Search NASA⌕ Search

SEARCH · Search NASA

Results for “Temporal modeling”

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 757 records · Page 42
All sourcesNASA NTRSDOE OSTI
DOE OSTI2017

STExNMF: Spatio-Temporally Exclusive Topic Discovery for Anomalous Event Detection

Understanding newly emerging events or topics associated with a particular region of a given day can provide deep insight on the critical events occurring in highly evolving metropolitan cities. We propose herein a novel topic modeling approach on text documents with spatio-temporal information (e.g., when and where a document was published) such as location-based social media data to discover prevalent topics or newly emerging events with respect to an area and a time point. We consider a map view composed of regular grids or tiles with each showing topic keywords from documents of the corresponding region. To this end, we present a tilebased spatio-temporally exclusive topic modeling approach called STExNMF, based on a novel nonnegative matrix factorization (NMF) technique. STExNMF mainly works based on the two following stages: (1) first running a standard NMF of each tile to obtain general topics of the tile and (2) running a spatiotemporally exclusive NMF on a weighted residual matrix. These topics likely reveal information on newly emerging events or topics of interest within a region. We demonstrate the advantages of our approach using the geo-tagged Twitter data of New York City. We also provide quantitative comparisons in terms of the topic quality, spatio-temporal exclusiveness, topic variation, and qualitative evaluations of our method using several usage scenarios. In addition, we present a fast topic modeling technique of our model by leveraging parallel computing.

Shin, Sungbok↗
DOE OSTI2026

Lethality is Local, but Survival is Systemic: Temporal and Multi-Organ Responses to Chlorine Gas Exposure in a Murine Model

Chlorine gas (Cl2) is a highly toxic chemical associated with both localized lung injury and systemic health effects. While pulmonary damage has been well characterized, the systemic inflammatory and metabolic responses remain poorly understood. We aimed to define the temporal and multi-organ responses to Cl2 exposure in a murine model, with a focus on identifying spatiotemporal inflammation and its impact on survival and lethality. SKH1 mice were exposed for 10 min to varying concentrations of Cl2 (94.4–810 ppm, representative of non-lethal, LD10, and LD50 doses) and monitored for respiratory function, perfusion, and acidosis using organ-specific imaging. At multiple time points (40 min, 6 h, 24 h, and 7 d), we measured phosphoproteins, cytokines, chemokines, growth factors, and metabolic hormones in the lungs, heart, cortex, and plasma. Statistical modeling and logistic regression were used to identify biomarkers associated with lethality and survival. We found that lung injury was the primary cause of potential lethality, particularly via early phosphoprotein signaling disruptions. However, survival correlated with early systemic coordination of inflammatory and metabolic signals across organs. Perfusion and acidosis imaging were strongly associated with chemokine and hormone responses. Key survival-associated plasma biomarkers included decreased insulin, increased ghrelin, and decreased eotaxin. While potential lethality from Cl2 exposure is locally driven by pulmonary injury, survival depends on systemic, multi-organ responses that occur rapidly post-exposure. Within this model, our findings identify a potential therapeutic window to enhance survival and suggest candidate biomarkers that may be explored translationally for both triage and treatment of chlorine-related incidents.

chlorine gas↗
DOE OSTI2026

A Statistician’s Overview of Physics-Informed Neural Networks for Spatio-Temporal Data

The recent success of deep neural network models with physical constraints (so-called, Physics-Informed Neural Networks, PINNs) has led to renewed interest in the incorporation of mechanistic information in predictive models. Statisticians and others have long been interested in this problem, which has led to several practical and innovative solutions dating back decades. In this overview, we focus on the problem of data-driven prediction and inference of dynamic spatio-temporal processes that include mechanistic information, such as would be available from partial differential equations, with a strong focus on the quantification of uncertainty associated with data, process, and parameters. Here, we give a brief review of several paradigms and focus our attention on Bayesian implementations given they naturally accommodate uncertainty quantification. We then show that it is straight-forward to include the Bayesian PINN (B-PINN) within the Bayesian hierarchical model (BHM) framework that has long been considered for modeling dynamic spatio-temporal processes. Such a BHM-PINN is illustrated via a simulation study in which a latent nonlinear Burgers’ equation PDE governs the dynamics of Poisson distributed spatio-temporal data. Supplementary materials for this article are available online, including a standardized description of the materials available for reproducing the work.

