About

Praveen Kumar holds a B.Tech. from the Indian Institute of Technology, Bombay, an M.S. from Iowa State University, and a Ph.D. from the University of Minnesota, all in civil engineering. He has been a faculty member in the Department of Civil and Environmental Engineering at the University of Illinois since 1995. Prior to this, he worked as a research scientist at the Universities Space Research Association and NASA-Goddard Space Flight Center. Dr. Kumar's teaching encompasses undergraduate and graduate courses in hydrosystems engineering, engineering modeling under uncertainty, surface water hydrology, hydroclimatology, stochastic hydrology, non-linear methods in hydrology, and hydroinformatics. His research focuses on Hydrocomplexity, which involves understanding and predicting emergent patterns arising from complex non-linear multi-scale interactions between soil, water, climate, vegetation, and human systems, with the aim of developing innovative solutions to water and sustainability challenges. He is currently the Director of the Critical Zone Observatory for Intensively Managed Landscapes and leads NSF-supported projects on Geo-Semantics, cyber-infrastructure for long-tail data and models, and EarthCube initiatives.

Research topics

• Environmental science
• Ecology
• Geology
• Water resource management
• Meteorology
• Agricultural engineering
• Geography
• Agronomy
• Biology
• Engineering
• Climatology

Selected publications

• Image authentication with topologically stable C-point singularities

  Applied Optics · 2026-04-15

  articleSenior author

  Optical authentication is a widely adopted approach in optical cryptography due to its inherent parallelism and resistance to digital attacks. We report here a novel, to our knowledge, optical image authentication scheme that utilizes the topological stability and uniqueness of C-point polarization singularity configurations within the vector light fields. C-points are utilized for multi-data encoding, resulting from their topological and polarization degrees of freedom (DOF) originating from their variable spatial distribution and singularity order. Authentication involves image encoding and validation through intensity-based analysis, eliminating the need for interferometric data retrieval. This approach significantly simplifies the optical implementation while maintaining high sensitivity to experimental deviations.

  PublisherDOI

• Comparison of Spherical and Cylindrical Weak Shock Waves in Highly Viscous Medium

  Procedure International Journal of Science and Technology · 2025-01-01

  article1st authorCorresponding
  PublisherDOI

• Phase-Space Divergence as the Driver of Information Flow in Dynamical Systems

  2025-03-14

  preprintOpen access1st authorCorresponding

  Understanding the mechanisms that generate information flow in dynamical systems is crucial for advancing causal inference and dependency characterization in natural and engineered systems. Information flow is defined as the exchange of predictive or uncertainty-reducing knowledge between variables in a coupled system, arising when fluctuations in one component influence the variability in another. This study establishes that information flow emerges as a direct result of trajectory divergence in phase-space, an effect encoded in the generalized dynamics of probability density functions. We show that when the divergence of flow fields in phase-space is non-zero, it induces temporal changes in the entropic structure of the system. This expands the traditional Liouville equation to non-conservative systems. This divergence creates, rather than merely propagates, informational dependencies among system components, highlighting the dynamic nature of mutual and multivariate information in such systems.Our results reveal that in conservative systems, where phase-space volume is preserved, the system entropy remains invariant, and informational dependencies are determined solely by initial conditions. In contrast, dissipative systems—exemplified by the damped harmonic oscillator and the Lorenz system—exhibit significant entropic and informational evolution driven by non-zero divergence. The mathematical framework presented quantifies the role of divergence in shaping joint, marginal, and conditional entropy, as well as bivariate and higher-order mutual information. This approach provides a comprehensive understanding of how phase-space dynamics underpin the flow and transformation of information.The findings have profound implications across multiple domains, including environmental science, climate dynamics, and engineered systems, where causal relationships often arise from interactions between variables in complex networks. By bridging physical principles with information theory, the work offers a new lens for exploring the dynamics of natural and artificial systems, with potential applications in predictive modeling, data assimilation, and the design of resilient systems under uncertainty.This investigation not only addresses a longstanding question about the origin of information flow in coupled systems but also lays the groundwork for future studies incorporating time-lagged dependencies and higher-order interactions in both theoretical and applied contexts. The framework proposed herein enables a more refined analysis of information flow in complex systems, advancing our ability to interpret, predict, and engineer their behavior. 

