Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “tool size”

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 1,027 records · Page 57

Social network structure and the spread of complex contagions from a population genetics perspective

Ideas, behaviors, and opinions spread through social networks. If the probability of spreading to a new individual is a non-linear function of the fraction of the individuals’ affected neighbors, such a spreading process becomes a “complex contagion”. This non-linearity does not typically appear with physically spreading infections, but instead can emerge when the concept that is spreading is subject to game theoretical considerations (e.g. for choices of strategy or behavior) or psychological effects such as social reinforcement and other forms of peer influence (e.g. for ideas, preferences, or opinions). Here we study how the stochastic dynamics of such complex contagions are affected by the underlying network structure. Motivated by simulations of complex contagions on real social networks, we present a framework for analyzing the statistics of contagions with arbitrary non-linear adoption probabilities based on the mathematical tools of population genetics. The central idea is to use an effective lower-dimensional diffusion process to approximate the statistics of the contagion. This leads to a tradeoff between the effects of ”selection” (microscopic tendencies for an idea to spread or die out), random drift, and network structure. Our framework illustrates intuitively several key properties of complex contagions: stronger community structure and network sparsity can significantly enhance the spread, while broad degree distributions dampen the effect of selection compared to random drift. Finally, we show that some structural features can exhibit critical values that demarcate regimes where global contagions become possible for networks of arbitrary size. Our results draw parallels between the competition of genes in a population and memes in a world of minds and ideas. Our tools provide insight into the spread of information, behaviors, and ideas via social influence, and highlight the role of macroscopic network structure in determining their fate.

59 BASIC BIOLOGICAL SCIENCES↗

REopt Lite Web Tool

NREL developed a free, publicly available web version of the REopt (TM) renewable energy integration and optimization platform called REopt Lite. REopt Lite recommends the optimal size and dispatch strategy for grid-connected photovoltaics (PV) and battery storage at a site. It also allows users to explore how PV and storage can increase a site's resiliency during a grid outage.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Microfield exposure tool enables advances in EUV lithography development

With demonstrated resist resolution of 20 nm half pitch, the SEMATECH Berkeley BUV microfield exposure tool continues to push crucial advances in the areas of BUY resists and masks. The ever progressing shrink in computer chip feature sizes has been fueled over the years by a continual reduction in the wavelength of light used to pattern the chips. Recently, this trend has been threatened by unavailability of lens materials suitable for wavelengths shorter than 193 nm. To circumvent this roadblock, a reflective technology utilizing a significantly shorter extreme ultraviolet (EUV) wavelength (13.5 nm) has been under development for the past decade. The dramatic wavelength shrink was required to compensate for optical design limitations intrinsic in mirror-based systems compared to refractive lens systems. With this significant reduction in wavelength comes a variety of new challenges including developing sources of adequate power, photoresists with suitable resolution, sensitivity, and line-edge roughness characteristics, as well as the fabrication of reflection masks with zero defects. While source development can proceed in the absence of available exposure tools, in order for progress to be made in the areas of resists and masks it is crucial to have access to advanced exposure tools with resolutions equal to or better than that expected from initial production tools. These advanced development tools, however, need not be full field tools. Also, implementing such tools at synchrotron facilities allows them to be developed independent of the availability of reliable stand-alone BUY sources. One such tool is the SEMATECH Berkeley microfield exposure tool (MET). The most unique attribute of the SEMA TECH Berkeley MET is its use of a custom-coherence illuminator made possible by its implementation on a synchrotron beamline. With only conventional illumination and conventional binary masks, the resolution limit of the 0.3-NA optic is approximately 25 nm, however, with EUV not expected in production before the 22-nm half pitch node even finer resolution capabilities are now required from development tools. The SEMATECH Berkeley MET's custom-coherence illuminator allows it to be used with aggressive modified illumination enabling kJ factors as low as 0.25. Noting that the lithographic resolution of an exposure tool is defined as k{sub 1}{lambda}/NA, yielding an ultimate resolution limit of 11 nm. To achieve sub-20-nm aerial-image resolution while avoiding forbidden pitches on Manhattan-geometry features with the centrally-obscured MET optic, a 45-degree oriented dipole pupil fill is used. Figure 1 shows the computed aerial-image contrast as a function of half pitch for a dipole pupil fill optimized to print down to the 19-nm half pitch level. This is achieved with relatively uniform performance at larger dimensions. Using this illumination, printing down to the 20-nm half pitch level has been demonstrated in chemically amplified resists as shown in Fig. 2. The SEMATECH Berkeley MET tool plays a crucial role in the advancement of EUV resists. The unique programmable coherence properties of this tool enable it to achieve higher resolution than other EUV projection tools. As presented here, over the past year the tool has been used to demonstrate resist resolutions of 20 half pitch. Although not discussed here, because the Berkeley MET tool is a true projection lithography tool, it also plays a crucial role in advanced EUV mask research. Examples of the work done in this area include defect printability, mask architecture, and phase shift masks.