Bayesian↗
NASA NTRS2012

Large Eddy Simulation (LES) of Particle-Laden Temporal Mixing Layers

High-fidelity models of plume-regolith interaction are difficult to develop because of the widely disparate flow conditions that exist in this process. The gas in the core of a rocket plume can often be modeled as a time-dependent, high-temperature, turbulent, reacting continuum flow. However, due to the vacuum conditions on the lunar surface, the mean molecular path in the outer parts of the plume is too long for the continuum assumption to remain valid. Molecular methods are better suited to model this region of the flow. Finally, granular and multiphase flow models must be employed to describe the dust and debris that are displaced from the surface, as well as how a crater is formed in the regolith. At present, standard commercial CFD (computational fluid dynamics) software is not capable of coupling each of these flow regimes to provide an accurate representation of this flow process, necessitating the development of custom software. This software solves the fluid-flow-governing equations in an Eulerian framework, coupled with the particle transport equations that are solved in a Lagrangian framework. It uses a fourth-order explicit Runge-Kutta scheme for temporal integration, an eighth-order central finite differencing scheme for spatial discretization. The non-linear terms in the governing equations are recast in cubic skew symmetric form to reduce aliasing error. The second derivative viscous terms are computed using eighth-order narrow stencils that provide better diffusion for the highest resolved wave numbers. A fourth-order Lagrange interpolation procedure is used to obtain gas-phase variable values at the particle locations.

Bellan, Josette↗
NASA NTRS2005

The Flow of Plasma in the Solar-Terrestrial Environment

The overall goal of our NASA theory research is to trace the flow of mass, momentum, and energy through the magnetosphere-ionosphere-atmosphere system taking into account the coupling, time delays, and feedback mechanisms that are characteristic of the system. Our approach is to model the magnetosphere-ionosphere-atmosphere (M-I-A) system in a self-consistent quantitative manner using unique global models that allow us to study the coupling between the different regions on a range of spatial and temporal scales. The uniqueness of our global models stems from their high spatial and temporal resolutions, the physical processes included, and the numerical techniques employed. Currently, we have time-dependent global models of the ionosphere, thermosphere, polar wind, plasmasphere, and electrodynamics. It is now becoming clear that a significant fraction of the flow of mass, momentum, and energy in the M-I-A system occurs on relatively small spatial scales. Therefore, an important aspect of our NASA Theory program concerns the effect that mesoscale (100-l000 km) density structures have on the macroscopic flows in the ionosphere, thermosphere, and polar wind. The structures can be created either by structured magnetospheric inputs (i.e., structured electric field, precipitation, or Birkeland current patterns) or by time variations of these inputs due to geomagnetic storms and substorms. Some of the mesoscale structures of interest include sun-aligned polar cap arcs, propagating plasma patches, traveling convection vortices, subauroral ion drift (SAID) channels, gravity waves, and the polar hole.

