ABSTRACT:

This resource contains 3 items: A Colab notebook demonstrating how to use the NWPS API and the NWM API through a simple flood warning system; Another Colab notebook that demonstrates the use of the GEOGLOWS API through a similar flood warning system; A PowerPoint that was presented at the 2026 Tethys Platform Summit. All three of these resources were used in a 90 minute workshop meant to familiarize participants with the use of public APIs. These notebooks are fully working and ready to be copied, but users must have access to a NWM API key for full functionality. One can request a key at https://hub.ciroh.org/docs/products/data-management/bigquery-api/ . For further information please see the README document.

Acknowledgements:
This research was supported by the Cooperative Institute for Research to Operations in Hydrology (CIROH) with funding under award NA22NWS4320003 from the NOAA Cooperative Institute Program. The statements, findings, conclusions, and recommendations are those of the author(s) and do not necessarily reflect the opinions of NOAA.

ABSTRACT:

This workshop introduces machine learning for hydrologic applications and works through a reproducible modeling workflow in PyTorch. The first half covers ML fundamentals: the supervised, unsupervised, and reinforcement learning paradigms, common task types and algorithm families, and the practical limitations that affect real projects, such as data quality, overfitting, and interpretability. The hands-on portion builds an LSTM model to predict streamflow from the CAMELS dataset, using 10 gauged basins in Southern Appalachia. Participants load and explore the data, then preprocess it with attention to leakage: a temporal train/validation/test split, normalization computed on training data only, log-transformed streamflow, and 365-day input sequences. The model is trained with Adam and a masked MSE loss, then evaluated on a held-out 2008–2014 test period using per-basin NSE scores, hydrographs, and residual plots. The second half turns to model improvement. After diagnosing whether the architecture or the underlying data is the limiting factor, the workshop compares four variants on the same test basins: a baseline LSTM, a version augmented with static basin attributes, a deeper two-layer LSTM, and a causal-convolution ConvLSTM. Throughout, the emphasis is on reading the full distribution of per-basin performance rather than a single average, and on deciding what to change next based on evidence.

Acknowledgements:
This research was supported by the Cooperative Institute for Research to Operations in Hydrology (CIROH) with funding under award NA22NWS4320003 from the NOAA Cooperative Institute Program. The statements, findings, conclusions, and recommendations are those of the author(s) and do not necessarily reflect the opinions of NOAA.

ABSTRACT:

This workshop makes the case that incorporating end-user and stakeholder perspectives strengthens the research-to-operations (R2O) pipeline, and gives participants concrete social science methods for doing it. The framing organizes the value of user integration around three goals: making research outputs valid (revealing operational pitfalls and ground-truthing findings), relevant (grounding work in users' lived experience and context), and actionable (producing tools that solve articulated problems and sustain long-term engagement). These ideas are illustrated with CIROH-funded project examples spanning flood warning communication, benefits analysis, flash flood forecasting, usability testing of NOAA flood inundation maps, and community-based work such as FloodSavvy and the Coupled Human-Water Systems Think Tank. The materials then introduce a "user integration toolbox" of audience research methods that hydrologic researchers can apply themselves or pursue through collaboration: community-based participatory research, research ethics considerations for human-subjects work, semi-structured interviews, focus groups, and surveys, along with a candid look at where AI tools can and cannot substitute for social science expertise. The deck is built around an interactive session format, including small-group discussions and a reflection exercise in which participants identify a key audience and draft questions that audience insights could help answer. It is intended for hydrologic and water-resources researchers who want to engage users more systematically but lack formal social science training.

Acknowledgements:
This research was supported by the Cooperative Institute for Research to Operations in Hydrology (CIROH) with funding under award NA22NWS4320003 from the NOAA Cooperative Institute Program. The statements, findings, conclusions, and recommendations are those of the author(s) and do not necessarily reflect the opinions of NOAA.

ABSTRACT:

This workshop introduces FLDPLN, a low-complexity flood inundation mapping model developed at the University of Kansas, and walks through a complete end-to-end mapping workflow using the 2018 Labor Day flood on Wildcat Creek, Kansas as a worked example. FLDPLN sits between full hydrodynamic models and simpler approaches: it takes a hydro-conditioned DEM as its primary input and combines two flooding mechanisms — backfill flooding ("water seeks its own level") and spillover flooding ("water flows downhill") — to produce depth-to-flood values across the floodplain. Using small depth increments and allowing spillover between iterations resolves the discontinuities that arise from ridgelines when backfill is used alone. The model represents the floodplain as a many-to-many relationship between flood stream pixels (FSPs) and floodplain pixels (FPPs), with depth-to-flood values stored in a precomputed library. At map time, observed or forecasted gauge stages are interpolated along the stream network — using either linear interpolation between gauges or a volume-based method with segment stage-volume rating curves — and a flooded pixel's depth is computed from the maximum difference between FSP flow depth and its DTF. The hands-on portion covers the full Kansas FIM workflow through four Jupyter notebooks: preparing the DEM and building the Python environment, generating a segment-based FLDPLN library, tiling that library for efficient lookup, mapping the 2018 Wildcat Creek event against USGS gauges and the NOAA/NWS reference inundation map, and producing specialized outputs such as minimum depth-to-flood, FSP count, and depression depth maps. Materials include notebooks and a Python package on GitHub, an ArcGIS toolbox, and a MATLAB Runtime–based installer; the operational Kansas Flood Mapping Dashboard, which serves real-time FIM products across eight river basin libraries statewide, illustrates how the same workflow scales beyond a single creek.

