Search NASA⌕ Search

SEARCH · Search NASA

Results for “variable importance”

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 505 records · Page 28

The role of impulse parameters in force variability

One of the principle limitations of the human motor system is the ability to produce consistent motor responses. When asked to repeatedly make the same movement, performance outcomes are characterized by a considerable amount of variability. This occurs whether variability is expressed in terms of kinetics or kinematics. Variability in performance is of considerable importance because for tasks requiring accuracy it is a critical variable in determining the skill of the performer. What has long been sought is a description of the parameter or parameters that determine the degree of variability. Two general experimental protocals were used. One protocal is to use dynamic actions and record variability in kinematic parameters such as spatial or temporal error. A second strategy was to use isometric actions and record kinetic variables such as peak force produced. What might be the important force related factors affecting variability is examined and an experimental approach to examine the influence of each of these variables is provided.

Carlton, L. G.↗

Relative contributions of mean-state shifts and ENSO-driven variability to precipitation changes in a warming climate

The El Niño–Southern Oscillation (ENSO) is an important driver of regional hydroclimate variability through far-reaching teleconnections. This study uses simulations performed with coupled general circulation models (CGCMs) to investigate how regional precipitation in the twenty-first century may be affected by changes in both ENSO-driven precipitation variability and slowly evolving mean rainfall. First, a dominant, time-invariant pattern of canonical ENSO variability (cENSO) is identified in observed SST data. Next, the fidelity with which 33 state-of-the-art CGCMs represent the spatial structure and temporal variability of this pattern (as well as its associated precipitation responses) is evaluated in simulations of twentieth-century climate change. Possible changes in both the temporal variability of this pattern and its associated precipitation teleconnections are investigated in twenty-first-century climate projections. Models with better representation of the observed structure of the cENSO pattern produce winter rainfall teleconnection patterns that are in better accord with twentieth-century observations and more stationary during the twenty-first century. Finally, the model-predicted twenty-first-century rainfall response to cENSO is decomposed into the sum of three terms: 1) the twenty-first-century change in the mean state of precipitation, 2) the historical precipitation response to the cENSO pattern, and 3) a future enhancement in the rainfall response to cENSO, which amplifies rainfall extremes. Lastly, by examining the three terms jointly, this conceptual framework allows the identification of regions likely to experience future rainfall anomalies that are without precedent in the current climate.

54 ENVIRONMENTAL SCIENCES↗

The Fall and Rise of the Global Climate Model

Abstract Global models are an essential tool for climate projections, but conventional coarse‐resolution atmospheric general circulation models suffer from errors both in their parameterized cloud physics and in their representation of climatically important circulation features. A notable recent study by Terai et al. (2020, https://doi.org/10.1029/2020ms002274) documents a global model capable of reproducing the regime‐based effect of aerosols on cloud liquid water path expected from observational evidence. This may represent a significant advance in cloud process fidelity in global models. Such models can be expected to give a better estimate of the effective radiative forcing of the climate. If this advance in cloud process representation can be matched by advances in the representation of circulation features such as monsoons, then such models may also be able to navigate the complex tangle between spatially heterogeneous aerosol–cloud interactions and regional circulation patterns. This tight link between aerosol and circulation results in anthropogenic perturbations of climate variables of societal importance, such as regional rainfall distributions. Upcoming global models with km‐scale resolution may improve the regional circulation and be able to take advantage of the Terai et al. (2020, https://doi.org/10.1029/2020ms002274) improvement in cloud physics. If so, an era of significantly improved regional climate projection capabilities may soon dawn. If not, then the improvement in cloud physics might spur intensified efforts on problems in model dynamics. Either way, based on the rapid changes in aerosol emissions in the near future, learning to make reliable projections based on biased models is a skill that will not go out of style.