Schunk, Robert W.↗
DOE OSTI2025

Spatial and Temporal Variability of Vertical Velocity under Shallow Cumulus

Vertical velocity distribution below cloud is one of the key determinants of cloud life cycle, but observations of this variable are extremely sparse in space. Doppler lidar retrievals and large-eddy simulations at the U.S. Department of Energy’s Atmospheric Radiation Measurement User Facility Southern Great Plains site are used to determine whether vertical velocity statistics from temporally dense profiles at a single location can be substituted for spatial vertical velocity statistics. We show that even a small number (five) of widely distributed [ O (1°) latitude/longitude spacing] lidars is sufficient sampling to reconstruct domainwide spatial vertical velocity variance, but not higher moments of the vertical velocity distribution. Spatial and temporal vertical velocity variances in the Doppler lidar observations are nearly interchangeable as long as the spatial variance is temporally averaged and the temporal variance is averaged across lidars. This is true even though the dominant spatial scales of vertical velocity variability are ≲ 3 km, more than an order of magnitude smaller than the spacing between the lidars. Further, in the limit where the temporal variance does not vary across a spatial domain (e.g., if the meteorological and surface forcing of the atmospheric turbulence is homogeneous across the domain) and the domain-mean vertical velocity is zero, the commonly available retrieval of temporal vertical velocity variance at one site is equivalent to the spatial variance over the domain. We use an updraft parcel model to show that substituting temporal for spatial vertical velocity statistics will have a relatively minor effect on cloud droplet number concentrations.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2018

Temporal Coupling of Subsurface and Surface Soil CO 2 Fluxes: Insights From a Nonsteady State Model and Cross-Wavelet Coherence Analysis

Inferences about subsurface CO 2 fluxes often rely on surface soil respiration (R soil ) estimates because directly measuring subsurface microbial and root respiration (collectively, CO 2 production, S Total ), is difficult. To evaluate how well R soil serves as a proxy for S Total , we applied the nonsteady state DEconvolution of Temporally varying Ecosystem Carbon componenTs model (0.01-m vertical resolution), using 6-hourly data from a Wyoming grassland, in six simulations that cross three soil types (clay, sandy loam, and sandy) with two depth distributions of subsurface biota. We used cross-wavelet coherence analysis to examine temporal coherence (localized linear correlation) and offsets (lags) between S Total and R soil and fluxes and drivers (e.g., soil temperature and moisture). Cross-wavelet coherence revealed higher coherence between fluxes and drivers than linear regressions between concurrent variables. Soil texture and moisture exerted the strongest controls over coherence between CO 2 fluxes. Coherence between CO 2 fluxes in all soil types was strong at short (~1 day) and long periods (>8 days), but soil type controlled lags, and rainfall events decoupled the fluxes at periods of 1–8 days for several days in sandy soil, up to 1 week in sandy loam, and for a month or more in clay soil. Concentrating root and microbial biomass nearer the surface decreased lags in all soil types and increased coherence up to 10% in clay soil. The assumption of high temporal coherence between R soil and S Total is likely valid in dry, sandy soil, but may lead to underestimates of short-term STotal in semiarid grasslands with fine-grained and/or wet soil.

54 ENVIRONMENTAL SCIENCES↗
NASA NTRS2012

Simulation of Oxygen Disintegration and Mixing With Hydrogen or Helium at Supercritical Pressure