36 MATERIALS SCIENCE↗

Tuning carbon molecular sieves for natural gas separations: A diamine molecular approach

This article introduces a new postsynthetic modification method for tuning the separation properties of carbon molecular sieve (CMS) membranes. Polymeric hollow fibers of 6FDA/BPDA-DAM were pyrolyzed to 550°C under inert Argon, and were then exposed to a tetrahydrofuran solution containing PPM levels of paraphenylenediamine dopant. The original goal of the treatment was to modify the morphology in a manner to prevent relaxation of the CMS to suppress physical aging, envisioned to be analogous to that in glassy polymer membranes where reduction of “unrelaxed free volume” causes reduced permeance with a mild increase in selectivity. Results of long-term CO 2 and CH 4 permeation experiments of doped fibers showed that aging-resistant membranes did not result. This fact notwithstanding, results revealed this approach provides a valuable new tuning tool for CMS transport properties. Complementary gas sorption experiments performed suggest the amine dopants tune large ultramicropores, thereby increasing size and shape based diffusion selectivity.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Retracted: Monitoring Hatchery Broodstock Composition and Genetic Variation of Spring/Summer Chinook Salmon in the Columbia River Basin with Multigeneration Pedigrees

Abstract Hatchery production of Chinook Salmon Oncorhynchus tshawytscha in the Columbia River basin comprises most of the anadromous salmonid production in this region. Hatchery facilities and programs serve to mitigate for impacts to salmonids due to the construction and operation of hydropower dams and habitat impacts from development in addition to the conservation and restoration of natural populations. A genetic method referred to as parentage-based tagging (PBT) enables highly reliable detection of hatchery-origin fish and inference of multigeneration pedigrees. This study compiles 11 years of PBT data from nearly 125,000 interior stream-type Chinook Salmon from 24 spawning hatcheries located in tributaries of the mid- and upper Columbia River as well as the Salmon, Clearwater, and Grande Ronde River subbasins. Multigenerational pedigrees allowed investigation of the proportion of natural- and hatchery-origin broodstock (pNOB and pHOB, respectively) for each hatchery and enumeration of the scale of production between segregated and integrated programs. We then compared how pHOB and the scale of production influenced the number of stray fish observed, genetic diversity, relatedness, and age-class compositions within broodstocks. Over 91.0% of hatchery broodstock could be assigned back to their parents, and there was an overall rate of less than 1.0% of broodstock that were unintentionally incorporated into nonnatal hatchery programs. We evaluated 14 segregated programs with 0.0–10.0% pNOB, 7 intermediate programs (10.1–50.0% pNOB), and 3 integrated programs (>50.0% pNOB). There was no correlation between the scale of production or pNOB with the level of genetic diversity, but as production size increased, so did the effective number of breeders. This study demonstrates the utility of PBT as a monitoring tool for hatchery broodstocks, and results suggest that segregated and integrated programs have tradeoffs that generally align with their intended broodstock management purpose of providing fish for harvest and/or fish for supplementation or reintroduction.

Horn, Rebekah L.↗

Multigeneration Pedigrees to Monitor Hatchery Broodstock Composition and Genetic Variation of Spring/Summer Chinook Salmon in the Columbia River Basin

