SYSongkun Yan

Songkun Yan

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Songkun Yan has a total of 57 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Hydrology's Three Clouds in the Age of Data and AI: Thinned, Thickened, and Crystallizing

    by Yang Hong, Songkun Yan

    Beven (1987) proposed two linked challenges for catchment hydrology: macroscale theory and a stochastic framework for predictive uncertainty. Nearing et al. (2021) revived these ideas as hydrology’s Kelvin-style “clouds” in the machine-learning era. We revisit that lineage considering large-sample hydrology, Earth observation, differentiable modeling, and agentic artificial intelligence (AI). We generalize the inherited problems as representability—whether hydrologically relevant behavior can be captured and shown to transfer across the scales and purposes required—and uncertainty/identifiability—whether evidence can discriminate among plausible representations and explanations. Representability has thinned as data-rich learning has revealed more transferable predictive regularities, although prediction alone does not establish hydrologic understanding; uncertainty/identifiability has thickened as model flexibility grows faster than discriminating evidence. A third cloud, verifiability, is beginning to crystallize as agentic AI changes the architecture of scientific inquiry. Its contours are not yet settled: the concern is how sufficiently independent challenge and guardrails can be preserved as increasingly autonomous systems generate claims, select evidence, estimate uncertainty, interpret results, and participate in checking their own outputs. Cloud III therefore is emerging rather than well defined. The three clouds can work together and frame a shift from predictive skill alone toward defensible hydrologic knowledge.

  • Open Access

    TORRENT: An observation-constrained and benchmark-ready community shared dataset in the contiguous United States for flash-flood research and model intercomparison

    by Songkun Yan, Mengye Chen, Jonathan J. Gourley, Jian Zhang, Günter Blöschl, Paul Bates, Zhi Li, Siyu Zhu, Abigail Snooks, Yixin Wen, et al.

    Flash floods develop within minutes to hours, occur mainly in small to medium-sized watersheds, and are documented through scattered sources such as news reports, social media, stream gauges, operational storm events databases, and post-flood field surveys. Such fragmentation historically limits flash flood research advances and model evaluation or intercomparisons. We present TORRENT (the Traceable, Observation-constrained, Rapid-Response, Episode–gauge–watershed Network of Testbeds), a benchmark-ready community dataset of flash-flood hydrological response in the contiguous United States (CONUS). Each testbed links one reported NOAA/NCEI flash-flood episode, one U.S. Geological Survey (USGS) stream gauge, and the gauge’s contributing watershed, combining the observed discharge hydrograph—the primary benchmark target—with watershed boundaries, temporally screened in-basin high-water marks, and quality-control metadata; full packages additionally provide Multi-Radar Multi-Sensor (MRMS) precipitation forcing and auxiliary gridded products. TORRENT follows a three-level structure distinguishing what was reported, what observations are available, and what can be evaluated reproducibly: a national inventory of 42,466 reported episodes for 1996–2025; observation-availability documentation for 5,424 recent CONUS episodes during 2021–2025, including episodes without usable external observations; and 806 episode–gauge–watershed evaluation packages representing 160 episodes and 208 gauges. Each Level-3 package has a retrievable watershed, at least one in-basin high-water mark meeting the stated temporal-screening criterion, and a detected discharge response. Flash-flood-response screening, deduplication, in-basin NCEI corroboration, station-relative severity screening, and post-selection curation yield a conservative primary benchmark of 21 testbeds representing 14 episodes and 21 gauges. This primary set is intended as a high-confidence reference for formal rainfall–runoff model intercomparison rather than as a statistically representative sample of CONUS flash floods; the broader Level-3 pool supports exploratory analysis and robustness testing. Primary testbeds exhibit rapid, pronounced responses, with estimated event-peak return periods ranging from about 2 to more than 100 years. The automated construction and screening stages are released as deterministic, re-runnable code, together with documented curation decisions, sensitivity tests, and quality-control diagnostics. TORRENT is openly available at https://doi.org/10.5281/zenodo.23359069 (Yan et al., 2026).

  • Open Access

    The evolution of hydrology (1900-) under changing scientific paradigms: From catchments to Earth-system synthesis

    by Yang Hong, Soroosh Sorooshian, Efi Foufoula-Georgiou, Steven M. Gorelick, Charles Vörösmarty, Songkun Yan, Siyu Zhu, Victor R. Baker, Yixin Wen, Mengye Chen

    Since 1900, hydrology has evolved from a predominantly empirical and applied discipline into an increasingly integrated Earth-system science. This Commentary interprets that evolution through eight cumulative hydrologic eras shaped by five broader scientific paradigms: empirical, theoretical, computational, data-intensive, and artificial intelligence. Twentieth-century developments established the foundation of catchment-based hydrology in quantitative observation, analytical and process explanation, numerical simulation, and uncertainty diagnostics. Tracer, remote sensing and geospatial approaches subsequently expanded hydrologic inquiry across time, space, and connected systems; artificial intelligence now adds new capacity for data-model integration and possible scientific discovery. Together, these capabilities point toward an emerging phase of Earth-system synthesis that connects local process realism with global hydrology and coupled natural-human dynamics. The challenge is not to integrate everything at every scale, but to combine complementary capabilities where scientific questions demand it. Future progress depends on whether such synthesis is observationally supported, physically constrained, uncertainty-aware, diagnostically transparent and transferable across changing environments, and actionable for advancing Earth science and benefiting society.