54 ENVIRONMENTAL SCIENCES↗

How Rains and Floods Become Groundwater: Understanding Recharge Pathways With Stable and Cosmogenic Isotopes

Anthropogenic climate change leads to increased precipitation intensity and exacerbated droughts in California, challenging the reliability and drought resiliency of water supply. Storing floodwater underground via managed aquifer recharge can mitigate these effects through direct infiltration or streambed infiltration. Seasonally dry streams (arroyos) already play an important part in managing groundwater recharge to the Livermore basin (CA). Understanding how, when and where stormwater and arroyo water infiltrate is critical to effectively utilise this strategy. To track water from recent storms (water year 2022–2023, WY23) into the Livermore Valley Groundwater Basin, we analysed stable water isotopes (δ 18 O and δ 2 H) in combination with naturally occurring radioactive isotopic tracers, sulphur-35 ( 35 S, t ½ = 87 days) and tritium ( 3 H, t ½ = 12.3 years). By comparing measurements of δ 18 O, 35 S and 3 H in arroyos to precipitation and groundwater, we classified the relative age and identified source of recharge to 16 wells near two arroyos. Two wells contained water with recent recharge (from WY23) from local precipitation. One well had recent recharge from variable (precipitation and imported water) sources. One well contained imported water recharge. Three wells contained water from mixed recent and older (pre-WY23) waters, from local precipitation sources. Two wells contained recent recharge from local mine settling ponds. Seven wells had older recharge from local precipitation sources. This combination of isotopes allows us to delineate where local and imported water recharges in this highly managed basin and identify locations where managed aquifer recharge is contributing to rapid groundwater infiltration. Our combined interpretation of isotopic water ages and sources in the context of land use shows that local infiltration of precipitation in open spaces is an important recharge mechanism, in addition to the managed arroyo recharge. Finally, a broader familiarity with 35 S will enable more extensive research on the infiltration of urban floodwaters.

58 GEOSCIENCES↗

Spatial and Temporal Variability of Aerosol Particles in Arctic Spring

The objective of this work is to investigate the variability in the particle number concentration that may affect climate change assessment for Arctic regions. The Indirect and Semi-Direct Aerosol Campaign (ISDAC) was conducted in April 2008, in the vicinities of Fairbanks and Barrow, Alaska. Measurements of particle number concentrations and size distributions were conducted using a Passive Cavity Aerosol Spectrometer Probe (PCASP-100X) mounted under the Convair-580 aircraft wing. Total number concentration of particles (Na) with diameters in the range 0.12-3 μm was determined for polluted and clean air masses during times when the air was free of clouds and/or precipitation. Variability in Na was considered for both vertical profiles and constant altitude (horizontal) flight legs. This variability can have important implications for estimates of particle properties used in global climate model (GCM) simulations. When aerosol particle layers were encountered, Na rapidly increased from 25 cm-3 up to 550 cm-3 within relatively clean air masses, and reached up to 2200 cm-3 within polluted air masses, dominated by biomass burning pollution. When averaging Na over different distance scales, it was found that Na=140 cm-3 represent an average value for the majority of the encountered clean cases; while Na=720 cm-3 is a mean for polluted cases dominated by biomass burning plumes. These estimates, however, would not capture the details of particle layers encountered during most of the flights. Average aerosol particle characteristics can be difficult to interpret, especially during polluted cases, due to small-scale spatial and temporal variability.

Shantz, Nicole C↗

Key environmental and production factors for understanding variation in switchgrass chemical attributes

Abstract Switchgrass ( Panicum virgatum L.) is a promising feedstock for bioenergy and bioproducts; however, its inherent variability in chemical attributes creates challenges for uniform conversion efficiencies and product quality. It is necessary to understand the range of variation and factors (i.e., field management, environmental) influencing chemical attributes for process improvement and risk assessment. The objectives of this study were to (1) examine the impact of nitrogen fertilizer application rate, year, and location on switchgrass chemical attributes, (2) examine the relationships among chemical attributes, weather and soil data, and (3) develop models to predict chemical attributes using environmental factors. Switchgrass samples from a field study spanning four locations including upland cultivars, one location including a lowland cultivar, and between three and six harvest years were assessed for glucan, xylan, lignin, volatiles, carbon, nitrogen, and ash concentrations. Using variance estimation, location/cultivar, nitrogen application rate, and year explained 65%–96% of the variation for switchgrass chemical attributes. Location/cultivar × year interaction was a significant factor for all chemical attributes indicating environmental‐based influences. Nitrogen rate was less influential. Production variables and environmental conditions occurring during the switchgrass field trials were used to successfully predict chemical attributes using linear regression models. Upland switchgrass results highlight the complexity in plant responses to growing conditions because all production and environmental variables had strong relationships with one or more chemical attributes. Lowland switchgrass was limited to observations of year‐to‐year environmental variability and nitrogen application rate. All explanatory variable categories were important for lowland switchgrass models but stand age and precipitation relationships were particularly strong. The relationships found in this study can be used to understand spatial and temporal variation in switchgrass chemical attributes. The ability to predict chemical attributes critical for conversion processes in a geospatial/temporal manner would provide state‐of‐the‐art knowledge for risk assessment in the bioenergy and bioproducts industry.