Abstract Hatchery production of Chinook Salmon Oncorhynchus tshawytscha in the Columbia River basin comprises most of the anadromous salmonid production in this region. Hatchery facilities and programs serve to mitigate for impacts to salmonids due to the construction and operation of hydropower dams and habitat impacts from development in addition to the conservation and restoration of natural populations. A genetic method referred to as parentage-based tagging (PBT) enables highly reliable detection of hatchery-origin fish and inference of multigeneration pedigrees. This study compiles 11 years of PBT data from nearly 125,000 interior stream-type Chinook Salmon from 24 spawning hatcheries located on tributaries of the mid- and upper Columbia River and in the Salmon, Clearwater, and Grande Ronde River subbasins. Multigenerational pedigrees allowed for investigation of the proportions of natural- and hatchery-origin broodstock (pNOB and pHOB, respectively) for each hatchery and enumeration of the scale of production between segregated and integrated programs. We then compared how pHOB and the scale of production influenced the number of stray fish observed, genetic diversity, relatedness, and age-class compositions within broodstocks. Over 91.0% of hatchery broodstock could be assigned back to their parents, and overall less than 1.0% of broodstock consisted of fish that were unintentionally incorporated into nonnatal hatchery programs. We evaluated 11 segregated programs with 0.0–10.0% pNOB, 9 intermediate programs (10.1–50.0% pNOB), and 3 integrated programs (>50.0% pNOB). There was no correlation between the scale of production or pNOB with the level of genetic diversity, but as production size increased, so did the effective number of breeders. This study demonstrates the utility of PBT as a monitoring tool for hatchery broodstocks, and results suggest that segregated and integrated programs have tradeoffs that generally align with their intended broodstock management purpose of providing fish for harvest and/or fish for supplementation or reintroduction.

Horn, Rebekah L.↗

Optimizing Tensor Contraction Expressions for Hybrid CPU-GPU Execution

Tensor contractions are generalized multidimensional matrix multiplication operations that widely occur in quantum chemistry. Efficient execution of tensor contractions on Graphics Processing Units (GPUs) requires several challenges to be addressed, including index permutation and small dimension-sizes reducing thread block utilization. Moreover, to apply the same optimizations to various expressions, we need a code generation tool. In this paper, we present our approach to automatically generate CUDA code to execute tensor contractions on GPUs, including management of data movement between CPU and GPU. To evaluate our tool, GPU-enabled code is generated for the most expensive contractions in CCSD(T), a key coupled cluster method, and incorporated into NWChem, a popular computational chemistry suite. For this method, we demonstrate speedup over a factor of 8.4 using one GPU (instead of one core per node) and over 2.6 when utilizing the entire system using hybrid CPU+GPU solution with 2 GPUs and 5 cores (instead of 7 cores per node). Finally, we analyze the implementation behavior on future GPU systems.

CUDA↗

Enhancing Network Visibility and Security through Tensor Analysis

The increasing size, variety, rate of growth and change, and complexity of network data has warranted advanced network analysis and services. Tools that provide automated analysis through traditional or advanced signature-based systems or machine learning classifiers suffer from practical difficulties. These tools fail to provide comprehensive and contextual insights into the network when put to practical use in operational cyber security. In this paper, we present an effective tool for network security and traffic analysis that uses high-performance data analytics based on a class of unsupervised learning algorithms called tensor decompositions. The tool aims to provide a scalable analysis of the network traffic data and also reduce the cognitive load of network analysts and be network-expert-friendly by presenting clear and actionable insights into the network. With this being said, In this paper, we demonstrate the successful use of the tool in two completely diverse operational cyber security environments, namely, (1) security operations center (SOC) for the SCinet network at the SuperComputing (SC) Conference in 2016 and 2017 and (2) Reservoir Labs’ Local Area Network (LAN). In each of these environments, we produce actionable results for cyber security specialists including (but not limited to) (1) finding malicious network traffic involving internal and external attackers using port scans, SSH brute forcing, and NTP amplification attacks, (2) uncovering obfuscated network threats such as data exfiltration using DNS port and using ICMP traffic, and (3) finding network misconfiguration and performance degradation patterns.

97 MATHEMATICS AND COMPUTING↗

Quantitative analysis of Cu XANES spectra using linear combination fitting of binary mixtures simulated by FEFF9

We report x-ray absorption near edge structure (XANES) is a powerful tool to probe the fingerprint of local structures, and when coupled with X-ray microscopy, the small spot size enables one to probe very specific regions of interest in a device or material, e.g. interfaces/bulk, different grains, good/bad electrical areas. In this work, we investigate the use of linear combination fitting (LCF) of XANES spectra for the particular case of Cu doping in CdTe. We show that the experimental data seem to be accurately represented by standards of Cu 2 Te and its substoichiometric counterpart, Cu 1.43 Te. We use Cu in CdTe as a case study to evaluate the accuracy of linear combination fitting using simulated standards, given that experimental standards for certain phases (e.g. Cu 1.43 Te) or defect structures (e.g. vacancies) cannot be readily obtained. We discuss how spectral features of the FEFF9-simulated standards, fitting ranges, and noise levels all dictate the accuracy of this type of analysis. We show that the greater the spectral difference between the two standards, the better the LCF is able to differentiate between the two structures and to tolerate experimental noise. Finally, we estimate the error of the fitted weights for different spectral features and noise levels and propose a framework to study local structures semi-quantitatively by using binary mixtures of FEFF9-simulated standards.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