The simulation of high-pressure turbulent flows, where the pressure, p, is larger than the critical value, p(sub c), for the species under consideration, is relevant to a wide array of propulsion systems, e.g. gas turbine, diesel, and liquid rocket engines. Most turbulence models, however, have been developed for atmospheric-p turbulent flows. The difference between atmospheric-p and supercritical-p turbulence is that, in the former situation, the coupling between dynamics and thermodynamics is moderate to negligible, but for the latter it is very significant, and can dominate the flow characteristics. The reason for this stems from the mathematical form of the equation of state (EOS), which is the perfect-gas EOS in the former case, and the real-gas EOS in the latter case. For flows at supercritical pressure, p, the large eddy simulation (LES) equations consist of the differential conservation equations coupled with a real-gas EOS. The equations use transport properties that depend on the thermodynamic variables. Compared to previous LES models, the differential equations contain not only the subgrid scale (SGS) fluxes, but also new SGS terms, each denoted as a correction. These additional terms, typically assumed null for atmospheric pressure flows, stem from filtering the differential governing equations, and represent differences between a filtered term and the same term computed as a function of the filtered flow field. In particular, the energy equation contains a heat-flux correction (q-correction) that is the difference between the filtered divergence of the heat flux and the divergence of the heat flux computed as a function of the filtered flow field. In a previous study, there was only partial success in modeling the q-correction term, but in this innovation, success has been achieved by using a different modeling approach. This analysis, based on a temporal mixing layer Direct Numerical Simulation database, shows that the focus in modeling the q-correction should be on reconstructing the primitive variable gradients rather than their coefficients, and proposes the approximate deconvolution model (ADM) as an effective means of flow field reconstruction for LES heat flux calculation. Further, results for a study conducted for temporal mixing layers initially containing oxygen in the lower stream, and hydrogen or helium in the upper stream, show that, for any LES, including SGS-flux models (constant-coefficient Gradient or Scale-Similarity models, dynamic-coefficient Smagorinsky/Yoshizawa or mixed Smagorinsky/Yoshizawa/Gradient models), the inclusion of the q-correction in the LES leads to the theoretical maximum reduction of the SGS heat-flux difference. The remaining error in modeling this new subgrid term is thus irreducible.

Bellan, Josette↗
NASA NTRS1991

A study of the sources and sinks of methane and methyl chloroform using a global three-dimensional Lagrangian tropospheric tracer transport model

Sources and sinks of methane and methyl chloroform are investigated using a global three-dimensional Lagrangian tropospheric tracer transport model with parameterized hydroxyl and temperature fields. Using the hydroxyl radical field calibrated to the methyl chloroform observations, the globally averaged release of methane and its spatial and temporal distribution were investigated. Two source function models of the spatial and temporal distribution of the flux of methane to the atmosphere were developed. The first model was based on the assumption that methane is emitted as a proportion of net primary productivity (NPP). The second model identified source regions for methane from rice paddies, wetlands, enteric fermentation, termites, and biomass burning based on high-resolution land use data. The most significant difference between the two models were predictions of methane fluxes over China and South East Asia, the location of most of the world's rice paddies, indicating that either the assumption that a uniform fraction of NPP is converted to methane is not valid for rice paddies, or that NPP is underestimated for rice paddies, or that present methane emission estimates from rice paddies are too high.

Taylor, John A.↗
DOE OSTI2017

Availability and Maintenance Modeling for GIS Equipment Served in High-Speed Railway Under Incomplete Maintenance

Gas-insulated substation (GIS) equipment served in the traction power-supply system (TPSS) of the high-speed railway (HSR) suffers from extremely severe traction load conditions during its lifetime. In this paper, three major issues related to the actual field maintenance operations of GIS equipment in TPSS are addressed: 1) s-dependent degradation and shock process; 2) both spatial and temporal failure thresholds of shocks; and 3) incomplete maintenance (i.e., nonignorable maintenance time). To model the deterioration and failure process of GIS equipment, the hard failure model and the soft failure model are proposed. The characteristic function is utilized to bridge the s-dependent degradation and shock process in the soft failure model, while the extreme shock model (with spatial threshold) and δ-shock model (with temporal threshold) are introduced in the hard failure model. Furthermore, the index of availability, which combines both reliability and maintainability, is derived under the incomplete maintenance consideration. Based on the balance of availability and maintenance cost, the maintenance strategy with periodic inspections for GIS equipment in the long-run time span is developed. In conclusion, optimizations of three decision parameters (i.e., inspection period, extreme shock threshold, and δ-shock threshold) are implemented to achieve the best performance (with high availability and low maintenance cost) of the maintenance strategy.

