changliao1025 / hexwatershed

The C++ backend of the HexWatershed a mesh independent flow direction model for hydrologic models

What's new

• HexWatershed produces the world's first flow direction map for the whole Arctic using the MPAS mesh.

• 2024-06-15, I gave a talk in the ‘Advanced Computational Methods for Climate Modeling and Analysis’ session of the ESCO 2024 - 9th European Seminar on Computing ESCO2024. This talk focused on km-scale river routing modeling in ESM.

• 2024-05-15 I gave a workshop ‘Mesh-independent flow direction modeling using HexWatershed 3.0’ in the CSDMS annual meeting at Montclair State University CSDMS 2024.

Introduction

I am an Earth Scientist at Pacific Northwest National Laboratory (PNNL).

My research mainly focuses on high-resolution Earth system modeling at both regional and global scales.

I wrote casual articles at https://changliao.github.io/

I developed models and tools to understand Earth's system.

• 🔭 I’m currently working on HexWatershed (www.hexwatershed.org and www.twitter.com/hexwatershed) model
• 🌱 I’m currently learning Hilbert Space, Graph Theory, and GPU
• 👯 I’m looking to collaborate on hydrology-related projects
• 🤔 I’m looking for help with math and physics
• 💬 Ask me about high-performance computing, hydrology, and RS/GIS.
• 📫 How to reach me: Bluesky (https://bsky.app/profile/changliao.bsky.social) or changliao.climate at gmail.com.
• 😄 Pronouns: ...
• ⚡ Fun fact: I can roll a kayak.

hexwatershed's Issues

Order in json file

The watershed cell is not ordered the same in different process. So in the post-processing, it is critical to re-order them if index is used.

Elevation profile based slope when topology is off

There seems to be a bug when the topology is off; the slope is not calculated.

Fix a bug in accumulation threshold

The minimal accumulation threshold was updated.

Vertex ID

Vertex ID is needed for VTK, but this information should not be built by hexwatershed, instead, it should be defined by the mesh generation process, see MPAS mesh specs.

Hierarchical data output

Since the model is using the hierarchical data model to calculate the attributes of each cell, it is desirable to update the output structure.
For example, the model can output individual segments/subbasins/watersheds.

Add a flag to disable the flow-to-ocean behavior

Currently, when in a coastal zone, the land to ocean requirement changes the normal flow direction algorithm. I will add a flag to control this feature.

Don't use `build` dir

The standard cmake sequence is:

mkdir build
cd build
cmake ..
make

That is, the directory build is often used as the place to put an out-of-source build. It would be better if you weren't using it to store files.

Domain or watershed data operation

Some algorithms are better handled at watershed scale. However, the model is expected to run with multiple watersheds. As a result, some redesign may be needed.

Preserve the stream info from pyflowline

The json file from PyFlowline contains a flag on whether a cell is a stream grid or not. However, the flow accumulation is also used to define the stream grid. To preserve the original information, it is better to keep the flag even though its accumulation may be smaller than the threshold.

TWI slope not updates

Global and local parameter for basin

If a local parameter is provided, then the global parameter will be ignored.
This allows different parameters for different basins.