Water Resource Forecasting

After more than 10 years of experience in hydrology, I still see surface runoff as the most visible and misunderstood part of the hydrological cycle. As rain falls on the land, some evaporates, some infiltrates, and some recharges groundwater. The remaining water flows over the surface as runoff. This simple process controls floods, soil erosion, reservoir inflow, urban drainage, and water quality. Understanding runoff means understanding how a catchment responds to climate, land use, and human activity. How surface runoff forms Runoff is generated when: • Rainfall intensity exceeds infiltration capacity • Soil becomes saturated • Land is sealed by roads and buildings • Slopes accelerate overland flow This is why rainfall alone never tells the full story. Simple ways to estimate runoff: For students, consultants, and early-career hydrologists, these methods still matter: • Runoff coefficient method • Rational method • SCS Curve Number method • Water balance approach • Infiltration index methods (phi and W index) • Unit hydrograph method • Regional empirical equations • Time of concentration-based estimates • Excel-based rainfall runoff calculations Simple does not mean wrong. Many design decisions rely on these methods every day. Widely used hydrological models When scale and complexity increase, models help us organize the hydrological cycle: • HEC-HMS for event-based flood modeling • SWAT for long-term basin-scale runoff and land use studies • MIKE SHE and MIKE 11 for integrated surface and groundwater analysis • VIC and TOPMODEL for regional and terrain-driven runoff processes • IHACRES for data-limited catchments Each model is a tool. None is universal. AI and machine learning in runoff estimation Data-driven methods are now common, especially for forecasting: • Artificial Neural Networks • Random Forest and Decision Trees • Support Vector Machines • Deep learning models such as LSTM They can predict runoff well but often explain little. Physical understanding still matters. A simple rule from experience Start simple. Match the method to your data. Always verify a model against real data. Surface runoff is not just a number. It is the heartbeat of a watershed and the link between climate, land, and society. If you work in water, you work with runoff, whether you realize it or not. #SurfaceRunoff #Hydrology #RainfallRunoff #HydrologicalCycle #WatershedHydrology #HECHMS #SWATModel #HydrologicalModeling #RunoffModeling #FloodModeling #HydrologyAndAI #MachineLearningInHydrology #AIForWater #DataDrivenHydrology #WaterResources #ClimateChangeImpacts #FloodRisk #SustainableWater #WaterSecurity #WaterProfessionals #HydrologyStudents #EnvironmentalEngineering #SWAT #HEC-HMS #AI #Sustainability #Flood #CivilEngineering #ResearchAndPractice #STEM #ScienceCommunication #KnowledgeSharing #LearningEveryday #CFBR

This week, Google Research open-sourced our entire hydrology modeling framework on GitHub. This release allows researchers and operational forecasters worldwide to build and train AI flood forecasting models using the exact same architecture that powers Google’s Flood Hub. Our GoogleHydrology Python package uses PyTorch to implement an upgraded ME-LSTM architecture. By processing multi-source meteorological inputs into a unified system, our latest benchmarking shows this model extends the reliable predictive horizon by six days in gauged basins and by one day in ungauged basins compared to our original 2024 version. By releasing the model architecture and training pipeline openly, national meteorological services and local authorities can integrate their own specialized data while retaining full control over their inputs. We’ve already seen the operational potential of this approach through our partnership with the Czech Hydrometeorological Institute (CHMI), who validated the model's quality against traditional forecasting methods and built an adapter to integrate our framework directly into standard industry workflows like the Delft-FEWS platform. While established institutions like CHMI demonstrate how global agencies can enhance existing water management workflows, this open-source release is also built to be highly accessible and computationally efficient. By removing the need for costly traditional infrastructure, it simultaneously democratizes access for resource-constrained regions and local teams who need high-caliber insights the most. To explore the model code, interactive tutorials, and technical benchmarks, read the full blog post: https://lnkd.in/djWtMM-b