  • Open Access

    Remote Sensing Hydrology Revisited: Advances, Persistent Challenges, and a Science‐Driven Agenda for the Next Decade

    by Yang Hong, Dennis P. Lettenmaier, Efi Foufoula‐Georgiou, Ana P. Barros, George J. Huffman, James S. Famiglietti, Dawen Yang, Tamlin M. Pavelsky, Soroosh Sorooshian, Taikan Oki, et al.

    In the WRR 50th Anniversary Special Collection, Lettenmaier et al. (2015, https://doi.org/10.1002/2015WR017616 ) provided a timely account of how satellite remote sensing evolved from a niche capability in hydrology into a mainstream driver of hydrologic discovery, as the community shifted from imagery toward validated Earth Observation (EO) products and open‐access data systems. In this Commentary for the WRR 60th Anniversary Special Collection, we revisit and extend that synthesis and argue that the next step is not just more EO data, but a more accountable remote sensing hydrology. The challenge is to integrate remote sensing more directly into hydrologic science questions in ways that ground networks alone cannot provide consistently across scales. We center this Commentary on a core question: How can EO be used, credibly and consistently, to quantify where water is stored, how it moves, and how the hydrologic cycle is changing, in ways that improve hydrologic understanding and decision value beyond what ground networks alone can provide? We summarize key advances since 2015 that sharpen water‐cycle observability, expand analysis‐ready data systems and computation, and accelerate hybrid inference, while highlighting persistent challenges that continue to limit credible hydrologic inference near thresholds and extremes. We conclude with a science‐driven agenda focused on closing observability gaps, improving record coherence, advancing uncertainty‐aware inference, strengthening end‐to‐end traceability from measurements to societal impacts, and sustaining climate‐quality EO records for hydrologic science in a changing climate, while recognizing that these advances will depend on continuity of observing capability and credible follow‐on mission strategies.


  • Open Access

    AI Agent for Hydrologic Modeling: Definition, Development, and Application

    by Songkun Yan, Mengye Chen, Zhi Li, Yixin Wen, Siyu Zhu, Mofan Zhang, Di Liu, Jie Cao, Xiaodong Chen, Chengbin Deng, et al.

    Hydrologic modeling supports flood forecasting and water resources management, but complex preprocessing, parameterization, and configuration limit broader use. This study defines a six‐level framework for artificial intelligence (AI)‐agent autonomy in hydrologic modeling and develops a Level‐4 agent, powered by large language models, that translates natural‐language requests into data retrieval, model execution, diagnostics, and reports with human oversight. In a proof‐of‐concept application to the July 2025 Texas flash flood, the agent reproduced key flood dynamics in the tested basin and reduced manual workflow effort. Observation‐driven simulations aligned with streamflow records, whereas a forecast‐driven run missed the flood response because the deterministic rainfall forecast displaced the storm core. These results suggest the feasibility of AI‐agent‐assisted hydrologic modeling in the tested case, while robustness and generalization across broader basins and events remain to be established through systematic validation, probabilistic meteorological forcing, and expert review of automated outputs.


  • SWD-Net: A Benchmark Dataset for Small Water Body Detection From Very High-Resolution Imagery

    by Di Liu, Chengbin Deng, Pontus Olofsson, Songkun Yan, Jiao Wang, Isabelle Hu, Yang Hong

    Small and seasonal water bodies store a substantial portion of inland surface water and play a key role in local hydrology, ecosystem resilience, and agricultural sustainability. Yet their small size, irregular geometry, and frequent seasonal variability hinder accurate detection in global water products derived from coarse-resolution imagery. To address this limitation, we present Small Water Body Detection (SWD-Net), a high-resolution, expert-annotated benchmark dataset for small water-body detection from aerial and satellite imagery. SWD-Net consists of 10,085 image–mask pairs (1,024 × 1,024 pixels) sourced primarily from NAIP, Google Earth, and LoveDA, spanning diverse climatic and land-cover conditions. Baseline experiments using five representative segmentation architectures, including U-Net, Mamba U-Net, ViT U-Net, SegFormer U-Net, and Swin U-Net, demonstrate clear architectural contrasts, with transformer-based networks achieving the most accurate delineation of fine-scale and fragmented ponds. Among them, Swin U-Net achieved the best overall performance, with an IoU of 93.53% and an F1-score of 96.56% on the test set. All data, pretrained weights, and evaluation codes are publicly released to facilitate transparent and reproducible research in fine-scale hydrological mapping, and seasonal water monitoring.