2-D energy analyzer for low energy electrons

A 2-D electron energy analyzer is designed and constructed to measure the transverse and longitudinal energy distribution of low energy (<1 eV) electrons. The analyzer operates on the principle of adiabatic invariance and motion of low energy electrons in a strong longitudinal magnetic field. The operation of the analyzer is studied in detail and a design to optimize the energy resolution, signal to noise ratio, and physical size is presented. An energy resolution better than 6 meV has been demonstrated. Such an analyzer is a powerful tool to study the process of photoemission which limits the beam quality in modern accelerators.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Goldilocks models of higher-dimensional inflation (including modulus stabilization)

We explore the mechanics of inflation within simplified extra-dimensional models involving an inflaton interacting with the Einstein-Maxwell system in two extra dimensions. The models are Goldilocks-like inasmuch as they are just complicated enough to include a mechanism to stabilize the extra-dimensional size (or modulus), yet simple enough to solve explicitly the full extra-dimensional field equations using only simple tools. The solutions are not restricted to the effective 4D regime with H || m {sub KK} (the latter referring to the characteristic mass splitting of the Kaluza-Klein excitations) because the full extra-dimensional Einstein equations are solved. This allows an exploration of inflationary physics in a controlled calculational regime away from the usual four-dimensional lamp-post. The inclusion of modulus stabilization is important because experience with string models teaches that this is usually what makes models fail: stabilization energies easily dominate the shallow potentials required by slow roll and so open up directions to evolve that are steeper than those of the putative inflationary direction. We explore (numerically and analytically) three representative kinds of inflationary scenarios within this simple setup. In one the radion is trapped in an inflaton-dependent local minimum whose non-zero energy drives inflation. Inflation ends as this energy relaxes to zero when the inflaton finds its own minimum. The other two involve power-law scaling solutions during inflation. One of these is a dynamical attractor whose features are relatively insensitive to initial conditions but whose slow-roll parameters cannot be arbitrarily small; the other is not an attractor but can roll much more slowly, until eventually transitioning to the attractor. The scaling solutions can satisfy H > m {sub KK}, but when they do standard 4D fluctuation calculations need not apply. When in a 4D regime the solutions predict η ≅ 0 and so r ≅ 0.11 when n {sub s} ≅ 0.96 and so are ruled out if tensor modes remain unseen. Assessment of general parameters is difficult until a full 6D fluctuation calculation is done.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Mechanism of surface modification in the plasma-surface interaction in electrical arcs

Electrical sparks and arcs are plasma discharges that carry large currents and can strongly modify surfaces. This damage usually comes in the form of micrometer-sized craters and frozen-in liquid on the surface. Using a combination of experiments, plasma and atomistic simulation tools, we now show that the observed formation of deep craters and liquidlike features during sparking in vacuum is explained by the impacts of energetic ions, accelerated under the given conditions in the plasma sheath to kiloelectron volt energies, on surfaces. The flux in arcs is so high that in combination with kiloelectron volt energies it produces multiple overlapping heat spikes, which can lead to cratering even in materials such as Cu, where a single heat spike normally does not.

36 MATERIALS SCIENCE↗

SU-E-J-36: Comparison of CBCT Image Quality for Manufacturer Default Imaging Modes

