function given training data. We demonstrate that this framework has the ability to provide for prediction confidence intervals based on the assumption of a probabilistic posterior, given an appropriate model architecture and adequate training data. The applicability of the present framework to cases with noisy measurements and limited observations is also assessed. To demonstrate the capabilities of this framework, we consider canonical regression problems of fluid dynamics from the viewpoint of reduced-order modeling and spatial data recovery for four canonical data sets. The examples considered in this study arise from (i) the shallow-water equations, (ii) a two-dimensional cylinder flow, (iii) the wake of a NACA0012 airfoil with a Gurney flap, and (iv) the NOAA sea surface temperature data set. Furthermore, the present results indicate that the probabilistic neural network not only produces a machine-learning-based fluid flow surrogate model but also systematically quantifies the uncertainty therein to assist with model interpretability.

42 ENGINEERING↗
DOE OSTI2025

Data Agnostic Feature-Target Analysis & Ranking Machine Learning Pipeline (DAFTAR-ML) v0.1.0

DAFTAR-ML is a specialized machine-learning pipeline that identifies relevant features based on their relationship to a target variable. Many ML pipelines focus solely on prediction, and feature ranking is often absent or lacks robust statistical methods. DAFTAR-ML performs its tasks with this outcome in mind. Model training is robust, using nested cross-validation and hyperparameter tuning. Instead of relying on native feature-importance scores, it employs SHAP (SHapley Additive exPlanations) to quantify feature importance. The pipeline also produces comprehensive results, including publication-quality visualizations.

Melie, Tina [Lawrence Berkeley National Laboratory↗
DOE OSTI2017

Maximum credibly yield for deuteriuim-filled double shell imaging targets meeting requirements for yield bin Category A

We are anticipating our first NIF double shell shot using an aluminum ablator and a glass inner shell filled with deuterium shown in figure 1. The expected yield is between a few 10 10 to a few 10 11 dd neutrons. The maximum credible yield is 5e+13. This memo describes why, and what would be expected with variations on the target. This memo evaluates the maximum credible yield for deuterium filled double shell capsule targets with an aluminum ablator shell and a glass inner shell in yield Category A (< 10 14 neutrons). It also pertains to fills of gas diluted with hydrogen, helium ( 3 He or 4 He), or any other fuel except tritium. This memo does not apply to lower z ablator dopants, such as beryllium, as this would increase the ablation efficiency. This evaluation is for 5.75 scale hohlraum targets of either gold or uranium with helium gas fills with density between 0 and 1.6 mg/cc. It could be extended to other hohlraum sizes and shapes with slight modifications. At present only laser pulse energies up to 1.5 MJ were considered with a single step laser pulse of arbitrary shape. Since yield decreases with laser energy for this target, the memo could be extended to higher laser energies if desired. These maximum laser parameters of pulses addressed here are near the edge of NIF’s capability, and constitute the operating envelope for experiments covered by this memo. We have not considered multiple step pulses, would probably create no advantages in performance, and are not planned for double shell capsules. The main target variables are summarized in Table 1 and explained in detail in the memo. Predicted neutron yields are based on 1D and 2D clean simulations.

38 RADIATION CHEMISTRY, RADIOCHEMISTRY, AND NUCLEA↗
DOE OSTI2025

WiDS Livermore Datathon 2025

The WiDS Datathon 2025 requires participants to build a model to predict both an individual’s sex and their ADHD diagnosis using functional brain imaging data of children and adolescents and their socio-demographic, emotions, and parenting information. The task is to create a multi-outcome model to predict two target variables: 1) ADHD (1=yes or 0=no) and 2) female (1=yes or 0=no).

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2022

Analysis of Inlier and Outlier Compounds with Respect to Artificial Neural Network Cetane Number Prediction Accuracy

Artificial neural networks (ANNs) are exceptional at forming non-linear correlations between multivariate input and target variables; however, they are often seen as a “black box” approach, since how ANNs form these correlations is somewhat ambiguous. Furthermore, the process underlying how ANNs learn from inlier and outlier samples within the input dataset is not fully understood. Intuitively, it is expected that training ANNs with inlier samples will increase prediction accuracy and training with outlier samples will reduce prediction accuracy; though, in practice, this is not always true. The present work identifies and analyzes inliers and outliers of existing experimental cetane number (CN) data encompassing a variety of compounds and compound groups. It also investigates how ANNs trained to predict CN perform with and without outliers included in the training data, and whether a relationship exists between inliers/outliers and ANN prediction accuracy across the whole dataset and for individual samples. Additionally, individual outlier compounds are analyzed, highlighting how they structurally differ from inlier compounds.

