Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “High frequency analysis”

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 217 records · Page 12
All sourcesNASA NTRSDOE OSTI
DOE OSTI2026

Impedance and wakefield studies of the EIC RCS 591 MHz five-cell cavity

The Electron-Ion Collider (EIC) is a next-generation accelerator complex designed to enable high-luminosity collisions between highly polarized electrons and light ions (e.g., He-3). A central component of its Electron Injection System (EIS) is the Rapid Cycling Synchrotron (RCS), which accelerates a single 28 nC electron bunch from 750 MeV to 5, 10, or 18 GeV using an array of 591 MHz five-cell superconducting RF (SRF) cavities—eight at the current design stage. To ensure stable acceleration of high-charge bunches, we conducted detailed impedance and wakefield studies of the SRF cavity structure using both frequency- and time-domain methods. Wakefield solvers (ECHO3D, ECHO1D, CST), eigenmode analysis, and multi-particle tracking with ELEGANT were employed to evaluate longitudinal and transverse impedance effects and to determine instability thresholds. These studies provide critical input for the cavity design and operating parameters required to preserve beam quality and stability in the RCS.

Accelerator Physics↗
DOE OSTI2025

Statistical Analysis of Trans‐Ionospheric Pulse Pairs and Inferences on Their Characteristics

Trans-ionospheric pulse pairs (TIPPs), first observed in 1993, are signatures of in-cloud lightning discharges observed by satellite-based broadband very high frequency (VHF) receivers. It has been definitively shown that TIPPs are the space-based signatures of compact intracloud discharges (CIDs), and that the associated pair of pulses that comprise a TIPP result from the direct VHF pulse from the discharge, followed by a pulse reflected from the Earth's surface. However, the ratio of the peak amplitudes of these two pulses can vary widely, with the second pulse often having considerably higher peak amplitude than the first. This observation has not been satisfactorily explained. Using data collected from geostationary orbit by the Radio Frequency Sensor (RFS) and matched to locations reported by the Global Lightning Dataset (GLD360), we assemble the largest database to date of 76,348 TIPPs with associated location, altitude, and amplitude ratio of the two pulses in the TIPP. We show that the amplitude ratio of TIPPs is strongly correlated to the altitude of the associated discharges and the geometry of the source location with respect to the Earth's surface and the receiver. These observations strongly suggest that the difference in amplitude of the two pulses is driven by a nondipole radiated beam pattern that is dependent on the polarity of the CID, velocity of the current wavefront, and viewing angle.

58 GEOSCIENCES↗
DOE OSTI2025

Bi-stability and period-doubling cascade of frequency combs in exceptional-point lasers

Abstract Recent studies have demonstrated that a laser can self-generate frequency combs when tuned near an exceptional point (EP), where two cavity modes coalesce. These EP combs induce periodic modulation of the population inversion in the gain medium, and their repetition rate is independent of the laser cavity’s free spectral range. In this work, we perform a stability analysis that reveals two notable properties of EP combs, bi-stability and a period-doubling cascade. The period-doubling cascade enables halving of the repetition rate while maintaining the comb’s total bandwidth, presenting opportunities for the design of highly compact frequency comb generators.

exceptional points↗
DOE OSTI2026

Wildfire-Power Grid Interactions: Feedback, Impacts, Monitoring, Modeling, and Mitigation Strategies

Wildfires are increasingly interacting with electric power systems through a two-way hazard chain: fires damage grid assets and trigger cascading outages, while grid faults can ignite new fires under hot, dry, and windy conditions. This review synthesizes the state of knowledge across five domains: (i) physical impacts of flames, heat, and smoke on lines, towers, insulators, and substations; (ii) power-infrastructure-initiated ignitions via conductor clash, high-impedance faults, and corona discharge; (iii) widespread blackouts and disproportionate societal impacts; (iv) multi-scale monitoring spanning laboratory tests, in-situ and grid-integrated sensors, and Earth observation; (v) coupled modeling that links fire behavior with grid operations; and (vi) technological and strategic mitigation pathways spanning prevention, response, and recovery. We integrate these domains into a novel 'feedback-aware' socio-technical framework. Through a longitudinal analysis (2005-2025) of global incidents, we identify that while vegetation contact remains the most frequent ignition source, aging infrastructure failure has emerged as a critical driver of catastrophic 'mega-fires'. We further identify persistent gaps, including limited interoperability of high-frequency grid and environmental data, scarce real-time data assimilation, and under-developed equity metrics for outage management. We conclude by outlining a research agenda to (1) deploy interoperable sensing architectures, (2) advance feedback-coupled fire-grid simulations, and (3) evaluate mitigation portfolios through techno-economic and fairness lenses. Recognizing wildfire-grid interactions as coupled socio-technical systems is essential for protecting infrastructure and communities and for ensuring reliable, sustainable electricity in a changing world.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2023