Purpose CBCT is being increasingly used in patient setup for radiotherapy. Often the manufacturer default scan modes are used for performing these CBCT scans with the assumption that they are the best options. To quantitatively assess the image quality of these scan modes, all of the scan modes were tested as well as options with the reconstruction algorithm. Methods A CatPhan 504 phantom was scanned on a TrueBeam Linear Accelerator using the manufacturer scan modes (FSRT Head, Head, Image Gently, Pelvis, Pelvis Obese, Spotlight, & Thorax). The Head mode scan was then reconstructed multiple times with all filter options (Smooth, Standard, Sharp, & Ultra Sharp) and all Ring Suppression options (Disabled, Weak, Medium, & Strong). An open source ImageJ tool was created for analyzing the CatPhan 504 images. Results The MTF curve was primarily dictated by the voxel size and the filter used in the reconstruction algorithm. The filters also impact the image noise. The CNR was worst for the Image Gently mode, followed by FSRT Head and Head. The sharper the filter, the worse the CNR. HU varied significantly between scan modes. Pelvis Obese had lower than expected HU values than most while the Image Gently mode had higher than expected HU values. If a therapist tried to use preset window and level settings, they would not show the desired tissue for some scan modes. Conclusion Knowing the image quality of the set scan modes, will enable users to better optimize their setup CBCT. Evaluation of the scan mode image quality could improve setup efficiency and lead to better treatment outcomes.

60 APPLIED LIFE SCIENCES↗

HAERO: High-performance aerosols

SAND2021-15059 O Haero is a software library for representing aerosol size distributions, chemical species, and parameterized processes to use with a coupled atmospheric dynamics model or as a standalone tool. Sandia National Laboratories is a multimission laboratory managed and operated by National Technology & Engineering Solutions of Sandia, LLC, a wholly owned subsidiary of Honeywell International Inc., for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA0003525.

Bosler, Peter↗

Towards predicting removal rate and surface roughness during grinding of optical materials

A series of controlled grinding experiments, utilizing loose or fixed abrasives of either alumina or diamond at various particle sizes, were performed on a wide range of optical workpiece materials [single crystals of Al 2 O 3 (sapphire), SiC, Y 3 Al 5 O 12 (YAG), CaF 2 , and LiB 3 O 5 (LBO); a SiO 2 -Al 2 O 3 -P 2 O 5 -Li 2 O glass ceramic (Zerodur); and glasses of SiO 2 :TiO 2 (ULE), SiO 2 (fused silica), and P 2 O 5 -Al 2 O 3 -K 2 O-BaO (Phosphate)]. The material removal rate, surface roughness, and the morphology of surface fractures were measured. Separately, Vickers indentation was performed on the workpieces, and the depths of various crack types as a function of applied load was measured. Single pass grinding experiments showed distinct differences in the spatial pattern of surface fracturing between the loose alumina abrasive (isolated indent-type lateral cracking) and the loose or fixed diamond abrasive (scratch-type elongated lateral cracking). Each of the grinding methods had a removal rate and roughness that scaled with the lateral crack slope, sℓ (i.e. the rate of increase in lateral crack depth with the applied load) of the workpiece material. A grinding model (based on the volumetric removal of lateral cracks accounting for neighboring lateral crack removal efficiency and the fraction of abrasive particles leading to fracture initiation) and a roughness model (based on the depth of lateral cracks or the interface gap between the workpiece and lap) are shown to quantitatively describe the material removal rate and roughness as a function of workpiece material, abrasive size, applied pressure and relative velocity. This broad, multi-process variable grinding model can serve as a predictive tool for estimating grinding rates and surface roughness for various grinding processes on different workpiece materials

36 MATERIALS SCIENCE↗

PRIMA-X - Performance Retargeting of Instrumentation, Measurement, and Analysis Technologies for Exascale Computing