  PublisherDOI

• Agentic LLM Framework for Adaptive Decision Discourse

  ArXiv.org · 2025-02-16

  preprintOpen accessSenior author

  Effective decision-making in complex systems requires synthesizing diverse perspectives to address multifaceted challenges under uncertainty. This study introduces an agentic Large Language Models (LLMs) framework for simulating decision discourse - the deliberative process through which actionable strategies are collaboratively developed. Unlike traditional decision-support tools, this framework simulates diverse stakeholder personas, each bringing unique priorities, expertise and value-driven reasoning to a dialogue that emphasizes trade-off exploration in a self-governed assembly. We present explorative results fostering robust and equitable recommendations, with two use cases: first, our framework simulates a response to the floods that occurred on July 2025 in Texas; second, a hypothetical extreme flooding in a Midwestern township under varying forecasting uncertainty. Recommendations made balance competing priorities considered through social, economic and environmental dimensions, setting a foundation for scalable and context-aware recommendations and transforming how decisions for real-world high-stake scenarios can be approached in digital environments. This research explores novel and alternate routes leveraging agentic LLMs for adaptive, collaborative, and equitable recommendations, with implications across domains where uncertainty and complexity converge.

  PublisherOA PDFDOI

• Improving Stream Solute Predictions With a Modified LSTM Model Incorporating Solute Interdependences and Hysteresis Patterns

  Journal of Geophysical Research Machine Learning and Computation · 2025-02-12 · 1 citations

  articleOpen accessSenior authorCorresponding

  Abstract Surface runoff and infiltrated water en route to the stream interact with dynamic landscape properties, ranging from vegetation and microbial activities to soil and geological attributes. Stream solute concentrations are highly variable and interconnected due to these interactions, flow paths, and residence times, and often exhibit hysteresis with flow. Significant unknowns remain about how point measurements of stream solute chemistry reflect interdependent hydrobiogeochemical and physical processes, and how signatures are encapsulated as nonlinear dynamical relationships between variables. We take a Machine Learning (ML) approach to understand and capture these dynamical relationships and improve predictions of solutes at short and long time scales. We introduce a physical process‐based “flow‐gate” into an Long Short‐Term Memory (LSTM) model, which enables the model to learn hysteresis behaviors if they exist. Further, we use information‐theoretic metrics to detect how solutes are interdependent and iteratively select source solutes that best predict a given target solute concentration. The “flow‐gate LSTM” model improves model predictions (1%–32% decreases in RMSE) relative to the standard LSTM model for all nine solutes included in the study. The predictive improvements from the flow‐gate LSTM model highlight the importance of lagged concentration and discharge relationships for certain solutes. It also indicates a potential limitation in the traditional LSTM model approach since flow rates are always provided as input sources, but this information is not fully utilized. This work provides a starting point for a predictive understanding of geochemical interdependencies using machine‐learning approaches and highlights potential improvements in model architecture.

  PublisherOA PDFDOI

• The contribution of sterodynamic changes to the uncertainty of regional sea-level projections over the 21st century

  2025-05-15 · 1 citations

  preprintOpen access

  Understanding the drivers of future sea-level uncertainty is essential for effective coastal planning. This study quantifies the contribution of sterodynamic sea-level change to regional sea-level uncertainty using the Framework for Assessing Changes To Sea-level (FACTS). We decompose total uncertainty across time, regions, scenarios, and projection methods. Even when considering deeply uncertain ice sheet contributions, sterodynamic processes remain the dominant source of uncertainty over the next decades (2020-2050), accounting for at least 80% over the North Atlantic and Southern Ocean. The contribution of sterodynamic sea-level change to the total uncertainty decreases towards the end of the 21st century as the contribution from ice sheets becomes more pronounced, but it likely remains an important source of regional sea-level uncertainty at 2100 along the North American east coast and European coastlines. The spatiotemporal pattern of sterodynamic significance does not change with future greenhouse gas emissions, yet its overall role depends on the approach taken to model the Antarctic ice sheet. Our results highlight the critical role of sterodynamic processes for near-term coastal planning decisions, especially along the North Atlantic coastlines. Targeted improvements in ocean modeling, particularly in the representation of ocean circulation changes, such as the Atlantic Meridional Overturning Circulation strength, provide potential pathways to reduce uncertainty in sea-level projections. Using the aforementioned criteria, we demonstrate a narrowing of the likely range of relative sea-level rise projections at 2100 by up to 10% along the U.S. northeast coast.

