SeismicX-Cont / scripts /README_download_data.md
cangyeone's picture
Sync canonical release scripts and config
96864ec verified
|
Raw
History Blame Contribute Delete
2.73 kB
# Data Download and Rebuild Notes
This mini package already contains the two hourly HDF5 files, the SQLite index,
the mini annotation JSON, station metadata, response metadata, example picker
outputs, and validation outputs needed for quick inspection.
## Download the Complete Dataset
The complete SeismicX-Cont v1.0.0 data product is hosted in two public dataset
repositories:
```text
https://huggingface.co/datasets/cangyeone/SeismicX-Cont
https://doi.org/10.57967/hf/8993
https://www.modelscope.cn/datasets/cangyeone/SeismicX-Cont
```
Both hosting repositories are aligned to the same release file organization.
The root `manifest_filesizes.tsv` lists the expected paths and file sizes for
quick package inspection.
Hugging Face example:
```bash
git lfs install
git clone https://huggingface.co/datasets/cangyeone/SeismicX-Cont
```
ModelScope example:
```bash
git lfs install
git clone https://www.modelscope.cn/datasets/cangyeone/SeismicX-Cont.git
```
For command-line downloads without a full git clone, use the official client
for the platform you prefer. For example:
```bash
huggingface-cli download cangyeone/SeismicX-Cont \
--repo-type dataset \
--local-dir SeismicX-Cont
```
or with ModelScope Python utilities:
```python
from modelscope.hub.snapshot_download import snapshot_download
snapshot_download(
"cangyeone/SeismicX-Cont",
repo_type="dataset",
local_dir="SeismicX-Cont",
)
```
## Rebuild From Raw Waveforms
The mini package includes the processing scripts needed to rebuild the mini
HDF5 derivative product after you obtain the source MiniSEED waveform archive.
The expected local raw-data layout is:
```text
$SEISMICX_RAW_ROOT/07/06
$SEISMICX_RAW_ROOT/11/14
```
Rebuild the two mini HDF5 files:
```bash
SEISMICX_RAW_ROOT=/path/to/continous_usa/data
python scripts/makeh5_flex_seg.py \
--input_dir "$SEISMICX_RAW_ROOT/07/06" \
--loc_file data/label/stations.csv \
--output data/hdf5/continuous_waveform_usa.h5 \
--split_interval hour \
--include_split_file_id 20190706_04 \
--compression gzip \
--compression_opts 4
python scripts/makeh5_flex_seg.py \
--input_dir "$SEISMICX_RAW_ROOT/11/14" \
--loc_file data/label/stations.csv \
--output data/hdf5/continuous_waveform_usa.h5 \
--split_interval hour \
--include_split_file_id 20211114_16 \
--compression gzip \
--compression_opts 4
```
Rebuild the SQLite waveform index and manifest:
```bash
./scripts/build_waveform_index.sh
./scripts/write_manifest.sh
```
The copied `essd_scripts/` directory is included for manuscript-statistic and
figure reproducibility. Those scripts are designed for the complete release; run
them from the full SeismicX-Cont repository when reproducing the ESSD paper.