09 BIOMASS FUELS↗
DOE OSTI2025

Flood Susceptibility Mapping Using Machine Learning and Geospatial-Sentinel-1 SAR Integration for Enhanced Early Warning Systems

This study presents a comprehensive framework for flood susceptibility mapping by integrating geospatial factors with both statistical and machine learning models. Thirteen Flood-related factors, including DEM, slope, TWI, NDVI, etc., are extracted as features of models, and historical flood data derived from Sentinel-1 SAR from 2018 to 2023 are used as the target variables of the models. These datasets are analyzed using a frequency-based statistical model and three machine learning models, including Random Forest, XGBoost, and CNN, to generate flood susceptibility maps. The performance of each model is evaluated through AUC; and SHAP scores are separately generated for Machine learning (ML) models to explain each feature contribution in the ML model. The generated susceptibility maps are validated by high-flood-risk locations monitored by flood sensors, BLE inundation models, and flood-prone areas suggested by the Local Community Task Force. The results indicate that the XGBoost model outperforms all other models, with an AUC of 0.92 and demonstrates the highest alignment with recommended high-flood-risk locations, while the frequency-based statistical model showed the weakest performance with an AUC of 0.65. SHAP value graphs highlight the elevation, slope, and TWI as the most influential features across all models. The susceptibility maps generated by the machine learning model show strong agreement with the BLE map and high-flood-risk areas identified by the local Community Task Force.

Google Engine↗
DOE OSTI2016

Transient dynamics of terrestrial carbon storage: Mathematical foundation and numeric examples

Terrestrial ecosystems absorb roughly 30% of anthropogenic CO 2 emissions since preindustrial era, but it is unclear whether this carbon (C) sink will endure into the future. Despite extensive modeling, experimental, and observational studies, what fundamentally determines transient dynamics of terrestrial C storage under climate change is still not very clear. Here we develop a new framework for understanding transient dynamics of terrestrial C storage through mathematical analysis and numerical experiments. Our analysis indicates that the ultimate force driving ecosystem C storage change is the C storage capacity, which is jointly determined by ecosystem C input (e.g., net primary production, NPP) and residence time. Since both C input and residence time vary with time, the C storage capacity is time-dependent and acts as a moving attractor that actual C storage chases. The rate of change in C storage is proportional to the C storage potential, the difference between the current storage and the storage capacity. The C storage capacity represents instantaneous responses of the land C cycle to external forcing, whereas the C storage potential represents the internal capability of the land C cycle to influence the C change trajectory in the next time step. The influence happens through redistribution of net C pool changes in a network of pools with different residence times. Furthermore, this and our other studies have demonstrated that one matrix equation can exactly replicate simulations of most land C cycle models (i.e., physical emulators). As a result, simulation outputs of those models can be placed into a three-dimensional (3D) parameter space to measure their differences. The latter can be decomposed into traceable components to track the origins of model uncertainty. Moreover, the emulators make data assimilation computationally feasible so that both C flux- and pool-related datasets can be used to better constrain model predictions of land C sequestration. We also propose that the C storage potential be the targeted variable for research, market trading, and government negotiation for C credits.

54 ENVIRONMENTAL SCIENCES↗
NASA NTRS2002

Science Goal Driven Observing: A Step Towards Maximizing Science Returns and Spacecraft Autonomy

While recent missions have made great strides in the ability to autonomously monitor and react to changing health and physical status of spacecraft, little progress has been made in responding quickly to science driven events. Also, the new generation of space-based telescopes will generate data at an unprecedented rate. Yet, while on-board data processing and storage capability will increase rapidly, bandwidth for downloading data will not increase as fast and can become a significant cost of a science program. For observations of inherently variable targets and targets of opportunity, the ability to recognize early if an observation will not meet the science goals, and react accordingly, can have a major positive impact on the overall scientific returns of an observatory and on its operational costs. If the observatory can reprioritize the schedule to focus on alternate targets, discard uninteresting observations prior to downloading, or download them at a reduced resolution its overall efficiency will be dramatically increased. We are investigating and developing tools for a science goal monitoring (SGM) system. The SGM will have an interface to help capture higher-level science goals from the scientists and translate them into a flexible observing strategy that SGM can execute and monitor. SGM will then monitor the incoming data stream and interface with data processing systems to recognize significant events. When an event occurs, the system will use the science goals given it to reprioritize observations, and react appropriately and/or communicate with ground systems - both human and machine - for confirmation and/or further high priority analyses.