09 BIOMASS FUELS↗

Modelling above-ground biomass stock over Norway using national forest inventory data with ArcticDEM and Sentinel-2 data

Boreal forests constitute a large portion of the global forest area, yet they are undersampled through field surveys, and only a few remotely sensed data sources provide structural information wall-to-wall throughout the boreal domain. ArcticDEM is a collection of high-resolution (2 m) space-borne stereogrammetric digital surface models (DSM) covering the entire land area north of 60° of latitude. The free-availability of ArcticDEM data offers new possibilities for aboveground biomass mapping (AGB) across boreal forests, and thus it is necessary to evaluate the potential for these data to map AGB over alternative open-data sources (i.e., Sentinel-2). This study was performed over the entire land area of Norway north of 60° of latitude, and the Norwegian national forest inventory (NFI) was used as a source of field data composed of accurately geolocated field plots (n=7710) systematically distributed across the study area. Separate random forest models were fitted using NFI data, and corresponding remotely sensed data consisting of either: i) a canopy height model (ArcticCHM) obtained by subtracting a high-quality digital terrain model (DTM) from the ArcticDEM DSM height values, ii) Sentinel-2 (S2), or iii) a combination of the two (ArcticCHM+S2). Furthermore, we assessed the effect of the forest- and terrain-specific factors on the models’ predictive accuracy. The best model (,i.e., ArcticCHM+S2) explained nearly 60% of the variance of the training set, which translated in the largest accuracy in terms of root mean square error (RMSE=41.4 t/ha). This result highlights the synergy between 3D and multispectral data in AGB modelling. Furthermore, this study showed that despite the importance of ArcticCHM variables, the S2 model performed slightly better than ArcticCHM model. This finding highlights some of the limitations of ArcticDEM, which, despite the unprecedented spatial resolution, is highly heterogeneous due to the blending of multiple acquisitions across different years and seasons. We found that both forest- and terrain-specific characteristics affected the uncertainty of the ArcticCHM+S2 model and concluded that the combined use of ArcticCHM and Sentinel-2 represents a viable solution for AGB mapping across boreal forests. The synergy between the two data sources allowed for a reduction of the saturation effects typical of multispectral data while ensuring the spatial consistency in the output predictions due to the removal of artifacts and data voids present in ArcticCHM data. While the main contribution of this study is to provide the first evidence of the best-case-scenario (i.e., availability of accurate terrain models) that ArcticDEM data can provide for large-scale AGB modelling, it remains critically important for other studies to investigate how ArcticDEM may be used in areas where no DTMs are available as is the case for large portions of the boreal zone.

space-borne imagery↗

Outcomes of the DOE Workshop on Atmospheric Challenges for the Wind Energy Industry

