parthawrk/as-dataset

Indian Meteorological Department (IMD) and IMDAA Climate Datasets

This repository hosts a curated collection of high-resolution historical gridded weather and climate datasets for the Indian subcontinent. It aggregates long-term daily observations from the Indian Meteorological Department (IMD) alongside high-frequency hourly reanalysis data from the Indian Monsoon Data Assimilation and Analysis (IMDAA) project.

The dataset totals approximately 100 GB and is stored in NetCDF (.nc) and zipped formats to facilitate climate research, monsoon modeling, extreme event analysis, and machine learning-based weather forecasting.

Dataset Contents

1. IMD Daily Rainfall Dataset

• Temporal Coverage: 1901 – 2025 (125 continuous years)
• Spatial Resolution: 0.25° × 0.25° gridded resolution
• Geographic Coverage: Pan-India (Whole Mainland India Grid)
• Frequency: Daily cumulative rainfall
• Format: NetCDF (.nc) / Compressed archives (.zip)

2. IMD Daily 2-Meter Temperature (t2m) Dataset

• Geographic Coverage: Pan-India
• Frequency: Daily values (including maximum, minimum, and mean temperatures where applicable)
• Format: NetCDF (.nc) / Compressed archives (.zip)

3. IMDAA Hourly Reanalysis Dataset

• Parameters: 2-Meter Temperature (t2m) and Rainfall (rf)
• Frequency: Hourly high-frequency intervals
• Format: NetCDF (.nc) / Compressed archives (.zip)
• Context: Sourced from the high-resolution regional reanalysis over the South Asian region, capturing diurnal cycles and localized convective systems.

Intended Use & Research Context

This repository is maintained for ongoing academic research, climate trend analysis, and regional hydrological modeling. Due to the large volume of the binary files (~100 GB total), users are highly encouraged to use the huggingface_hub caching mechanism rather than issuing frequent full repository downloads.

Recommended Data Access (Python)

To avoid unnecessary bandwidth usage and leverage local caching, download specific files via the huggingface_hub SDK:

from huggingface_hub import hf_hub_download

# Example: Download a specific NetCDF file
file_path = hf_hub_download(
    repo_id="parthawrk/as-dataset", 
    filename="Data/IMD/IMD_RF25_1901_2025_merged.nc", 
    repo_type="dataset"
)
print(f"File cached locally at: {file_path}")

Data Formats & Visualisation

Because the data is primarily stored in NetCDF (.nc) format, it cannot be rendered natively by the Hugging Face Dataset Viewer. To explore and interact with the arrays, use Python libraries such as xarray, netCDF4, or rioxarray.

import xarray as xr

# Open the cached NetCDF file
ds = xr.open_dataset(file_path)
print(ds)

Citation & Acknowledgments

• IMD Data: Kindly acknowledge the India Meteorological Department, Ministry of Earth Sciences (MoES), Government of India, for the primary gridded meteorological observations.
• IMDAA Data: Acknowledge the National Centre for Medium Range Weather Forecasting (NCMRWF), MoES, UK Met Office, and the National Centre for Atmospheric Science (NCAS) for the reanalysis products.