WindNinja Preprocessing and Postprocessing Pipeline

This pipeline prepares DEMs, runs WindNinja simulations, and postprocesses the output, including smoothing and mosaicking.

---

Dependency requirements for python scripts

GDAL Multiprocessing Subprocess Numpy

Step 1: Prepare DEMs for WindNinja

This step generates UTM-projected tiles with overlap for WindNinja input.

Strategy to geenrate square tile in UTM projection to get rid of NoData

(Note: we are cliping out above square tile from 350% overlapped tile. I am generating 350% overlapped tile as I won't have whole input landscape file of CONUS(237GB) into my memory to get data after. I am saving it as intermediatory output and deleting it after getting square tile.)

Modify Paths:

Edit prepare_dems.py and set:

input_raster_path = <path for landscape file>
output_directory = <path for output folder>

Run:

python3 prepare_dems.py

You will get below file structure

output_directory/
├── original_tiles/
│   ├── tile_0_x_y.tif
│   ├── tile_1_x_y.tif
│   └── ...
├── temp_albers_tiles/
│   ├── temp_350_albers_0.tif
│   ├── temp_overlap_albers_0.tif
│   ├── temp_utm_overlap_aoi_0_epsgXXXX.tif
│   ├── temp_utm_350_tile_0_epsgXXXX.tif
│   └── ...
├── utm_aoi_tiles/
│   ├── utm_overlap_aoi_0_epsgXXXX.tif
│   ├── utm_overlap_aoi_1_epsgXXXX.tif
│   └── ...
└── final_utm_tiles/
    ├── 0/
    │   ├── dems_folder/
    │   │   ├── dem0/
    │   │   │   ├── dem0.tif
    │   │   │   └── ...

Step 2: Run WindNinja Simulations on Slurm

The conus.sbatch script runs WindNinja simulations on an HPC cluster using Slurm.

Modify Paths

Before running the WindNinja simulations, ensure that the paths in conus.sbatch are correctly set according to your environment:

• SHARED_STORAGE: Path where WindNinja output folder final_utm_tiles is stored
• WINDNINJA_SIF: Path to the Singularity image for WindNinja.
• Scripts and Configs: Ensure run_caryWnCfg3.py, run.sh, and base_cli.cfg exist in the specified directory.

SHARED_STORAGE= <path>
WINDNINJA_SIF= <path>

sbatch conus.sbatch

To monitor logs:

tail -f windninja_pipeline_<job_id>.log

Step 3: Postprocessing (Smoothing and Mosaicking)

Modify Paths

Before running this step, update the paths in postprocess.py:

• base_dir: Path to the processed WindNinja tiles.
• aoi_tiles_dir: Directory containing AOI tiles. (which was generated by prepare_dems scripts(Step 1))
• output_dir: Directory where processed outputs will be saved.

Modify these in postprocess.py:

base_dir = <path>
aoi_tiles_dir = <path>
output_dir = <path>

Run:

python3 postprocess.py

You will get below file structure

output_directory/
├── windNinjax_tiles_momentum/
│   ├── 0/
│   │   ├── dem0_0_5_120m_reproj.tif
│   │   ├── dem0_0_5_120m_clipped.tif
│   │   ├── dem0_0_5_120m.tif
│   │   └── ...
│   ├── 1/
│   │   ├── dem0_0_5_120m_reproj.tif
│   │   ├── dem0_0_5_120m_clipped.tif
│   │   ├── dem0_0_5_120m.tif
│   │   └── ...
│   ├── final_mosaic_w_smoothing_momentum.tif