The U.S. Department of Energy-funded Mesoscale-to-Microscale Coupling (MMC) project team planned and conducted a virtual Workshop on Atmospheric Challenges for the Wind Energy Industry on October 19 and 20, 2020. The goal of the workshop was to forge a dialog with the community, including industry representatives, on how modeling tools are currently being used, the present active atmospheric modeling research in support of wind energy, and required advancements in capabilities and technology to continue to advance wind energy deployment. The workshop was planned in collaboration with an industry advisory panel that included representatives from wind power plant developers, turbine manufacturers, and companies that provide resource assessment and forecasting services. The format of the workshop included panels from government research sponsors, visionaries from industry, and mixed panels of researchers discussing research status and needs. A shared keynote presentation from the Technical University of Denmark experts anchored the second day of the workshop. An emphasis was placed on understanding the research needs in the offshore environment. In addition, breakout opportunities were provided each day. On the first day, the breakout discussions addressed predesigned questions configured to elicit participants’ thoughts on needed research directions. The second-day breakouts treated three important technical topics through a combination of presentations and group conversations. Each workshop participant chose their breakout preference from among downscaling details, modeling for turbines, and using artificial intelligence for atmospheric modeling. The discussions were robust and productive. The outcomes of the workshop include archiving a series of recommendations from industry and the research community on research directions required to further advance wind energy deployment. Discussions confirmed the need for high-fidelity modeling but that there are specific areas of applicability and other areas where the time and cost of computation is prohibitive. In those cases, the high-fidelity models can inform low-order models that are more practical for real-time or widely deployed applications. Industry must consider the financial cost of performing more expensive modeling approaches, but industry engineers and researchers are using these approaches where there appears to be a return on investment. An emerging type of low-order model is based on machine learning (ML). Participants confirmed that there are many atmospheric phenomena that need to be modeled better, including low-level jets, cold air outbreaks, land-sea induced circulations, diurnal variability, thin stable boundary layers, dynamic changes such as from frontal passage, interaction of wakes and blockage, and more. For the offshore environment, there is wide agreement that some level of ocean-wave-atmospheric coupling is necessary to capture variations in rotor-level winds needed to plan and operate offshore wind plants. Another recurring recommendation is that more observations are needed, particularly for the offshore environment. Those observations should consider the needs for model improvement, both for physically based models and for ML models. Observations must capture atmospheric profiles of variables that are important to understanding and modeling atmospheric and oceanic phenomena that impact boundary layer winds. Models must be validated with data and the uncertainty quantified, particularly those that are sensitive to initial and boundary conditions. Finally, a repeated request was to consider the holistic needs of hybrid plants of wind, solar, and storage resources because those types of plants are likely to be the wave of the future. In addition, industry wishes to understand impacts of the resource under a changing climate for long-term planning.

17 WIND ENERGY↗

Statistical characteristics of cloud variability. Part 2: Implication for parameterizations of microphysical and radiative transfer processes in climate models

The effects of subgrid cloud variability on grid-average microphysical rates and radiative fluxes are examined by use of long-term retrieval products at the Tropical West Pacific, Southern Great Plains, and North Slope of Alaska sites of the Department of Energy's Atmospheric Radiation Measurement program. Four commonly used distribution functions, the truncated Gaussian, Gamma, lognormal, and Weibull distributions, are constrained to have the same mean and standard deviation as observed cloud liquid water content. The probability density functions are then used to upscale relevant physical processes to obtain grid-average process rates. It is found that the truncated Gaussian representation results in up to 30% mean bias in autoconversion rate, whereas the mean bias for the lognormal representation is about 10%. The Gamma and Weibull distribution function performs the best for the grid-average autoconversion rate with the mean relative bias less than 5%. For radiative fluxes, the lognormal and truncated Gaussian representations perform better than the Gamma and Weibull representations. The results show that the optimal choice of subgrid cloud distribution function depends on the nonlinearity of the process of interest, and thus, there is no single distribution function that works best for all parameterizations. Examination of the scale (window size) dependence of the mean bias indicates that the bias in grid-average process rates monotonically increases with increasing window sizes, suggesting the increasing importance of subgrid variability with increasing grid sizes.

54 ENVIRONMENTAL SCIENCES↗

Identifying recharge sources and their impacts on a North Central New Mexico shallow aquifer using unsupervised machine learning