Design and Optimization of High-frequency Transformer for Isolated Single-Stage Three-Phase AC/DC Converter using Bidirectional Switches

This paper presents the design and optimization of the high-frequency transformer for an isolated single-stage threephase AC/DC converter enabled by the 4H-SiC BiDirectional Field Effect Transistor (BiDFET). Non-dominated Sorting Genetic Algorithm II (NSGA-II) is used to perform a multiobjective optimization to minimize loss, weight, and volume. A 20 kW, 50 kHz hardware prototype based on the selected optimal point is characterized by a power amplifier. The finite element analysis simulation results are also presented to validate the core magnetic flux density and winding electric field strength. The experimental prototype was tested at 800 Vdc/480 Vrms at 10 kW rated power.

BiDFET, High-frequency transformer, Multiobjective↗
DOE OSTI2021

The Pre-He White Dwarfs in Eclipsing Binaries. II. WASP 0843–11

We present the Transiting Exoplanet Survey Satellite (TESS) photometry and our high-resolution spectra of the EL CVn–type star 1SWASP J084356.46–113327.5 (WASP 0843–11), which consists of an F-type main-sequence (MS) star and a hotter low-mass helium white dwarf precursor (pre-He WD). In a spectral analysis, double-lined radial velocities (RVs) were obtained for the first time. From a simultaneous analysis of the TESS light and RV curves, the masses, radii, and temperatures of each component were determined to be M {sub 1} = 1.733 ± 0.031 M {sub ⊙}, M {sub 2} = 0.220 ± 0.008 M {sub ⊙}, R {sub 1} = 2.094 ± 0.013 R {sub ⊙}, R {sub 2} = 0.331 ± 0.003 R {sub ⊙}, T {sub 1} = 6960 ± 120 K, and T {sub 2} =9870 ± 310 K, respectively. According to their physical properties, the primary and secondary stars are located on the instability strips of δ Sct/γ Dor hybrids and extremely low-mass WDs, respectively. We applied multiple frequency analyses to the residual light curve after subtracting the binarity effects and detected a total of four independent frequencies. A frequency of f {sub 3} = 1.6356 day{sup −1} in the low-frequency region was reminiscent of γ Dor stars, while the other three frequencies of f {sub 2}, f {sub 15}, and f {sub 18} in the high-frequency region (150–196 day{sup −1}) are probably related to pre-He WD components. Our results demonstrate that the more massive MS primary star of WASP 0843–11 is a γ Dor pulsator, whereas the hotter secondary component is a pre-He WD with high-frequency pulsations.

47 OTHER INSTRUMENTATION↗
DOE OSTI2021

Quantifying Dam-Induced Fluctuations in Streamflow Frequencies Across the Colorado River Basin

Periodic fluctuations in natural streamflow are a major driver of river ecosystem dynamics and water resource management. However, most U.S. rivers are impacted both by long-term hydroclimatic trends and dams that alter flow variability. Despite these impacts, it remains largely unexplored how dams affect the dominant frequencies of natural streamflow over a highly regulated river network. We investigated the entire Colorado River Basin (CRB) to understand how the annual (10–14 months) and multi-annual (24–60 months) frequencies in natural flow regimes have been progressively altered by dams. Given the significant alteration over the CRB, we captured changes in streamflow frequencies between naturalized and observed monthly flows via wavelet analysis. Based on the similarity of changes in streamflow frequencies (annual and multi-annual) over the last 30 years, sections of the riverine network were classified into four groups. The annual frequency was relatively well preserved downstream of Hoover Dam, while showing a systematic trend of alteration downstream of Glen Canyon Dam until Hoover Dam. Meanwhile, the multi-annual frequency component was highly altered for the entire Lower Colorado main stem (i.e., downstream of Glen Canyon). We also identified dams with significant impacts on streamflow frequency by comparing wavelet coherence estimates. This study advances the notion that dams fundamentally alter river flow regimes across multiple frequencies and with varying amplitudes over time and space, with alteration propagating – or being dampened – by both hydroclimatic fluctuations and water resource management.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2026