Acknowledgements:
This research was supported by the Kansas Water Office (KWO) / KDEM and by the Cooperative Institute for Research to Operations in Hydrology (CIROH) with funding under award NA22NWS4320003 from the NOAA Cooperative Institute Program. The statements, findings, conclusions, and recommendations are those of the author(s) and do not necessarily reflect the opinions of NOAA.

ABSTRACT:

This workshop introduces the economic methods used to evaluate streamflow forecasts and quantify their impacts, framed within the broader practice of benefit-cost analysis (BCA) for federal water resource decisions. The first part reviews how agencies including USACE, FEMA, EPA, and NOAA apply BCA to flood control, and draws a contrast between structural investments such as dams and levees, for which tools like HAZUS and HEC-FDA already exist, and nonstructural investments such as forecast and early warning systems, where the benefits depend on human decisions and behaviors that those tools do not directly capture. The conceptual core of the workshop is the value-of-information (VOI) framework and its 2x2 cost-loss model, in which a decision maker chooses between a protective action and inaction under uncertainty, and the forecast's value is the expected reduction in cost plus loss relative to a without-forecast counterfactual. The relative economic value (REV) metric is introduced as a way to fold economic costs into forecast skill evaluation. Five worked examples build out the framework: a stylized 2x2 flood forecast (supported by the accompanying cost-loss spreadsheet), a multi-action flood forecast application for the City of St. Paul on the Upper Mississippi, a low-flow forecast for barge operations on the Lower Mississippi, a forecast-informed reservoir operations (FIRO) case study at Lake Mendocino, and a revealed-preference analysis of seasonal water supply forecasts used by Colorado farmers.

Acknowledgements:
This research was supported by the Cooperative Institute for Research to Operations in Hydrology (CIROH) with funding under award NA22NWS4320003 from the NOAA Cooperative Institute Program. The statements, findings, conclusions, and recommendations are those of the author(s) and do not necessarily reflect the opinions of NOAA.

ABSTRACT:

This workshop walks through a complete pipeline for building a relational flood inundation map (FIM) database that integrates outputs from multiple modeling sources and ties them to the National Water Model. The motivating problem is fragmentation: flood maps are produced by NOAA OWP, FEMA, USACE, USGS, InFRM, and FHWA in different formats and at different standards, with no common platform connecting them to operational NWM forecasts. The proposed solution is an 8-table SQLite schema storing extent polygons, depth and water-surface-elevation rasters, rating curves, and source metadata, with each flood map linked to NWM feature IDs so that a forecasted flow can be used to retrieve the matching map on demand. Participants work through a four-stage Python workflow split across three Jupyter notebooks. The first stage generates HAND-based FIMs using FIMServe: an area of interest and HUC8 are selected, hydrofabric is pulled from the CIROH S3 bucket, return-period flows are retrieved through the NWM API, and HUC-8 flood maps are merged and clipped as needed. The second stage harmonizes maps from different models by aligning projection, resolution, data type, no-data representation, and grid alignment, then dissolves and smooths vector extents and compresses rasters to reduce file size. Metadata files are then generated to populate the database tables. The third stage builds and populates the SQLite database with content from HAND, HEC-RAS, SRH-2D, AutoRoute, FIER, and satellite-derived sources, and uploads the result to HydroShare. The fourth stage uses the Tulane FIM visualization tool to display flood scenarios, compare maps across models, and overlay custom user files. The workshop closes with implications for research and operations and a look at planned extensions for velocity, impact, and ensemble/probabilistic visualization.

Acknowledgements:
This research was supported by the Cooperative Institute for Research to Operations in Hydrology (CIROH) with funding under award NA22NWS4320003 from the NOAA Cooperative Institute Program. The statements, findings, conclusions, and recommendations are those of the author(s) and do not necessarily reflect the opinions of NOAA.

ABSTRACT:

This workshop aimed at giving participants a sense of: forecast uncertainty, high consequence trade-offs, unclear decision thresholds, spatial variability in the forecast, and forecast latency. Participants were meant to take note of the factors influencing decision making, what information they wish they would've had, and certainty levels. Groups were made of different roles---Weather Forecast Office and Emergency Management staffs. They then went through multiple emergency scenarios based upon real historical storms in Hawaii. Participants were given real data and had to make high-pressure decisions. A debrief of the events, roleplaying, data, and questions from above was conducted after each scenario.