In this article, shallow aquifers are important but highly variable resources in arid to semi-arid regions. Limited shallow aquifer volume results in high sensitivity to recharge fluctuations, which can impact the local fauna and flora, and transport of contaminants in the aquifer or vadose zone. Aquifer response to external forcing (e.g., precipitation) is usually solved by estimating aquifer parameters and running physics-based models to match known fluctuations of hydraulic head. However, this technique is time and computationally expensive. Furthermore, high aquifer complexity decreases precision in physics-based models. Alternatively supervised machine learning is used to predict aquifer dynamics. However, these techniques rely on input data and struggle to interpret aquifer response for missing sources (i.e., snowpack data). To counter these problems, we propose an unsupervised machine learning technique (NMFk) to estimate the impact of different sources on aquifer recharge. NMFk is used to understand the influence of external forcing on shallow aquifer recharge in the Pajarito Plateau (Los Alamos, NM, USA). The results show how NMFk can be used to reduce the data dimension in a complex field dataset to three recharge signals that cause fluctuations within the field data. Here, the source signals are interpreted as rainfall, snowmelt, and a delayed aquifer response to the previous two signals. These results evidence how heterogeneous aquifers delimited by canyons incised into the Pajarito Plateau respond in similar ways to the source signals identified by NMFk. Furthermore, results show the importance of the local geology where faults act as sinks, and anthropogenic disturbances can facilitate infiltration amplifying the interpreted signal.

54 ENVIRONMENTAL SCIENCES↗

Exploring new topography-based subgrid spatial structures for improving land surface modeling

Topography plays an important role in land surface processes through its influence on atmospheric forcing, soil and vegetation properties, and river network topology and drainage area. Land surface models with a spatial structure that captures spatial heterogeneity, which is directly affected by topography, may improve the representation of land surface processes. Previous studies found that land surface modeling, using subbasins instead of structured grids as computational units, improves the scalability of simulated runoff and streamflow processes. In this study, new land surface spatial structures are explored by further dividing subbasins into subgrid structures based on topographic properties, including surface elevation, slope and aspect. Two methods (local and global) of watershed discretization are applied to derive two types of subgrid structures (geo-located and non-geo-located) over the topographically diverse Columbia River basin in the northwestern United States. In the global method, a fixed elevation classification scheme is used to discretize subbasins. The local method utilizes concepts of hypsometric analysis to discretize each subbasin, using different elevation ranges that also naturally account for slope variations. The relative merits of the two methods and subgrid structures are investigated for their ability to capture topographic heterogeneity and the implications of this on representations of atmospheric forcing and land cover spatial patterns. Results showed that the local method reduces the standard deviation (SD) of subgrid surface elevation in the study domain by 17 to 19 % compared to the global method, highlighting the relative advantages of the local method for capturing subgrid topographic variations. The comparison between the two types of subgrid structures showed that the non-geo-located subgrid structures are more consistent across different area threshold values than the geo-located subgrid structures. Altogether the local method and non-geo-located subgrid structures effectively and robustly capture topographic, climatic and vegetation variability, which is important for land surface modeling.

58 GEOSCIENCES↗

Shape optimization using a NURBS-based interface-enriched generalized FEM

This study presents a gradient-based shape optimization over a fixed mesh using a non-uniform rational B-splines-based interface-enriched generalized finite element method, applicable to multi-material structures. In the proposed method, non-uniform rational B-splines are used to parameterize the design geometry precisely and compactly by a small number of design variables. An analytical shape sensitivity analysis is developed to compute derivatives of the objective and constraint functions with respect to the design variables. Subtle but important new terms involve the sensitivity of shape functions and their spatial derivatives. As a result, verification and illustrative problems are solved to demonstrate the precision and capability of the method.

42 ENGINEERING↗

Tradeoffs between phenotypic plasticity and local adaptation influence the ecophysiology of the moss, Sphagnum magellanicum

Bryophytes are a diverse plant group and are functionally different from vascular plants. Yet, their peculiarities are rarely considered in the theoretical frameworks for plants. Currently, we lack information about the magnitude and the importance of intraspecific variability in the ecophysiology of bryophytes and how these might translate to local adaptation—a prerequisite for adaptive evolution. Capitalizing on two ecologically distinct (hummock and hollow) phenotypes of Sphagnum magellanicum, we explored the magnitude and pattern of intraspecific variability in this species and asked whether the environmental-mediated changes in shoot and physiological traits are due to phenotypic plasticity or local adaptation. Size, pigmentation, and habitat type that distinguished the phenotypes in the field did not influence the trait responses under a transplant and factorial experiment. In addition, the magnitude and pattern of trait variability (e.g., branch, stem and capitulum mass) changed with the treatments, which suggest that trait responses were due largely to phenotypic plasticity. The trait responses also suggest that the ecophysiological needs for mosses to grow in clumps, where they maintain a uniform growth may have an overriding effect over the potential for a fixed adaptive response to environmental heterogeneity, which would constrain local adaptation. In this paper, we conclude that extending the trait-based framework to mosses or making comparisons between mosses and vascular plants under any theoretical framework would only be meaningful to the extent that growth form and dispersal strategies are considered.