Automated Classification of Vehicle Movements at Signalized Intersections Using Vehicle Trajectories

Accurate vehicle movement classification through signalized intersections is of paramount importance to the analysis of intersection performance and the optimization of traffic control strategies. Conventional techniques for tracking vehicle turning movements depend on infrastructure-based strategies like human counts, loop detectors, and video analytics, all of which are costly, prone to errors, and spatially constrained. High-frequency trajectory data can be utilized to determine vehicle movement patterns in a scalable and infrastructure-independent method due to the adoption of connected vehicles (CVs). In recent years, several studies have utilized connected vehicle data to generate performance measures. Most of the trajectory-based performance measures approaches, however, require map matching-i.e., extracting geospatial references from maps to identify the movements that individual vehicles make at a signalized intersection. These approaches are often time-consuming and hinder scalability since geographic features need to be provided for an analysis to be conducted. Map matching methods are prone to errors as different map versions change these geographic features. This research presents a novel automatic classification pipeline that uses CV trajectory data to classify vehicle movements at signalized crossings, specifically pass-through left-turn and right-turn maneuvers. The process starts by filtering trips that cross a spatial bounding box that has been defined at the target intersection. Approach and departure headings for each trajectory crossing the boundary are computed and are clustered together to identify dominant movements. The proposed algorithm is used to classify the movement of vehicles at 10 intersections in the state of California, and the results indicate that the algorithm can classify movements at these intersections with varying traffic volumes and road network configurations, all in a map-less framework with no need for conflation of vehicle trajectories to a digital base map.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2020

Soft Switching Over the Entire Line Cycle for a Quadruple Active Bridge DCX in a DC to Three-Phase AC Module

This paper is focused on a transformer-isolated quadruple active bridge (QAB) dc-dc converter loaded by three full-bridge dc-ac inverters. The QAB functions as a "DC trans-former" (DCX) in the dc-to-three-phase ac module. The QAB outputs provide time-varying power at twice the line frequency, which presents challenges in maintaining zero voltage switching (ZVS) on the secondary sides during low-power portions of the line cycle. It is shown how ZVS can be achieved in a phase, even at zero-power transfer, using a relatively small circulating current provided by the magnetizing inductance of the high-frequency transformer. The approach is particularly effective in high-voltage applications using SiC MOSFETs, where reductions in switching loss outweigh additional conduction losses due to the circulating currents. A detailed analysis of ZVS switching waveforms at the zero power transfer is presented, including effects of nonlinear device capacitances. Analytical expressions are given for the optimal values of the magnetizing inductance and dead times of the QAB primary and secondary bridges. The approach is verified by experimental results on a 600 V, 4 kW prototype, demonstrating greater than 98.5% efficiency from 1 kW to 4 kW, with a peak efficiency of 99.0%. Compared to conventionally operated prototype, a 54% reduction in total loss is achieved at rated power.

41 EE - Solar Energy Technologies Office (EE-4S)↗
DOE OSTI2021

VLBI Data Processing on Coronal Radio-sounding Experiments of Mars Express

The ESA’s Mars Express solar corona experiments were performed at two solar conjunctions in the years 2015 and 2017 by a number of radio telescopes in the European VLBI Network. This paper presents the methods to measure the frequency and phase fluctuations of the spacecraft radio signal, and the applications to study the characteristics of the plasma turbulence effects on the signal at a single station and at multiple stations via cross correlation. The power spectra of the frequency fluctuations observed between 4.9 and 76.3 R {sub s} have a power-law shape close to a Kolmogorov spectrum over the frequency interval ν {sub lo} < ν < ν {sub up}, where the nominal value of ν {sub lo} is set to 3 mHz and ν {sub up} is in the range of 0.03–0.15 Hz. The rms of the frequency fluctuations is presented as a function of the heliocentric distance. Furthermore, we analyze the variations of the electron column density fluctuations at solar offsets 4.9 R {sub s} and 9.9 R {sub s} and the cross-correlation products between the VLBI stations. The power density of the differential fluctuations between different stations decreases at ν < 0.01 Hz. Finally, the fast flow speeds of solar wind >700 km s{sup -1} are derived from the cross correlation of frequency fluctuations at ν < 0.01 Hz. The fast flow speeds of solar wind correspond to the high heliolatitude of the coronal region that the radio rays passed. The VLBI observations and analysis methods can be used to study the electron column density fluctuations and the turbulence at multiple spatial points in the inner solar wind by providing multiple lines of sight between the Earth and the spacecraft.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2021