Jones, Jeremy↗
NASA NTRS2002

Science Goal Driven Observing and Spacecraft Autonomy

Spacecraft autonomy will be an integral part of mission operations in the coming decade. While recent missions have made great strides in the ability to autonomously monitor and react to changing health and physical status of spacecraft, little progress has been made in responding quickly to science driven events. For observations of inherently variable targets and targets of opportunity, the ability to recognize early if an observation will meet the science goals of a program, and react accordingly, can have a major positive impact on the overall scientific returns of an observatory and on its operational costs. If the onboard software can reprioritize the schedule to focus on alternate targets, discard uninteresting observations prior to downloading, or download a subset of observations at a reduced resolution, the spacecraft's overall efficiency will be dramatically increased. The science goal monitoring (SGM) system is a proof- of-concept effort to address the above challenge. The SGM will have an interface to help capture higher-level science goals from the scientists and translate them into a flexible observing strategy that SGM can execute and monitor. We are developing an interactive distributed system that will use on-board processing and storage combined with event-driven interfaces with ground-based processing and operations, to enable fast re-prioritization of observing schedules, and to minimize time spent on non-optimized observations. This paper will focus on our strategy for developing SGM and the technical challenges that we have encountered. We will discuss the SGM architecture as it applies to the proposed MIDEX-class mission Kronos. However, the architecture and interfaces will also be designed for easy adaptability to other observing platforms, including ground-based systems and to work with different scheduling and pipeline processing systems.

Jones, Jeremy↗
NASA NTRS2002

Science Goal Driven Observing and Spacecraft Autonomy

Spacecraft autonomy will be an integral part of mission operations in the coming decade. While recent missions have made great strides in the ability to autonomously monitor and react to changing health and physical status of spacecraft, little progress has been made in responding quickly to science driven events. For observations of inherently variable targets and targets of opportunity, the ability to recognize early if an observation will meet the science goals of a program, and react accordingly, can have a major positive impact on the overall scientific returns of an observatory and on its operational costs. If the onboard software can reprioritize the schedule to focus on alternate targets, discard uninteresting observations prior to downloading, or download a subset of observations at a reduced resolution, the spacecraft's overall efficiency will be dramatically increased. The science goal monitoring (SGM) system is a proof-of-concept effort to address the above challenge. The SGM will have an interface to help capture higher level science goals from the scientists and translate them into a flexible observing strategy that SGM can execute and monitor. We are developing an interactive distributed system that will use on-board processing and storage combined with event-driven interfaces with ground-based processing and operations, to enable fast re-prioritization of observing schedules, and to minimize time spent on non-optimized observations.

Koratkar, Amuradha↗
NASA NTRS2004

Hydrocode Simulations of the Chesapeake Bay Impact

The Chesapeake Bay Impact Crater (CBIC) formed about 35 million years ago (late Eocene), in a shallow marine environment (400-600 m water depth). The crater is complex and developed in a multi-layer, rheologically-variable target that comprised 400-1000 meters of soft, water-saturated sediments overlying crystalline basement. Seismic reflection data illustrates that the Chesapeake Bay crater morphology - often described as an "inverted sombrero" - is similar to other marine-target impact craters. It consists of a approx. 1 - 1.5-km deep, highly disturbed central crater, surrounded by a shallower, less deformed basin. The inner crater has a diameter of approx. 40 km; the edge of the outer basin extends to ~85-km diameter. The morphological divide between the inner and outer crater is termed the inner ring or peak ring. Little is known about the nature of the inner ring. Seismic reflection data show that the underlying basement is modestly uplifted; however, it is unclear whether the pristine surface expression of the inner ring was elevated above the floor of the outer crater.

Collins, G. S.↗
NASA NTRS2017

The GEOS-5 Neural Network Retrieval (NNR) for AOD

One of the difficulties in data assimilation is the need for multi-sensor data merging that can account for temporal and spatial biases between satellite sensors. In the Goddard Earth Observing System Model Version 5 (GEOS-5) aerosol data assimilation system, a neural network retrieval (NNR) is used as a mapping between satellite observed top of the atmosphere (TOA) reflectance and AOD, which is the target variable that is assimilated in the model. By training observations of TOA reflectance from multiple sensors to map to a common AOD dataset (in this case AOD observed by the ground based Aerosol Robotic Network, AERONET), we are able to create a global, homogenous, satellite data record of AOD from MODIS observations on board the Terra and Aqua satellites. In this talk, I will present the implementation of and recent updates to the GEOS-5 NNR for MODIS collection 6 data.