33 ADVANCED PROPULSION SYSTEMS↗
DOE OSTI2015

Combined modeling of acceleration, transport, and hydrodynamic response in solar flares. II. Inclusion of radiative transfer with RADYN

Solar flares involve complex processes that are coupled and span a wide range of temporal, spatial, and energy scales. Modeling such processes self-consistently has been a challenge in the past. Here we present results from simulations that couple particle kinetics with hydrodynamics (HD) of the atmospheric plasma. We combine the Stanford unified Fokker–Planck code that models particle acceleration and transport with the RADYN HD code that models the atmospheric response to collisional heating by accelerated electrons through detailed radiative transfer calculations. We perform simulations using two different electron spectra, one an ad hoc power law and the other predicted by the model of stochastic acceleration by turbulence or plasma waves. Surprisingly, the later model, even with energy flux ≪10{sup 10} erg s{sup −1} cm{sup −2}, can cause “explosive” chromospheric evaporation and drive stronger up- and downflows (and HD shocks). This is partly because our acceleration model, like many others, produces a spectrum consisting of a quasi-thermal component plus a power-law tail. We synthesize emission-line profiles covering different heights in the lower atmosphere, including Hα 6563 Å, He ii 304 Å, Ca ii K 3934 Å, and Si iv 1393 Å. One interesting result is the unusual high temperature (up to a few times 10{sup 5} K) of the formation site of He ii 304 Å, which is expected owing to photoionization-recombination under flare conditions, compared to those in the quiet Sun dominated by collisional excitation. When compared with observations, our results can constrain the properties of nonthermal electrons and thus the poorly understood particle acceleration mechanism.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2015

How bad or good are the external forward shock afterglow models of gamma-ray bursts?

The external forward shock models have been the standard paradigm to interpret the broadband afterglow data of gamma-ray bursts (GRBs). One prediction of the models is that some afterglow temporal breaks at different energy bands should be achromatic; that is, the break times should be the same in different frequencies. Multiwavelength observations in the Swift era have revealed chromatic afterglow behaviors at least in some GRBs, casting doubts on the external forward shock origin of GRB afterglows. In this paper, using a large sample of GRBs with both X-ray and optical afterglow data, we perform a systematic study to address the question: how bad or good are the external forward shock models? Our sample includes 85 GRBs up to 2014 March with well-monitored X-ray and optical light curves. Based on how well the data abide by the external forward shock models, we categorize them into five grades and three samples. The first two grades (Grade I and II) include 45 of 85 GRBs. They show evidence of, or are consistent with having, an achromatic break. The temporal and spectral behaviors in each afterglow segment are consistent with the predictions (the “closure relations”) of the forward shock models. These GRBs are included in the Gold sample. The next two grades (Grade III and IV) include 37 of 85 GRBs. They are also consistent with having an achromatic break, even though one or more afterglow segments do not comply with the closure relations. These GRBs are included in the Silver sample. Finally, Grade V (3/85) shows direct evidence of chromatic behaviors, suggesting that the external shock models are inconsistent with the data. These are included in the Bad sample. We further perform statistical analyses of various observational properties (temporal index α, spectral index β, break time t{sub b}) and model parameters (energy injection index q, electron spectral index p, jet opening angle θ{sub j}, radiative efficiency η{sub γ}, and so on) of the GRBs in the Gold sample, and derive constraints on the magnetization parameter ϵ{sub B} in the forward shock. Overall, we conclude that the simplest external forward shock models can account for the multiwavelength afterglow data of at least half of the GRBs. When more advanced modeling (e.g., long-lasting reverse shock, structured jets, arbitrary circumburst medium density profile) is invoked, up to >90% of the afterglows may be interpreted within the framework of the external shock models.