Physics-Based Feature Extraction from Bulk Time-Series PMU Datasets for Event Detection

In this work, two physics-based feature extraction techniques are developed for bulk time-series phasor measurement unit (PMU) datasets collected from the field to train the machine learning model for anomaly detection. Two approaches have been developed to extract useful features for different types of events. An admittance-based feature extraction technique is developed to detect events that involve line outages and system topology variations. The developed algorithm extracts the system equivalent admittance variation. Additionally, Fielder’s Theory is utilized to further reduce the potential computation burden by sectionalizing large-scale grids and datasets into smaller areas. Second, an oscillation-based feature extraction technique is developed to detect low-frequency oscillations in power grids. The dominant oscillation modes in the grids are extracted using energy-sorted Prony analysis. The extracted dominant oscillation modes by the developed work exhibit a high fitting resolution. Finally, the developed techniques have been validated using large-scale and real-world datasets.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2021

Diffraction of radio-frequency waves by plasma turbulence in the edge of a tokamak

The Mie-Lorenz-Debye theory of scattering of radio frequency waves (RF) by filamentary structures in the edge of a tokamak plasma has been established in prior publications. These results have been verified. Theory and computations show that RF can be reflected, refracted, and diffracted by turbulence. Furthermore, the filamentary structures lead to side scattering of waves and excitation of wave modes different from the launched RF. It is important to quantify the overall effect of turbulence on the propagation of RF since efficiency of heating and current drive by RF waves depends upon the wave characteristics. The commonly used RF ranges vary from the low frequency ion-cyclotron (IC) waves, to medium frequency lower-hybrid (LH) waves, to high-frequency electron cyclotron (EC) waves with wavelengths respectively, large, comparable, or smaller than the radial dimensions of the filaments. Currently, we discuss the scattering as well as the forces exerted by different RF in the edge plasma. Also, we present a code for 3D full wave scattering analysis of arbitrary tensor permittivity and plasma density profiles. The study is applicable to ITER-like plasmas, as well as to plasmas in medium sized tokamaks such as TCV, ASDEX-U and DIII-D and high field concepts like SPARC.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2021

Subcritical Californium Source Drive Noise Analysis Measurements With Unreflected Uranium (93.15) Hydride

On July 6, 1989, subcritical californium source-driven noise analysis (CSDNA) measurements were performed with bare uranium hydride cylindrical assemblies at the Los Alamos National Laboratory Critical Experiments Facility in KIVA 1. Three configurations of ~7.5 cm radius-enriched (93.15 wt. % 235 U) uranium hydride with a density of ~10 g/cm 3 cylinders were assembled with uranium hydride heights of ~11, ~14, and ~16 cm. The uranium hydride (~2 and ~3 cm high) was in thin-welded stainless steel cans. The neutron multiplication factors obtained on-line from the measured ratios of spectral densities were 0.942 ± 0.002, 0.917 ± 0.003, and 0.867 ± 0.004 for uranium hydride heights of ~16, ~14, and ~11 cm., respectively. Neutron multiplication factors for these three configurations calculated by Los Alamos National Laboratory Monte Carlo methods were 0.952 ± 0.005, 0.922 ± 0.005, and 0.860 ± 0.005, all of which are in good agreement with the measurements. The break frequency noise analysis data could be fitted to obtain the prompt neutron decay constant at all subcritical states. Some data presented in this report are from notes that are not in the logbooks. A discussion of the two different point kinetics theories of the measurements that illustrated the deficiencies of the rigorous theory at low neutron multiplication factors (k eff < 0.80) is given in an Appendix. These measurements may be the only configuration of unreflected, highly enriched uranium hydride configuration assembled. The presence of hydrogen diluted the uranium density, decreasing the reactivity, but this is offset by slowing down the neutron energy to the energy ranges of higher-fission cross sections. Extrapolations to delayed critical indicate that adding one more 3 cm can and one more 2 cm can would result in the unreflected cylindrical system being slightly above critical with a critical mass of ~37 kg, which is considerably less than the critical mass of an unreflected and unmoderated highly enriched uranium metal sphere. An unreflected, highly enriched uranium hydride sphere would have a critical mass less than 37 kg.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2022