  PublisherOA PDFDOI

• Drought constrictions on lateral carbon transport

  Nature Geoscience · 2025-09-26

  article
  PublisherDOI

• The Motion of Weak Plane Shock Waves in Highly Viscous Medium

  Procedure International Journal of Science and Technology · 2025-01-01

  article1st authorCorresponding
  PublisherDOI

• AI-Based Assistive System for Correcting Dysgraphia and Dyscalculia in Children

  2025-08-27

  articleSenior author

  Dysgraphia and dyscalculia make it to the list of learning disabilities that are most frequently noticed in students as a significant hindrance towards their academic performance. The aim of this research is to formulate strategies for the specific difficulties faced by students suffering from dysgraphia and dyscalculia. Dysgraphia students' written sentences which contain grammatical and spelling errors were reconstructed using the T5 transformer model. Similarly, the students' works were analyzed through the use of BERT model which focused on mathematical reversals, operator, and other errors made by the dyscalculic students. In a similar way a program is developed in python to assess and predict the types of errors based on the values obtained from the BERT model. Evidence from this research demonstrates how everything is about transformer models these days, as they can actually allow you to create practical assets and adopt learning styles suitable for students with learning disabilities.

  PublisherDOI

• Integrating flow and solute flux dynamics in an adaptive LSTM model for stream chemistry predictions

  2025-11-29

  articleSenior author

  High-frequency measurements of stream solute dynamics reveal fluctuations across a multitude of time-scales. At the storm-event time-scales they expose solute specific rapid mobilization or dilution effects, in addition to hysteresis. For various applications, predictions of these dynamics at measurement site are desired to inform impending events such as harmful algal blooms. To aid in such efforts, we present a machine learning adaptive model. It builds on the well-established Long Short-Term Memory (LSTM) model, but introduces important modifications to capture dynamical attributes such as hysteresis, mobilization, and dilution. These modifications incorporate flow and flux gates that embed flow- and flux-gradients directly into the traditional model architecture alongside the input, output, and forget gates. These gates are activated when flow or flux gradients or both exceed certain thresholds. The success of these activations in improving model performance indicates that they capture processes that better inform predictability of dynamical attributes embedded in the observed solute variability. The study is performed using RiverLab data, a stream chemistry laboratory in the field located on the bank of a stream, at two sites. The first site is the Upper Sangamon Basin in Illinois, U.S.A., and the second in Orgeval, France. Both are tile drained loess covered agricultural watersheds, but of different drainage areas and in different climates. We assess the model performance using various metrics aligned with our goal to capture key dynamical features. The results justify the need for going beyond the standard formulations of LSTM to better accommodate the nuanced dynamics of natural systems.

  PublisherDOI

Recent grants

• EarthCube Building Blocks: Collaborative Proposal: A Geo-Semantic Framework for Integrating Long-Tail Data and Models

  NSF · $647k · 2014–2017

• EAGER: Comparative Study of Critical Zones of Intensively Managed Loess Landscapes in China and United States

  NSF · $50k · 2017–2019

• Collaborative Research: NeTS-NOSS: Towards a Theory of In-Network Computation for Surveillance and Monitoring in Wireless Sensor Networks

  NSF · $420k · 2005–2010

• Interactions Between Water, Energy and Carbon Dynamics as Predictors of Canopy to Ecosystem Scale Vegetation Pattern and Function in a Changing Environment

  NSF · $1.6M · 2006–2011

• Critical Zone Observatory for Intensively Managed Landscapes (IML-CZO)

  NSF · $6.5M · 2013–2021

Frequent coauthors

• Phong V. V. Le

  Oak Ridge National Laboratory

  57 shared

• Andrew J. Stumpf

  University of Illinois Urbana-Champaign

  42 shared

• Jinhui Yan

  University of Illinois Urbana-Champaign

  36 shared

• Thanh H. Nguyen

  University of Illinois Urbana-Champaign

  36 shared

• Andrew W. Taylor‐Robinson

  University of Pennsylvania

  36 shared

• Brian F. Allan

  University of Illinois Urbana-Champaign

  36 shared

• Lei Zhao

  36 shared

• Efi Foufoula‐Georgiou

  Irvine University

  28 shared

Labs

• Climate, Meteorology & Atmospheric SciencesPI

Education

• Ph.D., Civil Engineering

  University of Minnesota System

  1993

Awards & honors

• Universities Space Research Association Award for Promise an…
• NASA New Young Investigator Award (1996)
• Xerox Award for Faculty Research (2005)
• Engineering Council Award for Excellence in Advising (2013)
• Featured Article, Water Resources Research (2011)