  • Open Access

    FloodSimBench: A Benchmark Dataset for Training Foundational Flood Inundation Models

    by Zhi Li, Songkun Yan, Mofan Zhang, Siyu Zhu, Yixin Wen, Alexander Sun, Meihong Ma, Yang Hong

  • A Review of the Past Half Century of Geostationary Satellite Thermal Observations for Global Precipitation Estimation: Developments, achievements, and future prospects

    by Siyu Zhu, Guoqiang Tang, Soroosh Sorooshian, George J. Huffman, Qingyun Duan, Songkun Yan, Ali Behrangi, Simon Michael Papalexiou, Phu Nguyen, Kuolin Hsu, et al.

    Over the past half century since 1975, the development and improvement of meteorological geostationary (GEO) satellites have played a pivotal role in observing cloud dynamics and subsequently in advancing spaceborne precipitation estimation. Infrared (IR) observation from GEO satellites offers unique advantages, such as broad spatial coverage, high temporal resolution, and long-term consistency, motivating extensive research to unlock the potential of GEO IR-based data for capturing precipitation structure and dynamics. This article reviews the major developments and milestone achievements of GEO IR-based precipitation estimation over the past five decades and summarizes the future prospects, including potential directions and remaining challenges. By examining the history of global GEO satellites and more than 100 references in this domain, we categorize the development of GEO IR-based precipitation estimation methodology and technology into three distinct stages: 1) the Exploration Phase (1975 to ca. 1995), 2) the Growth Phase (ca. 1995 to ca. 2015), and 3) the Exploitation Phase (ca. 2015 to the present). As we transition into an emerging new stage, these efforts collectively point toward multispectral retrievals, lifecycle-aware machine learning (ML), smart sensing, and advanced multisource integration as key directions shaping the future of GEO IR-based precipitation estimation. In summary, GEO IR-based precipitation estimation has made substantial contributions over the past half century and will play an increasingly important role with great potential in future precipitation science.


  • Open Access

    AI Agent for Hydrologic Modeling: Definition, Development and Application

    by Songkun Yan, Mengye Chen, Zhi Li, Yixin Wen, Siyu Zhu, Mofan Zhang, Di Liu, Jie Cao, Xiaodong Chen, Chengbin Deng, et al.

    This study defines a six-level framework of AI-agent autonomy for hydrologic modeling and develops a Level-4 agent that autonomously configures simulations, analyzes results, and generates reports with human oversight. The agent integrates natural language queries with data retrieval, model execution, and post-simulation diagnostics. Applied to the July 2025 Texas flash flood, the agent reproduced flood dynamics within minutes, reducing manual operations by over 90%. While observation-driven simulations aligned with streamflow records, forecast-driven runs failed to capture the event due to spatial displacement of rainfall forecasts, underscoring limitations of current warn-on-forecast practices. This work demonstrates that AI agents can democratize hydrologic modeling by lowering operational barriers for education, research, and decision support, while highlighting the need for improved meteorological inputs for reliable flood forecasting.

  • Open Access

    A New Approach to Identifying and Analyzing Precipitation Events and Their Typical Lifecycles Over Conterminous United States

    by Siyu Zhu, Guoqiang Tang, Songkun Yan, Yu Du, Yue Xu, Mofan Zhang, Mengye Chen, Huan Li, Yang Hong

    Precipitation is a key driver of the global water cycle and energy circulation, yet its complex formation and dynamic lifecycle make both observation and simulation challenging. Tracking precipitation events and analyzing their lifecycle stages (development, maturity, and dissipation) are essential for understanding precipitation dynamics. This study proposes a morphological precipitation event extraction (MPEE) method and a lifecycle fitting model, applied to CONUS404, Integrated Multi‐satellitE Retrievals for GPM, and ECMWF Reanalysis 5 from 2001 to 2021. Intercomparison results confirm the method's robustness, with most correlation coefficients exceeding 0.5 and most root mean squared errors less than 0.7 mm/hr, between extracted events and simulated life cycles. K‐means clustering reveals four precipitation types: common, high‐peak, long‐duration, and slow‐developing events with delayed peaks. The method effectively captures precipitation variability across data sets and provides a scalable approach for studying long‐term precipitation trends. This work lays a foundation for analyzing climate‐scale precipitation lifecycle changes, improving our understanding of precipitation dynamics and their implications for climate variability.