Sinan Rasiya Koya

Thank you for visiting my website!
I'm Sinan (if you enjoy garlic naan, my name should be easy to remember 😉). Haven't you intrigued by the Earth's marvels around? Its geological features, climate, and ecosystems? I am!

This enthusiasm led me to pursue research focusing on Hydrology. It all started at the Indian Institute of Technology, Gandhinagar, where I received my bachelor's degree in Civil Engineering. Later, I received my PhD in Water Resources Engineering/Hydrology from University of Nebraska-Lincoln, where I explored Artificial Intelligence (AI) and Deep Learning applications in hydrology. I study the hydrological extremes, like floods and droughts, and ways to improve predicting them using AI. During my doctoral studies, I focused on snow-induced floods and droughts. Currently, I am a postdoctoral associate at the Rutgers University, primarily working on operational-level high-resolution flash flood forecasting.

Outside of academia, I enjoy playing badminton and soccer. I am a philomath - learning new things brings me joy. You can learn more about me from the links below.

Publications

Google Scholar
2025
Trends and causal structures of rain-on-snow flooding
Nishant Kumar, Kanak Kanti Kar, Shivendra Srivastava, Sinan Rasiya Koya, Sudan Pokharel, Molly Likins, Tirthankar Roy
Journal of Hydrology
Read here
Seasonal Influences of Precipitation and River Stage on Groundwater Levels in Platte River Watersheds Vulnerable to Spring Floods
Incheol Kim, Sinan Rasiya Koya, Tirthankar Roy, Jongwan Eun
Journal of Hydrologic Engineering
Read here
2024
Temporal fusion transformers for streamflow prediction: Value of combining attention with recurrence
Sinan Rasiya Koya and Tirthankar Roy
Journal of Hydrology
Read here Code
Northern Pacific sea-level pressure controls rain-on-snow in North America
Sinan Rasiya Koya, Kanak Kanti Kar, and Tirthankar Roy
Communications Earth & Environment
Read here Data & Code
2023
An autoencoder-based snow drought index
Sinan Rasiya Koya, Kanak Kanti Kar, Shivendra Srivastava, Tsegaye Tadesse, Mark Svoboda, and Tirthankar Roy
Scientific Reports
Read here
Applicability of a flood forecasting system for Nebraska watersheds
Sinan Rasiya Koya, Nicolas Velasquez Giron, Marcela Rojas, Ricardo Mantilla, Kirk Harvey, Daniel Ceynar, Felipe Quintero, Witold F. Krajewski, and Tirthankar Roy
Environmental Modelling & Software
Read here Code
Snow-detonated floods: Assessment of the US midwest march 2019 event
Nicolás Velásquez, Felipe Quintero, Sinan Rasiya Koya, Tirthankar Roy, and Ricardo Mantilla
Journal of Hydrology: Regional Studies
Read here
2022
The Great Lakes Runoff Intercomparison Project Phase 4: the Great Lakes (GRIP-GL)
Juliane Mai, Hongren Shen, Bryan A. Tolson, Etienne Gaborit, Richard Arsenault, James R. Craig, Vincent Fortin, Lauren M. Fry, Martin Gauch, David Klotz, Frederik Kratzert, Nicole O'Brien, Daniel G. Princz, Sinan Rasiya Koya, Tirthankar Roy, Frank Seglenieks, Narayan K. Shrestha, Agnès G. T. Temgoua, Vincent Vionnet, and Jason W. Waddell
Hydrology and Earth System Sciences
Read here
Application of Neural Networks for Hydrologic Process Understanding at a Midwestern Watershed
Annushka Aliev, Sinan Rasiya Koya, Incheol Kim, Jongwan Eun, Elbert Traylor, and Tirthankar Roy
Hydrology
Read here

Projects

Over the past years, I have been fortunate to be part of various impactful projects. This has been a great learning experience. Huge thanks to my mentors and colleagues who made these projects possible.