Let’s talk about the future of water resources and how AI can help protect communities. I'm incredibly proud that foundational Google research on AI's role in hydrology was just highlighted by Water Resources Research as one of five significant articles from the past 60 years! This work demonstrated how AI models can achieve superior streamflow prediction, even in data-scarce regions, fundamentally shifting perspectives in the field. This kind of pioneering research is exactly what enables us to build tools with real-world impact. Here’s how we're translating that potential into action: 𝟭) 𝗚𝗹𝗼𝗯𝗮𝗹 𝗥𝗲𝗮𝗰𝗵 & 𝗘𝗮𝗿𝗹𝘆 𝗪𝗮𝗿𝗻𝗶𝗻𝗴𝘀: We're leveraging AI to provide life-saving flood forecasts globally through the Google Flood Hub, now live in over 100 countries, covering locations in which over 700 million people live. 𝟮) 𝗔𝗰𝘁𝗶𝗼𝗻𝗮𝗯𝗹𝗲 𝗔𝗜: Our sophisticated AI models provide crucial real-time forecasts, 5-7 days in advance, giving individuals and authorities valuable time to prepare and stay safe. 𝟯) 𝗔𝗰𝗰𝗲𝘀𝘀𝗶𝗯𝗹𝗲 𝗜𝗻𝗳𝗼𝗿𝗺𝗮𝘁𝗶𝗼𝗻: These warnings are integrated directly into tools people use daily – Google Search, Google Maps, and Android notifications – making critical information widely accessible when it matters most. This journey from fundamental AI discovery to operational, global-scale tools like the Google Flood Hub exemplifies our commitment to using AI for climate adaptation and resilience. Congratulations to Grey Nearing , Frederik Kratzert, and the entire team whose influential work continues to bridge AI, hydrology, and community safety! Link to the WRR-highlighted article: https://lnkd.in/gW9yWB3A Read more about our global flood forecasting initiative here: https://lnkd.in/g7b2gBNt Google Flood Hub: https://lnkd.in/g6pbzqQZ

⚡ The hydropower sector is splitting in two: those who are fluidly adapting to new information and those still clinging to The Old Way. That gap will only increase. If you're still making water/weather decisions the same way you did two (five ((TEN?!)) years ago… 😬 I wrote a piece for National Hydropower Association on the four ways AI is fundamentally reshaping how we make and use forecasts: AI forecasts are becoming fit-for-purpose → Run a solid model on your desktop, not a supercomputer. The resource efficiency alone is quietly revolutionary. (Also: resilience during cyber attacks. Air-gapped forecasting in an outage? Now we're talking.) The single-best-model mindset is dead → Decisions are leveraging many forecasts. The world is too heterogeneous for a singular "best." Get comfortable with ensemble thinking. Uncertainty isn't the enemy → Navigate it with probability. Define a 90th percentile spill protocol and you're essentially buying insurance against the long tail of a chaotic atmosphere. End-to-end forecasts are the next inflection point in weather There's been a gasp-inducing paper every month. Organizations attuned to these advances will navigate weather extremes and grid chaos more effectively. Those that don't will be flying blind in increasingly … interesting times. Bonus: Zero-shot forecasts are producing results that would've seemed impossible 18 months ago. Still nascent, but already showing up in industrial applications. Worth watching. Full article: https://lnkd.in/eUGYMcX5

🚀 Exploring Global Surface Water Storage Dynamics using GRACE Data in Google Colab (Python) + Google Earth Engine 🌍💧📊 I’ve recently been working on a geospatial workflow for monitoring and analyzing terrestrial water storage changes using the NASA GRACE Mascon dataset in Google Earth Engine (GEE) with Python. This workflow focuses on transforming satellite-derived gravimetric data into meaningful hydrological insights through a structured analytical pipeline that includes: ✅ Automatic extraction of the selected country/region of interest ✅ Preprocessing of GRACE monthly liquid water equivalent thickness (LWE) data ✅ Conversion of Earth Engine ImageCollection into xarray for time-series analysis ✅ Monthly mean water storage mapping ✅ Annual water storage trend analysis ✅ Computation of long-term water storage anomalies ✅ Visualization of spatial patterns and temporal dynamics ✅ Interactive time-series plotting for hydrological interpretation Using tools such as Google Earth Engine, geemap, xarray, xee, matplotlib, and plotly, the workflow enables efficient analysis of large-scale groundwater and surface water storage variability over time. 🌐 Why this matters: Monitoring water storage anomalies is essential for understanding: ~Hydrological droughts ~Groundwater depletion ~Climate variability impacts ~Water resource planning and management ~Catchment-scale and national-scale water security This kind of geospatial-hydrological integration has strong potential for supporting climate adaptation, water governance, and evidence-based decision-making, especially in water-stressed regions. I’m increasingly interested in applying remote sensing, GIS, Earth observation, and GeoAI to solve practical problems in: 💧 Water Resources 🌱 Agriculture & Irrigation 🌍 Climate Risk 🛰️ Environmental Monitoring #GoogleEarthEngine #Python #GRACE #RemoteSensing #GIS #Hydrology #WaterResources #ClimateChange #EarthObservation #Geospatial #DataScience #GeoAI #WaterSecurity #EnvironmentalMonitoring #SatelliteData #Hydroinformatics