The PRIMA-X (Performance Retargeting of Instrumentation, Measurement, and Analysis Technologies for Exascale Computing) project is the successor of the DOE PRIMA (Performance Refactoring of Instrumentation, Measurement, and Analysis Technologies for Petascale Computing) project, which addressed the challenge of creating a core measurement infrastructure that would serve as a common platform for both integrating leading parallel performance systems (notably TAU and Scalasca) and developing next-generation scalable performance tools. The PRIMA-X project shifts the focus away from refactorization of robust performance tools towards a re-targeting of the parallel performance measurement and analysis architecture for extreme scales. The massive concurrency, asynchronous execution dynamics, hardware heterogeneity, and multi-objective prerequisites (performance, power, resilience) that identify exascale systems introduce fundamental constraints on the ability to carry forward existing performance methodologies. In particular, there must be a deemphasis of per-thread observation techniques to significantly reduce the otherwise unsustainable flood of redundant performance data. Instead, it will be necessary to assimilate multi-level resource observations into macroscopic performance views, from which resilient performance metrics can be attributed to the computational features of the application. This requires a scalable framework for node-level and system-wide monitoring and runtime analyses of dynamic performance information. Also, the interest in optimizing parallelism parameters with respect to performance and energy drives the integration of tool capabilities in the exascale environment further. Initially, PRIMA-X was a collaborative project between the University of Oregon (lead institution) and the German Research School for Simulation Sciences (GRS). Because Prof. Wolf, the PI at GRS, accepted a position as full professor at Technische Universitaet Darmstadt (TU Darmstadt) starting February 1st, 2015, the work of GRS was continued at TU Darmstadt. This report covers the work at TU Darmstadt after the transition. The first main accomplishment of TU Darmstadt is the development of different techniques to aggregate performance data on the level of threads. TU Darmstadt evaluated different schemes designed at GRS during the first phase of this project and integrated them into Score-P, a widely used performance measurement framework for HPC applications. The second main accomplishment is a substantial increase of Score-P’s scalability, achieved by improving its internal representation of the underlying system. Third, we extended our performance-modeling tool Extra-P to automatically create models with more than one input parameter (e.g., process count and input size), which was previously impossible. These models allow the performance behavior of an application to be extrapolated to larger machines. Specific applications of the new multi-parameter capability realized in this project include modeling the isoefficiency function of task-based applications and a novel co-design methodology. Further extensions of Extra-P comprise online modeling functionality, automatic refinement of the model search space, and segmented performance modeling. Further contributions include new locality metrics for both shared- and distributed-memory applications, a new way of measuring the impact of inter-application inference on the execution time of an application, and finally an automatic method that selects appropriate OpenMP constructs for the parallelization of sequential code and classifies the affected variables based on their sharing semantics.

97 MATHEMATICS AND COMPUTING↗

HPC4Mfg with Samsung: Making semiconductor devices cool through HPC ab initio simulations

For decades, the semiconductor technology has followed the Moore’s law, making newer devices more powerful and energy efficient. Recently, however, it has reached a point where the performance and the energy efficiency of the device do not improve with the shrinking device size. One of the fundamental reasons of this deviation from the past trend is the interconnect resistance, which becomes larger with the shrinking size. The devices size is so small that the quantum mechanical effects can no longer be ignored and the traditional continuum simulation tools such as TCAD become inadequate. In this project, Samsung Semiconductor Inc. and Lawrence Berkeley National Laboratory has collaborated to perform first of kind device-scale ab initio simulations to optimize materials and interconnect morphology to minimize interconnect resistance. We have tested the use of LS3DF method and the PEtot_trans approach on top of the folded spectrum method (FSM) Escan code to calculate the scattering state, and to study various effects influence the interconnect conductivity. We found that, the LS3DF can be used to calculate such metallic system. On the other hand, the use of Escan code to solve the linear equation is not practical due to the slow convergence. We have implemented a Chebyshev filter technique to calculate a few hundred eigen states near the scattering state energy E, then use these eigen states as preconditioner to solve the linear equation. We have used this approach to study the different factors which affect the interconnect conductivity, including the shape, the point defect, the temperature, and the grain boundary.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Automated Cloud Based Long Short-Term Memory Neural Network Based SWE Prediction

Snow derived water is a critical component of the US water supply. Measurements of the Snow Water Equivalent (SWE) and associated predictions of peak SWE and snowmelt onset are essential inputs for water management efforts. This paper aims to develop an integrated framework for real-time data ingestion, estimation, prediction and visualization of SWE based on daily snow datasets. In particular, we develop a data-driven approach for estimating and predicting SWE dynamics using the Long Short-Term Memory neural network (LSTM) method. Our approach uses historical datasets (precipitation, air temperature, SWE, and snow thickness) collected at NRCS Snow Telemetry (SNOTEL) stations to train the LSTM network and current year data to predict SWE behavior. The performance of our prediction was compared for different prediction dates and prediction training datasets. Our results suggest that the proposed LSTM network can be an efficient tool for forecasting the SWE timeseries, as well as Peak SWE and snowmelt timing. Results showed that the window size impacts the model performance (where the Nash Sutcliffe efficiency (NSE) ranged from 0.96 to 0.85 and the Rooted Mean Square Error (RMSE) ranged from 0.038 to 0.07) with an optimum number that should be calibrated for different stations and climate conditions. In addition, by implementing the LSTM prediction capability in a cloud based site-monitoring platform, we automate model-data integration. By making the data accessible through a graphical web interface and an underlying API which exposes both training and prediction capabilities. The associated results can be made easily accessible to a broad range of stakeholders.

54 ENVIRONMENTAL SCIENCES↗