Hydrology

Over the past year I was fortunate to have the opportunity to develop and lead the execution of Aero-Graphics' USGS 3DHP - Elevation Derived Hydrography contract.

USGS' 3DHP program is the first systematic remapping of our nations hydrology since the original 1:24,000-scale topographic mapping program, which was active from 1947 to 1992. That means there has been nearly 30 years of technical advances to support more accurate hydrological mapping efforts.

In 2013, USGS established the 3DEP program, with the goal of achieving nationwide high-resolution lidar coverage at 4–8+ points per square meter. Today, with roughly 95% of the United States mapped under 3DEP, the need and opportunity for a modern, high-resolution hydrographic framework became exteremely clear.

This program is special to me because it is an excellent culmination of technical skills and hydrological/ geomorphological knowledge. In the sections that follow, I outline my general approach to the work, along with key challenges I've encountered and the solutions that have proven most effective.

Tools

The EDH workflow is built around a combination of open-source GIS tools and custom Python scripts. Key tools include:

• QGIS for data visualization, manipulation, and analysis.
• GDAL for raster(dems) data manipulation and analysis.
• WhiteboxTools for advanced terrain analysis and hydrologic modeling.
• Custom Python scripts for workflow automation, data validation, and QA/QC processes.
• GRASS GIS for robust hydrologic modeling tools (flow accumulation and strahler ordering).

Hydro-Enforcement

Running a flow accumulation algorithm on raw DEMs does not always produce an accurate flowline network. Water in the real world often flows in areas that LiDAR cannot reach. Obvious examples are present in built environments, namely culverts, bridges, and dams. Less obvious examples might be water flowing through dense vegetation, karst terrain, or beneath natural debris.

This shows the importance of hydro-enforcing DEMs for hydrologic mapping

Feature Classification

A critical step in the EDH workflow is the classification of hydrologic features. This includes identifying and delineating waterbodies (rivers, lakes/ponds, ocean) and flowlines (streams, canals, pipelines).

A canal draining a reservoir for irrigation

Accurate classification ensures that the derived hydrography reflects real-world conditions. This process involves a combination of automated algorithms that I developed, and a strong eye for detail during manual review.

An excellent example of this is a bank detection algorithm developed to delineate riverbanks using a combination of slope, elevation, and flow accumulation thresholds. This algorithm significantly improved the accuracy of waterbody boundaries. The results were then fed into a convolutional neural network to achieve superb bank predictions at scale.

Nuanced topological relationships between flowlines and waterbodies

Topology

Ensuring topological correctness in the derived hydrography is paramount. Topological requirements for EDH is extremely nuanced. Flowlines must connect as a clean directed graph with all nodes touching in the X,Y, and Z dimensions, maintaining consistently downward flow. Waterbodies must split flowlines, and have the Z value of the outflowing vertex, with island ring vertices digitized in the clockwise direction.

Knowing the depth of the topological requirements and the size of the dataset, I implemented a series of QC tools and topological checks throughout the workflow.

Elevation profile of DEM surface and a stream that flows beneath a road

Why it matters

High-resolution hydrography is crucial for a wide range of applications, including flood modeling, water resource management, environmental conservation, and urban planning. USGS 3DHP program which ingests the Elevation Derived Hydrography aims to provide a more accurate and detailed representation of the nation's waterways, which will allow for more informed decisions for a wide variety of stakeholders.