MAGNETICALLY SUSPENDED CANNED ROTOR PUMPS FOR THE INTEGRAL MOLTEN SALT REACTOR

This project developed a molten salt pump design for small modular reactors, concentrated solar, and Gen IV nuclear reactors with a magnetically levitated rotor that can operate at temperatures of up to 700 °C. This eliminates the need for rotating seals and roller element bearings which require maintenance and are prone to failure. The pump design is also more compact than existing molten salt pump designs which require long shafts to thermally isolate the electric motor. The aims of the project are to: 1) design and test the molten salt pump, 2) develop and test a high temperature position sensor needed for magnetic bearing control, and 3) develop and test high temperature coil fabrication methods for electric motor and magnetic bearing fabrication. To develop the high temperature coils, conductor materials were analyzed for oxidation, diffusion, cladding, and resistivity. Of the candidate conductor materials that met the requirements, silver was chosen. The long-term impact of radiation on silver resistivity was analyzed by developing a model of a generic molten salt fast reactor to model the conductor irradiation in worst case scenarios. The analysis showed that increased resistivity due to transmutation of silver decreases exponentially with the reactor reflector thickness. Different electrical insulation materials were tested and the final high temperature coils developed were able to withstand conductor to ground voltages up to 1800 V before insulation breakdown. Magnetic bearings require high-precision high-speed measurements of the rotor position to operate. Current commercial position sensors can only operate up to 550 °C so a novel position sensor was developed that can operate up to 800 °C. These sensors were also designed to measure the position of the metallic rotor through a thin metallic corrosion barrier that protects that coils, insulation, and magnetic cores so that the salt containment barrier does not have any penetrations. The high-temperature sensor was tested and had comparable performance to commercially available inductive position sensors. The sensor design and signal processing has led to several invention disclosures. The magnetically levitated molten salt pump design was completed and analyzed. The chloride salt test loop at Oak Ridge National Laboratory was chosen to test the pump and the pump hydraulics and power were designed for the test loop flow and pressure requirements. Classes of motor designs that do not utilize permanent magnetic were considered for the high-temperature motor. After extensive analysis of their performance, a synchronous reluctance motor was chosen for the pump. The design was refined and analyzed to optimize performance. The magnetic bearings were designed to meet the force and frequency requirements necessary to levitate the rotor and reject disturbances from rotor imbalance, motor forces, and hydraulic forces. Compliant mounts for the position sensors were designed that would not damage the ceramic sensor components and maintain sensor alignment over a wide temperature range. A thermal hydraulic analysis of the molten salt coolant flow used to cool the rotor and stator was performed to ensure that the maximum temperature in the rotor and stator will not exceed 725 °C. The corrosion barrier and casing were designed to maintain the salt fluid boundary by using compliant features to absorb large expansion mismatches due to different coefficients of thermal expansion (CTE). Finally, the salt drain and fill systems were designed along with the pump support structure, integration with the test loop, and thermal insulation and heating. Corrosion barrier fabrication experiments were performed to assess using laser cladding to apply a sub-millimeter corrosion barrier to the surface of a Fe-Co-V magnetic core material. The cladding method showed minimal mixing between materials at their interface and withstood temperature cycling without damage to the cladding. The technology commercialization is focused on the high-temperature position sensors and there has been significant interest from the nuclear industry in utilizing these sensors along with some interest from the aerospace industry. The project also led to a research collaboration with a fusion energy industrial partner to study barriers to scaling the pump design from the kW size to the MW size. This research collaboration will also study methods to create high-temperature coils that have conductor to ground breakdown voltages above 5 kV. The successful development of high-temperature electromagnetic coils and high-temperature position sensors greatly increases the temperature limits for electromagnetic devices that don’t rely on permanent magnets. This includes electric motors, magnetic bearings, linear actuators, rotary position and velocity sensing, actuated valves, and generators to name a few. This will fill a current need in molten salt reactors, concentrated solar, and fusion energy for these critical peripheral devices needed for practical reactor designs that do not currently have commercial solutions.