🌊 New Research Published in Earth’s Future | IIT Gandhinagar How will India’s rivers respond to a warming climate — and can we actually trust the projections? Our latest work tackles exactly this. In a paper just published in Earth’s Future (AGU), Dipesh Singh Chuphal and I present observation-constrained streamflow projections for nine major Indian rivers using a Bayesian detection-attribution framework (the KCC method). Why does this matter? Climate models disagree sharply on future precipitation over India — and that uncertainty cascades directly into river flow projections, making long-term water planning extremely difficult. What we did differently: We anchored future projections to 160+ years of observed streamflow and global temperature records, dramatically narrowing the uncertainty range. Key findings: ✅ Uncertainty in future streamflow reduced by ~⅓ compared to raw climate model ensembles ✅ Most Indian rivers — Ganga, Brahmaputra, Indus, Narmada, Krishna, Godavari — project significant flow increases by mid-to-late century ⚠️ The Cauvery basin is an exception, facing near-term declines and potential water scarcity 🎯 Using only the 8 best-performing climate models further reduces uncertainty to ~20% The bigger picture: Higher future flows aren’t just good news — they also signal elevated flood risk downstream. Our results carry direct implications for hydropower planning, transboundary water governance, food security, and climate-resilient infrastructure across a region home to nearly 2 billion people. Congratulations to Dipesh for leading this work with great rigor and dedication. This research was supported by the DST Major Research and Development Program. 📄 Read the full paper: https://lnkd.in/g2j9KwbG 📦 Data available on Zenodo: https://lnkd.in/g-cnPAZP #ClimateChange #WaterResources #Hydrology #IndianRivers #ClimateAdaptation #CMIP6 #StreamflowProjections #IITGandhinagar #EarthsFuture #AGU #MonsoonClimate #WaterSecurity #ClimateScience #HydrologicalModeling #BayesianStatistics #DST #FoodSecurity #FloodRisk #Drought #NetZero #SouthAsia #ClimateResilience #IPCC #OpenScience #OpenData

Interpretable Physics-Informed Graph Neural Networks For Flood Forecasting -- https://lnkd.in/gS369ty7 <-- shared paper -- https://lnkd.in/gftbrmrA <-- shared GitHub repository -- H/T Mehdi Taghizadeh “Climate change has intensified extreme weather events, with floods causing significant socioeconomic and environmental damage. Accurate flood forecasting is crucial for disaster preparedness and risk mitigation, yet traditional hydrodynamic models, while precise, are computationally prohibitive for real-time applications. Machine learning surrogates, such as graph neural networks (GNNs), improve efficiency but often lack physical consistency and interpretability. This paper introduces HydroGraphNet, a novel physics-informed GNN framework that, for the first time, integrates the Kolmogorov–Arnold Network (KAN) to enhance model interpretability in unstructured mesh-based flood forecasting. The framework embeds mass conservation laws into the loss function, ensuring physically consistent predictions. Additionally, it employs an autoregressive encoder–processor–decoder architecture that captures spatiotemporal flood dynamics while mitigating error accumulation over long forecasting horizons. Validation on flood data from the White River near Muncie, Indiana, demonstrates a 67% reduction in prediction error, near-zero mass balance error, and a 58% improvement in the critical success index for major flood events compared to a baseline GNN model. These results highlight the potential of the proposed framework to advance real-time flood forecasting with improved physical consistency and interpretability. #GIS #spatial #mapping #FloodForecasting #GraphNeuralNetworks #PhysicsInformedAI #Hydrology #ClimateResilience #ScientificMachineLearning #CivilEngineering #HydroGraphNet #PhyiscsNeMo #hydrology #water #flood #flooding #spatialanalysis #spatiotemporal #climatechange #model #modeling #machinelearning #AI #extremeweather #prediction #forecasting #planning #humanimpacts #loss #lossoflife #publicsafety #cost #economics #infrastructure #risk #hazard #disaster #mitigation #preparedness #naturaldisaster #HydroGraphNet #floodforecasting #dynamics #remotesensing #earthobservation