59 BASIC BIOLOGICAL SCIENCES↗

Post-fire soil emissions of nitric oxide (NO) and nitrous oxide (N 2 O) across global ecosystems: a review

Wildfires may increase soil emissions of trace nitrogen (N) gases like nitric oxide (NO) and nitrous oxide (N 2 O) by changing soil physicochemical conditions and altering microbial processes like nitrification and denitrification. When 34 studies were synthesized, we found a significant increase in both NO and N 2 O emissions up to 1 year post-fire across studies spanning ecosystems globally. However, when fluxes were separated by ecosystem type, we found that individual ecosystem types responded uniquely to fire. Forest soils tended to emit more N 2 O after fire, but there was no significant effect on NO. Shrubland soils showed significant increases in both NO and N 2 O emissions after fires; often with extremely large but short-lived NO pulses occurring immediately after fire. Grassland NO emissions increased after fire, but the size of this effect was small relative to shrublands. N 2 O emissions from burned grasslands were highly variable with no significant effect. To better understand the variation in responses to fire across global ecosystems, more consistent measurements of variables recognized as important controls on soil fluxes of NO and N 2 O (e.g., N cycling rates, soil water content, pH, and substrate availability) are needed across studies. We also suggest that fire-specific elements like burn severity, microbial community succession, and the presence of char be considered by future studies. Our synthesis suggests that fires can exacerbate ecosystem N loss long after they burn, increasing soil emissions of NO and N 2 O with implications for ecosystem N loss, climate, and regional air quality as wildfires increase globally.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Root exudation links root traits to soil functioning in agroecosystems

Root exudation is a key process for plant nutrient acquisition, but the controls on root exudation and its relationship to soil C and N processes in agroecosystems are unclear. We hypothesized that root exudation rates would be related to root morphological traits, N fertilization, and soil moisture. We also anticipated that root exudation would be correlated with bulk soil enzyme activity. Root exudation, root traits, and bulk soil extracellular enzyme activity were assessed in maize (Zea mays L.), soybean (Glycine max (L.) Merr.), biomass sorghum (Sorghum bicolor (L.) Moench), giant miscanthus (Miscanthus × giganteus), and switchgrass (Panicum virgatum L.). Measurements were taken in situ during two growing seasons with contrasting precipitation regimes, and N fertilization rate was varied in sorghum during one year. Specific root exudation (per unit root surface area) was negatively related to root diameter and was generally higher in annuals than perennials. Sorghum N fertilization did not affect root exudation rates, and soil moisture regime had no effect on annual root exudation rates within maize, sorghum, and miscanthus. Specific root exudation was negatively related to bulk soil C- and N-degrading soil enzyme activities. Intrinsic plant characteristics appeared more important than environmental variables in controlling in situ root exudation rates. The relationships between root diameter, root exudation, and soil C and N processes link root morphological traits to soil functions and demonstrate the potential tradeoffs among plant nutrient acquisition strategies in agroecosystems.

54 ENVIRONMENTAL SCIENCES↗

Variation in Biomass Yield and Cell Wall Composition in Switchgrass Natural Variants Under Two Nitrogen Regimes