Hines, J Wesley↗
DOE OSTI2025

Directional Variations in Tidal Flow Multifractality and Intermittency

We explored distinct directional variations of multifractal and intermittent characteristics of ebb and flood flow velocities at nodule point, WA, tidal energy site, and complementary inspection on the East River, NY using scaling exponents of the structure function, distribution flatness, detrending moving average (DMA) analysis, multifractal detrended fluctuation analysis (MF-DFA), and high-order spectral moments. Our findings reveal that tidal flow presents higher long-range dependence (LRD) and intermittent levels for the ebb flow sections, whereas the flood flow sections exhibit a higher degree of multifractality and greater sensitivity to larger magnitude of turbulent fluctuations. We demonstrate that long-range dependence predominantly contributes to multifractal behavior in both ebb and flood flows, as evidenced by the significantly reduced multifractal spectrum width for temporally randomly permuted time series. Moreover, spectral kurtosis analysis uncovers a higher intermittent level across all frequency scales for ebb flow sections and reveals a distinct pattern of tidal flow intermittency differing from the monotonically increasing intermittent level observed in wall-bounded and grid turbulence. Finally, we demonstrate that DMA, MF-DFA, and high-order spectral moments provide more comprehensive insights than structure function scaling exponents and PDF flatness methods.

Cheng, Shyuan [Univ. of Illinois at Urbana-Champai↗
DOE OSTI2026

Transverse BBU Suppression with Feedback for Energy-Recovery Linacs

The Electron–Ion Collider (EIC) achieves its design luminosity by cooling the ion beam with a high-current electron beam generated in an energy-recovery linac (ERL). The baseline ERL lattice employs a BNL five-cell cavity that is frequency-scaled to 197 MHz, 591 MHz, and 1.773 GHz, raising concerns about multibunch beam-breakup (BBU) instabilities. Threshold currents for each frequency option are established with two independent BBU tracking codes, providing cross-validated operating margins that guide cavity selection. To further increase the current limit, we incorporate a digital transverse feedback (FB) system that targets the dominant higher-order modes (HOMs). Simulations show the FB raises the BBU threshold by roughly an order of magnitude: power-spectral-density analysis of the beam centroid at the linac exit confirms strong suppression of the aliased HOM peak, although some spectral growth appears at secondary frequencies. This combined study quantifies baseline BBU limits, demonstrates effective active mitigation, and charts a practical path toward robust, high-current operation of the EIC ERL.

Accelerator Physics↗
DOE OSTI2026

A hybrid Penman-Monteith and machine learning model for simulating evapotranspiration and its components

Integrating physical processes with machine learning has advanced evapotranspiration (ET) simulation, yet most hybrid models fail to partition total ET into its components: soil evaporation (E) and vegetation transpiration (T). This study introduces Residual Neural Network–Penman–Monteith (RNN-PM), a novel hybrid dual-source ET model designed to overcome this limitation. The model synergizes the physically-based Penman–Monteith framework with three specialized residual neural networks trained to estimate key conductance parameters (canopy conductance, soil surface conductance, and aerodynamic conductance). Furthermore this explicit parameterization allows for the direct partitioning of total ET. Validation at National Ecological Observatory Network (NEON) flux sites using high-frequency partitioned E and T shows that RNN-PM reliably reproduces ET and the transpiration fraction (T/ET). For ET, the model achieves an average Kling–Gupta efficiency (KGE) of 0.89 and a root-mean-square error (RMSE) of 0.55 mm/day; for T/ET, the KGE is 0.87 with an RMSE of 0.06. Furthermore, RNN-PM demonstrates robust generalization, accurately simulating ET and its components well beyond the initial training dataset, even under extreme climatic conditions. This study extended the analysis by comparing the RNN-PM model with seven established dual-source ET models. The results indicate that RNN-PM outperforms both conventional machine learning models and purely physical process-based models in simulating ET components in most cases. Among the purely physical process-based dual-source models, those based on surface temperature decomposition showed improved performance as the leaf area index (LAI) decreased when evaluated against high-frequency ET component datasets. In contrast, the performance of conductance-based dual-source models declined with decreasing LAI. Although purely machine learning-based models can produce relatively accurate simulations of ET components, they often exhibit limited generalization capability, an issue that the RNN-PM model effectively overcomes. Ultimately, the RNN-PM model represents a significant advance in simulating ET components, offering a novel and scalable approach for improving the representation of land–atmosphere interactions in Earth system models.

54 ENVIRONMENTAL SCIENCES↗