79 ASTRONOMY AND ASTROPHYSICS↗
NASA NTRS1991

Reasoning about real-time systems with temporal interval logic constraints on multi-state automata

Models of real-time systems using a single paradigm often turn out to be inadequate, whether the paradigm is based on states, rules, event sequences, or logic. A model-based approach to reasoning about real-time systems is presented in which a temporal interval logic called TIL is employed to define constraints on a new type of high level automata. The combination, called hierarchical multi-state (HMS) machines, can be used to model formally a real-time system, a dynamic set of requirements, the environment, heuristic knowledge about planning-related problem solving, and the computational states of the reasoning mechanism. In this framework, mathematical techniques were developed for: (1) proving the correctness of a representation; (2) planning of concurrent tasks to achieve goals; and (3) scheduling of plans to satisfy complex temporal constraints. HMS machines allow reasoning about a real-time system from a model of how truth arises instead of merely depending of what is true in a system.

Gabrielian, Armen↗
DOE OSTI2025

Investigation of Main Bearing Fatigue Estimate Sensitivity to Synthetic Turbulence Models Using a Novel Drivetrain Model Implemented in OpenFAST

ABSTRACT A coupled medium‐fidelity drivetrain model is developed and implemented in OpenFAST for a 10‐MW land‐based reference turbine. The implementation is verified against a fully coupled multibody wind turbine model, including a detailed drivetrain. The new model can simultaneously and accurately estimate main bearing loads and represent elastic bending of the drivetrain. It has low computational cost and is useful for early design phases, sensitivity analyses and complex systems like wind farms (where computational expense must be expended elsewhere). Here, the model is implemented for a monopile offshore wind turbine and used to investigate the sensitivity of main bearing basic rating life to different synthetic turbulence models. Large‐eddy simulations (LES) targeting stable, neutral, and unstable atmospheric conditions at below‐, near‐ and above‐rated wind speeds are used as a reference. The turbulence models recommended by the International Electrotechnical Commission, the Mann spectral tensor model, and the Kaimal spectral model with exponential coherence are fitted to the LES data. Additionally, a constrained turbulence generator, PyConTurb (short for Python Constrained Turbulence ), based on LES data, is applied in the aero‐hydro‐servo‐elastic simulations. Taking PyConTurb as the baseline, the Kaimal model significantly underestimates fatigue of the downwind main bearing, with between 10% and 40% less damage. The Mann model also underestimates the downwind main bearing fatigue by up to 30%. The upwind main bearing damage is driven by mean loads, and differences between models are less significant, although the trends are similar. Reasons for these discrepancies are investigated and attributed to differences in spatial and temporal variations among the turbulence models.

17 WIND ENERGY↗
DOE OSTI2013

Modeling microalgal growth in an airlift-driven raceway reactor

In previous proof-of-concept studies, feasibility of a new airlift-raceway configuration and its energetic advantage and improved CO2 utilization efficiency over the traditional raceways and photobioreactors have been documented. In the current study, a mathematical model for predicting biomass growth in the airlift-raceway reactor is presented, which includes supply and transfer of CO2 and the synergetic effects of light, CO2, nitrogen, and temperature. The model was calibrated and validated with data from prototype scale versions of the reactor on two test species: Nannochloropsis salina and Scenedesmus sp., cultivated under indoor and outdoor conditions. Predictions of biomass concentrations by the proposed model agreed well with the temporal trend of the experimental data, with r(2) ranging from 0.96 to 0.98, p<0.001. A sensitivity analysis of the 10 model parameters used in this study revealed that only three of them were significant, with sensitivity coefficients ranging from 0.08 to 0.13.

09 BIOMASS FUELS↗
DOE OSTI2014

A connectivity-based modeling approach for representing hysteresis in macroscopic two-phase flow properties