Switchgrass (Panicum virgatum) is a promising lignocellulosic biofuel crop for which biomass and processing quality are important. Inherent plant variability across genotypes and environments challenges uniformity and product quality. In this study, the impact of nitrogen (N) application on switchgrass yield and quality was examined under field conditions using a highly diverse switchgrass panel over a 4-year period at Knoxville, TN. Overall, biomass production was correlated between low (0 kg of added N/ha) and moderate (135 kg of added N/ha) nitrogen treatments, suggesting that the N impact is largely uniform across the genotypes. Nonetheless, high biomass genotypes were identified with high nitrogen-use efficiency ; biomass was congruent or even higher (up to 9-fold) in the low N treatment. Genotypes were also identified with up to 94% nitrogen-remobilization efficiency. Furthermore, nitrogen application appeared to impact lignin content in whole tillers but was neutral to lignin monomer syringyl-to-guaiacyl (S/G) ratio. The same panel grown under natural conditions in Watkinsville, GA, produced significantly more biomass than the Tennessee panel in the first 3 years, but biomass was similar across both sites and treatments in year 4. Top performing genotypes overlapped between sites by 20-37%. There were low correlations in lignin content in whole tillers across the two field sites, but moderate correlations were observed for S/G ratios. The high yielding genotypes from the low N plot identified in this study can be used in breeding programs and management strategies in switchgrass to evade adverse environmental and economic effects.

09 BIOMASS FUELS↗

Evaluation and modification of ELM seasonal deciduous phenology against observations in a southern boreal peatland forest

Phenological transitions determine the timing of changes in land surface properties and the seasonality of exchanges of biosphere-atmosphere energy, water, and carbon. Accurate mechanistic modeling of phenological processes is therefore critical to understand and correctly predict terrestrial ecosystem feedbacks with changing atmospheric and climate conditions. However, the phenological components in the land model of the US Department of Energy's (DOE) Energy Exascale Earth System Model (ELM of E3SM) were previously unable to accurately capture the observed phenological responses to environmental conditions in a well-studied boreal peatland forest. In this research, we introduced new seasonal-deciduous phenology schemes into version 1.0 of ELM and evaluated their performance against the PhenoCam observations at the Spruce and Peatland Responses Under Changing Environments (SPRUCE) experiment in northern Minnesota from 2015 to 2018. We found that phenology simulated by the revised ELM (i.e., earlier spring onsets and stronger warming responses of spring onsets and autumn senescence) was closer to observations than simulations from the original algorithms for both the deciduous conifer (Larix laricina) and mixed shrub layers. Moreover, the revised ELM generally produced higher carbon and water fluxes (e.g., photosynthesis and evapotranspiration) during the growing season and stronger flux responses to warming than the default ELM. A parameter sensitivity analysis further indicated the significant contribution of phenology parameters to uncertainty in key carbon and water cycle variables, underscoring the importance of precise phenology parameterization. Furthermore, this phenological modeling effort demonstrates the potential to enhance the E3SM representation of land-climate interactions at broader spatiotemporal scales, especially under anticipated elevated CO 2 and warming conditions.

54 ENVIRONMENTAL SCIENCES↗

Constrained or unconstrained? Neural-network-based equation discovery from data

Throughout many fields, practitioners often rely on differential equations to model systems. Yet, for many applications, the theoretical derivation of such equations and/or the accurate resolution of their solutions may be intractable. Instead, recently developed methods, including those based on parameter estimation, operator subset selection, and neural networks, allow for the data-driven discovery of both ordinary and partial differential equations (PDEs), on a spectrum of interpretability. The success of these strategies is often contingent upon the correct identification of representative equations from noisy observations of state variables and, as importantly and intertwined with that, the mathematical strategies utilized to enforce those equations. Specifically, the latter has been commonly addressed via unconstrained optimization strategies. Representing the PDE as a neural network, we propose to discover the PDE (or the associated operator) by solving a constrained optimization problem and using an intermediate state representation similar to a physics-informed neural network (PINN). The objective function of this constrained optimization problem promotes matching the data, while the constraints require that the discovered PDE is satisfied at a number of spatial collocation points. We present a penalty method and a widely used trust-region barrier method to solve this constrained optimization problem, and we compare these methods on numerical examples. Our results on several example problems demonstrate that the latter constrained method outperforms the penalty method, particularly for higher noise levels or fewer collocation points. This work motivates further exploration into using sophisticated constrained optimization methods in scientific machine learning, as opposed to their commonly used, penalty-method or unconstrained counterparts. For both of these methods, we solve these discovered neural network PDEs with classical methods, such as finite difference methods, as opposed to PINNs-type methods relying on automatic differentiation. Here, we briefly highlight how simultaneously fitting the data while discovering the PDE improves the robustness to noise and other small, yet crucial, implementation details.

Data-driven discovery↗