README.md

---
license: cc-by-4.0
task_categories:
  - tabular-classification
  - tabular-regression
  - other
language:
  - en
tags:
  - biology
  - ecology
  - ornithology
  - birds
  - biodiversity
  - sri-lanka
  - species-distribution-modeling
  - environmental-science
  - climate-change
  - geospatial-machine-learning
  - remote-sensing
  - ecological-modeling
size_categories:
  - 1M<n<10M
pretty_name: Sri Lanka Bird Diversity Dataset
---

# Sri Lanka Bird Diversity Dataset

## Dataset Summary

The **Sri Lanka Bird Diversity Dataset with Environmental and Climate Features** is a large-scale biodiversity dataset containing **1,552,048 bird occurrence records** covering **429 bird species** across all **25 provinces of Sri Lanka**.

The dataset integrates bird observation records with spatial, climatic, environmental, and land-cover variables extracted at each observation location. It provides a comprehensive resource for machine learning, ecological modeling, and biodiversity research.

The dataset spans observations collected between **2014 and 2024**, enabling studies on:

- Species distribution modeling (SDM)
- Climate–biodiversity relationships
- Habitat suitability prediction
- Bird species classification
- Environmental impact analysis
- Ecological machine learning applications


---

# Dataset Details

## Dataset Description

Each record represents a bird observation associated with geographic coordinates, observation metadata, and environmental conditions at the observation location and time.

The dataset contains:

- **1,552,048 observation records**
- **429 unique bird species**
- **25 provinces of Sri Lanka**
- **10 years of observations (2014–2024)**
- **24 feature variables**

The dataset combines biodiversity observations with environmental variables derived from satellite imagery, climate models, and atmospheric datasets.

---

## Dataset Creators

Developed by:

- Dilusha Chandrasiri
- Maneesha Herath
- Muditha Herath
- Yasith Hewarathna
- Gishan Bandara

---

# Data Sources

The dataset integrates information from multiple publicly available sources:

### Bird Occurrence Data

- **Global Biodiversity Information Facility (GBIF)**  
  Bird observation records with species taxonomy and geographic information.

### Environmental Variables

| Variable Group | Source |
|---|---|
| Vegetation indices | Satellite remote sensing data (MODIS/Sentinel-derived products) |
| Land cover | ESA WorldCover / land-use datasets |
| Elevation | SRTM Digital Elevation Model |
| Climate variables | ERA5 climate reanalysis |
| Atmospheric variables | NASA MERRA-2 aerosol reanalysis |
| Geographic information | Sri Lankan administrative boundaries |

Users should acknowledge and cite the original data providers when using this dataset.

---

# Dataset Structure

The dataset is provided as:


The file contains **24 columns**.

---

# Features

| Feature | Type | Description |
|---------|------|-------------|
| `index` | Integer | Unique row identifier |
| `verbatimScientificName` | String | Scientific name of observed bird species |
| `stateProvince` | String | Province where observation occurred |
| `individualCount` | Float | Number of observed individuals |
| `decimalLatitude` | Float | Latitude coordinate (WGS84) |
| `decimalLongitude` | Float | Longitude coordinate (WGS84) |
| `eventDate` | Date | Observation date |
| `avg_rad` | Float | Average surface radiation |
| `NDVI_raw` | Float | Raw vegetation index |
| `NDVI` | Float | Normalized vegetation index |
| `LandCover_Class` | Integer | Land cover category identifier |
| `elevation_meters` | Integer | Elevation above sea level |
| `Carbon_Mass` | Float | Atmospheric carbon aerosol mass |
| `Dust_Mass` | Float | Atmospheric dust aerosol mass |
| `SO2_Mass` | Float | Sulfur dioxide aerosol concentration |
| `Sulfate_Mass` | Float | Sulfate aerosol concentration |
| `Sea_Salt_Mass` | Float | Sea salt aerosol concentration |
| `Total_Aerosol_Extinction` | Float | Aerosol optical extinction |
| `temp_mean` | Float | Mean temperature |
| `rainfall` | Float | Rainfall measurement |
| `wind_mean` | Float | Average wind speed |
| `humid_mean` | Float | Relative humidity |
| `shortwave_radiation` | Float | Solar shortwave radiation |
| `lka_general_2020` | Float | Sri Lankan environmental baseline indicator |