A really nice Unconventional Resources Technology Conference (URTeC) publication by the team at Texas Tech University and the Texas Produced Water Consortium: "Produced Water Management in the Permian Basin to Accommodate the Arid West Texas Region: Historical Production, Forecast, and Water Composition" Marshall Watson Talal Gamadi, PhD. Elvin Hajiyev Ayann Tiam Produced Water Society #water #brine #energy #environment "The Permian Basin has been the US largest producing basin since 2018, contributing to 28% of the US hydrocarbon energy ever since. Large volumes of produced water (PW) are left to be managed as a result. If not used in oilfield applications, PW is injected into saltwater disposal (SWD) wells. SWD is undesirable for increasing seismic activity and expediting regulatory action that could restrict hydrocarbon production. New Mexico has limited SWD more than Texas, leading to 2 MMbwpd to be disposed of in West Texas. A total of about 16 MMbwpd is disposed of in West Texas, increasing seismic risk in a region faced with 22 MMbwpd in water shortage through 2070. This study evaluates the basin's PW quality and forecasts production to minimize SWD and repurpose PW for beneficial use in West Texas. The forecast details 3 cases: low, base and high. The Permian Basin's PW from unconventional resources is set to peak at 28.8 MMbwpd in 2039 (base case) whilst producing 7.6 MMbopd and 42.1 bcf/d. The PW from unconventional wells is often injected into SWD wells, if not used for hydraulic fracturing. But if treated at a 50% recovery rate, the PW from tight-oil wells in the TX part of the Permian would meet 18% to 45% of the West-Texas irrigation water shortage through 2070. This was done by considering recent PW recycling percentages in fracking and estimating future trends. The Delaware Basin in Texas seems to be the most convenient region to treat PW from beneficial reuse, due to low water demand for fracking and low salinity. This study is novel as it provides a detailed rigorous forecast by county of the Permian Basin through 2070 and lays out the basis for future work to help evaluate reuse of PW in beneficial reuse throughout West Texas. The study also looks at geographical distribution in lithium and boron, amongst other chemical properties that can be provided on demand." https://lnkd.in/g7ruxS7M

“Karst aquifer recharge in Israel and the West Bank”   Throughout the southern and eastern Mediterranean region, river catchments, like others around the world, are expected to face a significant increase in groundwater STRESS, with freshwater resources predicted to DECLINE by as much as 30–50% by 2050.    I typically don’t like to promote long-term modeling projections, since most are still highly dubious and heavily conditioned (or at least should be), but it is important to know the reported magnitudes of change.   This recent study utilized an ensemble of event-based recharge calculation methods to address the challenge of structural uncertainty for the Western Mountain Aquifer, a Mediterranean karst aquifer located in Israel and the West Bank.   The Soil and Water Assessment Tool (SWAT) and the process-based infiltration model (PIM) were compared to site-specific, empirical regression models.    The SWAT and PIM mean annual recharge estimates ranged from 32–34.6% of precipitation, almost equating to the results of empirical regression models (32–36%).  Future recharge predictions under the influence of climatic forcings were quantified by parameterizing the SWAT and PIM methods with a downscaled regional climate model of Israel.   SWAT predicted a 23% DECREASE in recharge by 2051–2070, relative to 1981–2001.  By contrast, PIM showed only a 9% DECREASE, possibly due to the representation of infiltration through preferential flow pathways and the exclusion of surface runoff processes.   While key methodological differences in the representation of hydrological processes were evident, both methods effectively provided good estimates of groundwater recharge.     The recharge rates estimated from the various methods were then integrated into MODFLOW to assess their relative influence on groundwater storage dynamics.  The ensemble of MODFLOW projected groundwater storage outputs should be able to help provide useful guidance for sustainable groundwater management in the region. Study details are provided in Hepach et al. (2024) in Hydrogeology Journal, “Comparison of methods to calculate groundwater recharge for karst aquifers under a Mediterranean climate”

Excited to share that our manuscript, “Physics-informed deep learning reveals climate-driven snowpack decline and threatens ecological water availability in a Californian snow-fed catchment”, led by my PhD student Surendra Maharjan has been published in Ecological Informatics (Impact Factor – 7.3, CiteScore – 11.4). This study focuses on the Upper West Walker River Watershed in the eastern Sierra Nevada, California, a mountainous snow-fed region where seasonal snowpack acts as a natural reservoir, storing water in winter and releasing it gradually to sustain streamflow and ecological systems downstream. However, climate warming is shifting precipitation from snow to rain and accelerating melt, increasing vulnerability and creating an urgent need for advanced tools that can forecast ecological water risks. To address this challenge, this study evaluates three modeling approaches: the process-based SWAT hydrological model, a data-driven Long Short-Term Memory (LSTM) deep learning model, and a Physics-Informed LSTM (PIML) that integrates melt physics and precipitation-phase constraints. Key Findings: The PIML model demonstrated the most robust and well-balanced performance across key hydrologic metrics (NSE, KGE, RMSE). Future climate projections indicate that peak SWE may decline by up to 60%. Peak discharge may decrease by about 33% under warming conditions. Snowmelt and runoff may shift 10–19 days earlier, shortening the hydrologic season. These changes compress the hydrologic season, threaten summer ecological water availability, and heighten drought risk across snow-fed systems. The results underscore the growing challenges of managing water resources in snow-dominated basins under climate change. Coupling physics with deep learning offers a promising path toward more reliable forecasting of snowpack dynamics and streamflow in mountain watersheds. EssDs Chapman Chapman University Schmid College of Science and Technology Surendra Maharjan Wenzhao Li Rejoice Thomas Shahryar Fazli Hesham Morgan Mohamed Allali Ali Elgendy Link : https://lnkd.in/g4f6VpiK