Castellanos, Patricia↗
NASA NTRS2019

Definition of a Technology Validation Mission for P-band Reflectometry using Signals of Opportunity

Root-Zone Soil Moisture (RZSM) (moisture profile in the top meter of soil) and Snow Water Equivalent (SWE) (total snow pack water content) are identified as priority target variables in the ESAS 2017 decadal survey [1] with critical roles in hydrology and water management. RZSM estimates are vital for understanding multiple Earth system processes and forecasting (for example, droughts [2]). Simultaneous knowledge of surface and RZSM could enable a breakthrough in estimating key unobserved hydrologic fluxes and reduce uncertainty in net ecosystem exchange (NEE), carbon balance [3] discharge estimates, and crop yield forecasts [4] .With the high albedo and insulating properties of snow, monitoring, SWE accumulation would provide a key constraint on the potential runoff during spring ablation while monitoring SWE disappearance rates would provide a key constraint on SWE partition into runoff vs. infiltration/recharge. [5] demonstrated that knowledge of early-spring SWE generally contributes most to streamflow forecast skill in the Western U.S. SWE is also a source of water storage that provides the water resources during spring snowmelt. Despite such potentially transformative contributions, accurate RZSM and SWE measurements are unattainable with current technology. While active/passive L-band methods (e.g. SMAP, SMOS) can reliably retrieve surface soil moisture in the top 5 cm of soil [6], [7]. RZSM estimates are only available through model assimilation of brightness temperatures with a radiative transfer and land surface models [8]. SWE estimation uses multi-frequency passive microwave techniques (e.g. [9]-[11]), which have significant problems with deeper snow and in forested and mountainous environments [12]. Signals of opportunity (SoOp) in P-band (200-400 MHz) is a new remote sensing technique with the capability of estimating both essential hydrologic variables, RZSM and SWE, circumventing many of the aforementioned limitations under all weather conditions day and night. SoOp is the re-utilization of existing powerful satellite transmissions within bands allocated for communications or navigation. P-band SoOp sensitivity to soil moisture has been demonstrated in an airborne experiment over Oklahoma in 2016 [13]. Recent theory [14] and experiments [15] have also confirmed that the reflection coefficient phase is proportional to SWE.

Garrison, J. L.↗
NASA NTRS2019

Accelerated Simulation of Air Pollution Using NVIDIA RAPIDS

Atmospheric chemistry models are a central tool to study and forecast the impact of air pollution on the environment, vegetation, and human health. However, the numerical simulation of chemical kinetics is computationally expensive due to the stiffness of the system of ordinary differential equations that describes atmospheric chemistry. Here we present an alternative approach to the computation of atmospheric chemistry based on machine learning. Our training data set is produced using the NASA Goddard Earth Observing System (GEOS) model with GEOS-Chem chemistry, run on the NASA Center for Climate Simulation (NCCS) Discover supercomputing cluster on 384 Intel Xeon Haswell cores. This model spends more than 50% of total run time on solving atmospheric chemistry. The data set contains as input features the air pollution concentrations before solving the differential equations, together with some key physical parameters such as temperature and sun intensity. As target variables we define the air pollution concentrations after solving the differential equations. Using Dask-cuDF and Dask-XGBoost on the NVIDIA RAPIDS platform on 8 Tesla V100 GPUs, we generate from this training set gradient boosted decision tree models that can reproduce the simulation of chemical kinetics. We do this on the NCCS Advanced Data Analytics Platform (ADAPT) science cloud environment. Our application takes full advantage of recent advances in Dask-XGBoost, such as multi-node and multi-GPU scaling for distributed training with large data sets. The increase in training data size enabled by this is critical to capture the full range of chemical environments encountered across the globe and all annual seasons.The boosted tree models offer good predictability and show many of the features of the full chemistry reference simulation. Further improvements can be achieved through mass balance considerations and by accounting for error correlations. We incorporate the boosted tree models into the GEOS reference model using XGBoost's C API. This enables a seamless integration of the GPU trained models into GEOS-Chem, which is written in Fortran and optimized for use in a massively parallel CPU environment. We show the benefits of this approach and discuss the potential speedup of this machine learning accelerated atmospheric chemistry model.

Keller, Christoph A.↗