Scientific machine learning for atmospheric data assimilation and large-scale hydrological forecasting at Penn State
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
3
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-13) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.
Research profile
Output—
Not yet scored: the underlying output record is incomplete, so no volume or consistency score can be computed. This axis would measure cumulative and per-year publication counts and venue mix; only fragmentary signals are visible so far.
Impact—
Not yet scored: citation data are insufficient for an influence score. Impact would measure citations, citing works, and uptake beyond the home field; h-index 3 and i10-index 0 reflect an early-stage record, not a computed score on this axis.
Focus0/100
Radar focus scores 0, reflecting the even 0.5/0.5 split between scientific machine learning / atmospheric data assimilation and hydrological forecasting. Topical distribution is nonetheless concentrated: Meteorological Phenomena and Simulations (13 works), Model Reduction and Neural Networks (12), and Climate Variability and Models (11) dominate, with minor clusters in Oceanographic and Atmospheric Processes (2) and Hydrology and Watershed Management (1) — one coherent weather/climate-plus-ML agenda pursued through two application tracks.
Collaboration53/100
Collaboration score 53 reflects a team-science pattern: 13 distinct coauthors in the recent window, with recurring partners including Romit Maulik, Dibyajyoti Chakraborty, Rao Kotamarthi, Hong Zhang, Sen Wang, Yang Xu, Jeremy Feinstein, John K. Hutchison, Vijay P. Ramalingam, and Jason Stock. In both cataloged papers Guan is a middle author and not corresponding, indicating contribution within larger ML/Earth-system groups rather than solo-led projects.
Activity79/100
Activity score 79 is the strongest axis in the radar, grounded in an accelerating publication cadence: counts by year rise from 1 work in 2024 to 5 in 2025 and 8 in 2026, giving 14 works over the recent three-year window. This steep ramp is consistent with an early-career researcher rapidly expanding their footprint in scientific machine learning for Earth-system applications.
Reproducibility55/100
Reproducibility score 55 (mean over 2 assessed papers). Both cataloged works are rated medium readiness: the atmospheric super-resolution paper's evidence is entirely ERA5-referenced with no classical data-assimilation baseline for comparison, while Hapi's headline flood-forecasting wins rely on perfect reanalysis forcing. Rigorous statistics are noted in both, but independent operational or classical-baseline validation remains open.
Research summary
Haiwen Guan is an early-career researcher at Pennsylvania State University working at the intersection of scientific machine learning and Earth-system prediction, with a profile split evenly between scientific machine learning / atmospheric data assimilation and hydrological forecasting (field distribution 0.5/0.5). Two 2026 arXiv preprints anchor the Evolve catalog. The first, on generative atmospheric super-resolution, introduces composable likelihood interfaces that fuse heterogeneous in situ observations — radiosonde, aircraft, and surface data — with a fixed atmospheric diffusion prior via diffusion posterior sampling; Evolve rates the gains over radar-only conditioning as solid but notes all evidence is ERA5-referenced and no classical data-assimilation baseline is included. The second, Hapi, is a multivariable land-surface Swin Transformer for medium-range continental flood forecasting, with rigorous statistics and state-of-the-art claims that depend on perfect reanalysis forcing. Bibliometric coverage from OpenAlex remains thin: h-index 3, i10-index 0, with 14 works in the last three years ramping from 1 (2024) to 5 (2025) to 8 (2026). Collaboration is team-based (13 distinct coauthors; score 53), with recurring partners including Romit Maulik, Dibyajyoti Chakraborty, Rao Kotamarthi, Hong Zhang, Sen Wang, and Yang Xu; both cataloged papers list Guan as a middle author rather than lead or corresponding. Reproducibility readiness for both papers is rated medium (mean 55).
Collaboration pattern
Haiwen Guan works in a team-science setting: 13 distinct coauthors in the recent three-year window (collaboration score 53), with recurring partners spanning machine learning and Earth-system expertise — Romit Maulik, Dibyajyoti Chakraborty, Sen Wang, Rao Kotamarthi, Hong Zhang, John K. Hutchison, Jeremy Feinstein, Yang Xu, Vijay P. Ramalingam, and Jason Stock. In both Evolve-cataloged papers Guan is a middle author and not the corresponding author, consistent with contributing specialist roles within larger research groups.
Research keywords
Highlights
Composable likelihood interfaces for generative atmospheric super-resolution
Hapi: continental-scale multivariable flood forecasting with a land-surface Transformer
Timeline
- 2024First indexed output appears: 1 work recorded in the OpenAlex counts-by-year series.
- 2025Output rises to 5 indexed works in the OpenAlex counts-by-year series.
- 2026Most active year to date: 8 indexed works, including two arXiv preprints reviewed by Evolve — generative atmospheric super-resolution from heterogeneous in situ observations, and the Hapi land-surface Transformer for continental-scale hydrological forecasting.
Papers in the Evolve catalog
Frequent co-authors
About this profile
This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.
- Bibliometric indicators (h-index, i10-index, counts by year, coauthor counts) are derived from OpenAlex only; the output and impact axes are unscored due to insufficient data.
- Only two papers (both 2026 arXiv preprints) have been reviewed by Evolve; the researcher's broader portfolio is not yet assessed.
- Identity is resolved from public bibliographic data (OpenAlex profile and ORCID 0009-0008-3005-1497); name-based ambiguity cannot be fully excluded.
- No citation or venue acceptance data beyond OpenAlex coverage is available; preprint status means results may not yet be peer-reviewed.