Acknowledgements:
This research was supported by the Cooperative Institute for Research to Operations in Hydrology (CIROH) with funding under award NA22NWS4320003 from the NOAA Cooperative Institute Program. The statements, findings, conclusions, and recommendations are those of the author(s) and do not necessarily reflect the opinions of NOAA.

ABSTRACT:

The aim of this workshop was to translate research into language people can understand, trust, and use. It matters for R2O because the conducted research continues through the understanding and application of information. The workshop goes through common problems with science communication, arguing that a key fix is to start with how the problem relates to the audience. This can be followed with the "so what?" to build a story that really captures audiences. The workshop continues by discussing strategies for reaching audiences and making a high-quality podcast.

Acknowledgements:
This research was supported by the Cooperative Institute for Research to Operations in Hydrology (CIROH) with funding under award NA22NWS4320003 from the NOAA Cooperative Institute Program. The statements, findings, conclusions, and recommendations are those of the author(s) and do not necessarily reflect the opinions of NOAA.

ABSTRACT:

This workshop covers the geodetic and cartographic foundations for working with satellite data, then introduces Satpy as a Python tool for projecting and georeferencing it. The first part lays out the reference framework: geographic coordinate systems, the geoid, and the role of geodesy. It distinguishes horizontal from vertical datums and contrasts local datums (NAD27, ED50) with geocentric ones (NAD83, WGS84), using a timeline of NAD, WGS, and ITRF realizations to show how datums shift over time. The next section covers projected coordinate systems and the geometry of flattening a curved surface onto a plane, walking through planar, cylindrical, and conical projections with common examples including UTM, State Plane, Web Mercator, Lambert Conformal Conic, and Albers Equal Area Conic. The final section introduces Satpy, an open-source Python library developed by the Pytroll community and used operationally by NOAA, EUMETSAT, and meteorological services. It reads more than 70 satellite file formats, calibrates sensor counts to physical units, builds RGB composites, resamples swath data to arbitrary projections, and exports to GeoTIFF, NetCDF, and PNG, with support for major geostationary and polar-orbiting platforms. An accompanying Jupyter notebook demonstrates these capabilities on real satellite data.

Acknowledgements:
This research was supported by the Cooperative Institute for Research to Operations in Hydrology (CIROH) with funding under award NA22NWS4320003 from the NOAA Cooperative Institute Program. The statements, findings, conclusions, and recommendations are those of the author(s) and do not necessarily reflect the opinions of NOAA.

ABSTRACT:

This workshop introduces HydroServer, an open-source platform developed at Utah State University for collecting, managing, and sharing time series data from hydrologic and environmental monitoring sites. Built on the OGC SensorThings standard and extended with environmental metadata attributes, HydroServer provides a web application for site registration and data visualization, a Python client package (hydroserverpy) for programmatic data loading and retrieval, and APIs supporting both real-time sensor streaming and scheduled ETL workflows. Participants work through hands-on Jupyter notebooks covering site setup, data ingestion from CSV files, and metadata and data retrieval using hydroserverpy, with an overview of additional platform capabilities including automated data loading and a data quality control application currently in development.

Acknowledgements:
This research was supported by the Cooperative Institute for Research to Operations in Hydrology (CIROH) with funding under award NA22NWS4320003 from the NOAA Cooperative Institute Program. The statements, findings, conclusions, and recommendations are those of the author(s) and do not necessarily reflect the opinions of NOAA.

ABSTRACT:

This presentation introduces HydroSuite, an open-source ecosystem of more than 140 software tools and libraries developed by the HydroInformatics Lab at Tulane University to support hydrological education, research, and operations. The collection is organized around four areas — data, computing, communication, and community portals — and is built on modern web technologies including WebAssembly, Web Workers, WebRTC, and WebGPU, with the goal of bringing complex hydrologic analysis into the browser with minimal server-side dependency. Core components include HydroLang, a client-side programming framework for hydrological analysis; HydroCompute, a multi-CPU and GPU parallel computing library; HydroRTC, a real-time communication library for decentralized data streaming and sharing; BMI-JS, a JavaScript implementation of the CSDMS Basic Model Interface for coupling models to models and data; and markup-based component libraries (HydroLang-ML and Geo-WC) that expose data retrieval and visualization from agencies such as USGS, EPA, NWS, and FEMA through custom HTML elements. The presentation also lays out a five-phase development roadmap that moves from technical client- and server-side coding toward more intuitive interfaces, including HydroBlox (a block-based visual programming environment), the HydroSuite AI Helper (LLM-driven code assistance), and HydroAI (a voice-enabled agentic interface that generates workflows on a map from natural language).

Acknowledgements:
This research was supported by the Cooperative Institute for Research to Operations in Hydrology (CIROH) with funding under award NA22NWS4320003 from the NOAA Cooperative Institute Program. The statements, findings, conclusions, and recommendations are those of the author(s) and do not necessarily reflect the opinions of NOAA.