During CO 2 injection and storage in deep reservoirs, the injected CO 2 enters into an initially brine saturated porous medium, and after the injection stops, natural groundwater flow eventually displaces the injected mobile-phase CO 2 , leaving behind residual non-wetting fluid. Accurate modeling of two-phase flow processes are needed for predicting fate and transport of injected CO 2 , evaluating environmental risks and designing more effective storage schemes. The entrapped non-wetting fluid saturation is typically a function of the spatially varying maximum saturation at the end of injection. At the pore-scale, distribution of void sizes and connectivity of void space play a major role for the macroscopic hysteresis behavior and capillary entrapment of wetting and non-wetting fluids. This paper presents development of an approach based on the connectivity of void space for modeling hysteretic capillary pressure-saturation-relative permeability relationships. The new approach uses void-size distribution and a measure of void space connectivity to compute the hysteretic constitutive functions and to predict entrapped fluid phase saturations. Two functions, the drainage connectivity function and the wetting connectivity function, are introduced to characterize connectivity of fluids in void space during drainage and wetting processes. These functions can be estimated through pore-scale simulations in computer-generated porous media or from traditional experimental measurements of primary drainage and main wetting curves. The hysteresis model for saturation-capillary pressure is tested successfully by comparing the model-predicted residual saturation and scanning curves with actual data sets obtained from column experiments found in the literature. A numerical two-phase model simulator with the new hysteresis functions is tested against laboratory experiments conducted in a quasi-two-dimensional flow cell (91.4cm×5.6cm×61cm), packed with homogeneous and heterogeneous sands. Initial results show that the model can predict spatial and temporal distribution of injected fluid during the experiments reasonably well. However, further analyses are needed for comprehensively testing the ability of the model to predict transient two-phase flow processes and capillary entrapment in geological reservoirs during geological carbon sequestration.

58 GEOSCIENCES↗
DOE OSTI2021

A spatial evaluation of Arctic sea ice and regional limitations in CMIP6 historical simulations

The Arctic sea ice response to a warming climate is assessed in a subset of models participating in Phase 6 of the Coupled Model Intercomparison Project (CMIP6), using several metrics in comparison with satellite observations and results from the Pan-Arctic Ice Ocean Modeling and Assimilation System and the Regional Arctic System Model. Our study examines the historical representation of sea ice extent, volume, and thickness using spatial analysis metrics, such as the integrated ice-edge error, Brier score, and Spatial Probability Score. We find that the CMIP6 multi-model mean captures the mean annual cycle and 1979-2014 sea ice trends remarkably well. However, individual models experience a wide range of uncertainty in the spatial distribution of sea ice when compared against satellite measurements and reanalysis data. Our metrics expose common and individual regional model biases, which sea ice temporal analyses alone do not capture. We identify large ice edge and ice thickness errors in Arctic sub-regions, implying possible model specific limitations in or lack of representation of some key physical processes. We postulate that many of them could be related to the oceanic forcing, especially in the marginal and shelf seas, where seasonal sea ice changes are not adequately simulated. We therefore conclude that an individual model’s ability to represent the observed/reanalysis spatial distribution still remains a challenge. We propose the spatial analysis metrics as useful tools to diagnose model limitations, narrow down possible processes affecting them, and guide future model improvements critical to the representation and projections of Arctic climate change.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2021

Semi-automated Design of Artificial Intelligence Earth Systems Models

Prediction and observation of water cycles at various scales involve not only patterns isolated in space and time, but also modeling of complex spatio-temporal relationships across multiple domains. For instance, evapotranspiration (ET) and leaf area indexes (LAI) are two parameters that are needed to accurately model and understand land-atmosphere processes. Accurate assessments of ET and LAI are critical for understanding hydrological processes, deforestation, crop yield, and irrigation impacts. However, ET estimates for global simulations are available at very coarse spatial resolution. They are usually derived from satellite data based on broad plant functional types (PFTs), which fail to capture fine-scale variations because of changes in vegetation type across the globe. Similarly LAI estimates have typically been derived from vegetation indices at global scales or estimated locally using physical models, both of which suffer from a range of uncertainties that impact model sensitivity. The new era of AI model development for Earth systems (ES) calls for data-driven methods that provide domain scientists with uncertainty-aware estimations of biophysical parameters such as ET and LAI in a generalizable, interpretable, and discoverable manner.

54 ENVIRONMENTAL SCIENCES↗