---

# Dataset Statistics

| Statistic | Value |
|-----------|-------|
| Total records | 1,552,048 |
| Bird species | 429 |
| Geographic coverage | Sri Lanka |
| Provinces covered | 25 |
| Observation period | 2014–2024 |
| Features | 24 |
| Missing values | None after preprocessing |

---

# Geographic Coverage

The dataset covers all provinces of Sri Lanka:

- Ampara
- Anuradhapura
- Badulla
- Batticaloa
- Colombo
- Galle
- Gampaha
- Hambantota
- Jaffna
- Kalutara
- Kandy
- Kegalle
- Kilinochchi
- Kurunegala
- Mannar
- Matale
- Matara
- Monaragala
- Mullaittivu
- Nuwara Eliya
- Polonnaruwa
- Puttalam
- Ratnapura
- Trincomalee
- Vavuniya


---

# Data Processing Pipeline

The dataset was generated through the following processing workflow:

1. Bird occurrence records were collected and filtered for Sri Lanka.
2. Records with invalid geographic coordinates were removed.
3. Species names were standardized using taxonomic information.
4. Environmental raster datasets were spatially sampled at each observation location.
5. Climate variables were matched based on observation date and location.
6. Land-cover and elevation information were extracted.
7. Environmental features were normalized where required.
8. Missing records were removed.
9. Final feature vectors were generated for machine learning applications.


---

# Intended Uses

This dataset is suitable for:

## Ecological Applications

- Species distribution modeling
- Habitat suitability analysis
- Biodiversity assessment
- Environmental impact studies
- Climate change research

## Machine Learning Applications

- Multi-class species classification
- Regression-based abundance prediction
- Geospatial prediction models
- Feature importance analysis
- Explainable AI studies in ecology


---

# Example Machine Learning Tasks

| Task | Target Variable | Input Features |
|-|-|-|
| Species Classification | `verbatimScientificName` | Environmental + geographic features |
| Bird Abundance Prediction | `individualCount` | Climate + habitat variables |
| Habitat Modeling | Species presence | Location + environmental variables |
| Regional Biodiversity Analysis | Province/species distribution | Full feature set |

---

# Dataset Limitations and Biases

## Sampling Bias

Bird occurrence datasets collected from citizen science platforms may contain:

- Uneven geographic coverage
- Higher observation density near accessible locations
- Seasonal observation biases

## Environmental Resolution

Environmental variables are derived from remote sensing and climate products. Their spatial resolution may not perfectly represent local habitat conditions.

## Taxonomic Limitations

Species identification accuracy depends on the quality of original observation records.

## Recommended Use

Researchers should consider sampling bias correction and ecological validation before deploying models for conservation decision-making.

---

# Ethical Considerations

This dataset is intended for scientific research and educational purposes.

Users should avoid:

- Using predictions without ecological validation
- Drawing conservation conclusions from biased samples
- Misinterpreting correlations as causal relationships

---

# License

This dataset is released under the:

**Creative Commons Attribution 4.0 International (CC BY 4.0)**

Users are free to share and adapt the dataset with appropriate attribution.

---

## Related Paper

This dataset accompanies the following research paper:

**How Environment and Urbanization Shape Bird Diversity in Sri Lanka**

**arXiv:** https://arxiv.org/abs/2607.00582

The paper describes the methodology for constructing this dataset, including data collection, environmental feature extraction, preprocessing, and the machine learning analyses performed using the dataset. Readers interested in the complete methodology and experimental results are encouraged to refer to the paper.

If you use this dataset in your research, please consider citing both the dataset and the accompanying paper.

# Citation

If you use this dataset, please cite:

```bibtex
@dataset{sri_lanka_bird_diversity_2026,
  title        = {Sri Lanka Bird Diversity Dataset with Environmental and Climate Features},
  author       = {Chandrasiri, Dilusha and Herath, Maneesha and Herath, Muditha and Hewarathna, Yasith and Bandara, Gishan},
  year         = {2026},
  publisher    = {Hugging Face},
  version      = {1.0},
  license      = {CC-BY-4.0}
}