Operational-level High-resolution Flash Flood Forecasting System for New Jersey

Flash floods are among the most devastating natural disasters, causing significant damage and loss of life worldwide. With Federal Emergency Management Agency (FEMA) and New Jersey Office of Emergency Management (NJOEM), we are developing an operational-level high-resolution flash flood forecasting system for New Jersey. This system leverages advanced hydrological models with real-time and forecasted data to provide accurate and timely flood forecasts, enhancing the state's preparedness and response capabilities.
Water Cycle Diagram

Global Composite Drought Indicator Hot Spot Early Warning and Information System

Extreme weather can contribute to civil unrest. In the effort lead by Dr. Mark Svoboda, I have been actively involved in the ML implementation aiding the National Drought Mitigation Center to predict civil unrest from droughts. Read more in the news coverage below.
News coverage
Water Cycle Diagram

Streamflow Forecasting with Temporal Fusion Transformers

Does combining recurrence with attention improve time series modeling of streamflow? We addressed this question and showed that it does, by implementing advanced AI forecasting models (Transformers, Temporal Fusion Transformers - TFT, LSTM) for streamflow prediction in 2,610 catchments worldwide.
PyTorch, Lightning, CUDA, HPC, Shell, Xarray, NetCDF
Journal Article | AGU 2023 Conference | Code
Water Cycle Diagram

Prototype Flood Forecasting System for Nebraska

With a team from the Iowa Flood Center (IFC) and Nebraska DOT, I led the implementation of a prototype for a flood forecasting system in Nebraska. We . This project involved enhancement the Hillslope Link Model (HLM) to incorporate snow processes, installation of streamflow sensors, assimilation of streamflow measurements into the model simulation, integration of radar-based precipitation, flood risk vulnerability assessment, and development of a web interface to show the flood conditions.
C++, PostgreSQL, QGIS, MRMS-QPE, HTML, Python, Matlab
Journal Article | AGU 2021 Conference | Code
Water Cycle Diagram

Impacts of Snow droughts and Associated Compound Events on Streamflow

Snow droughts are escalating in the context of climate change. Less snow means less river flow in warmer seasons. We assess global rivers vulnerable to snow droughts by analyzing large-scale climate data.
Python, Google Earth Engine, HPC Parallel computing, ERA5-Land, MERRA2, CMIP6
Water Cycle Diagram

Drivers and Impacts of Rain-on-Snow in United States

Warm rain hitting snow, does that sound trivial? Losses from catastrophic floods induced by rain-on-snow (ROS) events are in the orders of billions of dollars. We address two parts of ROS mechanism. 1. What lead to ROS occurrence, specifically the effects of large-scale climate patterns 2. How does ROS events lead to floods. In the first part, we found an important causal link between the North Pacific sea-level pressure (SLP) pattern and ROS frequency in North America (find the link to journal article below). In the ongoing second part, by simulating land surface processes using Noah-MP, we assess the critical hydrological processes during ROS events that lead to floods.
Noah-MP, FORTRAN, Convergent Cross Mapping, NLDAS-2, GHCN-D
Journal Article
Water Cycle Diagram

Diffusion based rainfall runoff modeling

This ongoing study introduces a rainfall-runoff modeling framework based on Denoising Diffusion Probabilistic Models (DDPM), originally used in image generation. The model predicts streamflow by conditioning each day's discharge on hydroclimatic variables and the model's previous estimate, allowing sequential simulation.
PyTorch, Lightning, HPC, Numba, Dask
Water Cycle Diagram

Great Lakes Runoff Intercomparison Project – Great Lakes (GRIP-GL)

I was part of the study led by Dr. Juliane Mai which benchmarked various hydrologic models across the Great Lakes basin. Our candidate model, HYMOD2, was one of the top performing model in general. HYMOD2-lumped demonstrated exceptional performance in simulating actual evapotranspiration (AET). Read more in the paper below.
Hymod-2, Raven, Conceptual hydrological models, Matlab
Journal Article
Water Cycle Diagram

Topology-based Flood Vulnerability of Global River Networks

A graph theoretic analysis of global river networks to assess topology-based flood risk. Using Maximum flow algorithms on directed tree networks representing major global rivers, we revealed regions prone to flooding on a topological basis.
Python, NetworkX, Cytoscape, iGraph, ArcGIS
Code
Water Cycle Diagram

Synthetic Global Precipitation Maps by Generative Adversarial Networks

Generative Adversarial Networks (GANs) shifted the paradigm of AI with its remarkable ability to generate realistic images. Here, we use GANs to generate global precipitation maps at a resolution of 2.5°x2.5°. Read more in the report below.
PyTorch, GANs
Report | Code
Water Cycle Diagram

Learning

Thoughts

Contact