Hydrocarbon Spill Hyperspectral Dataset (HSHD

Instructions for Using the Dataset

1. Accessing the Dataset

1. Repository Location:

   • The dataset is organized in a cloud repository. Ensure you have access to the provided repository link (e.g., Google Drive, IEEE DataPort, or equivalent platform).

2. Repository Structure:

   • Each file contains information on specific samples, hierarchically organized into folders:
     • Clean: Reference samples without contamination (m).
     • Gasoline: Samples contaminated with gasoline (g).
     • Thinner: Samples contaminated with thinner (d).
     • Motor Oil: Samples contaminated with used motor oil (a).
   • Each file is labeled following the scheme:
     ContaminantType_Number (e.g., g0001 for the first gasoline sample).

2. Data Format Description

• File Format:
  The data is stored in the ENVI format, commonly used for remote sensing applications.

  • .hdr Files: Contain metadata such as wavelength ranges, pixel dimensions, and exposure conditions.
  • .dat Files: Contain the hyperspectral data cubes.

• Resolution and Specifications:

  • Spatial resolution: 1024 × 1024 pixels.
  • Spectral bands: 20 bands ranging from 400 nm to 1000 nm.
  • Bit depth: 12 bits per pixel.

3. Software Requirements

1. Hyperspectral Processing Software:
   • Use tools like ENVI, QGIS, or any software compatible with the ENVI format to visualize and analyze the data.
2. Custom Programming:
   • If you prefer using scripts, libraries such as Spectral Python (SPy), OpenCV, or NumPy can be used to manipulate hyperspectral data.
3. System Requirements:
   • Processor: Intel Core i5 or higher.
   • RAM: Minimum 8 GB.
   • Disk Space: At least 5 GB free.

4. Analysis Workflow

1. Load the Data:

   • Download the files from the repository and save them to your local system.
   • Maintain the folder structure and metadata to ensure proper traceability during the analysis.

2. Data Preprocessing:

   • Use the .hdr files to extract metadata and calibrate the spectral data.
   • Apply preprocessing techniques such as noise reduction or normalization to enhance data quality.

3. Data Visualization:

   • Use hyperspectral visualization tools to explore the spectral and spatial characteristics of the samples.
   • Compare clean samples (m) with contaminated samples (g, d, a) to identify unique spectral signatures.

4. Advanced Analysis:

   • Implement machine learning models or statistical tools to classify contaminants based on their spectral signatures.
   • Develop predictive models using the temporal evolution of the contaminant data.

5. Citation

• Please cite this dataset in your publications or projects using the provided citation format in the dataset metadata.