SeasonWatch Project

Overview

The SeasonWatch Project aims to analyze the phenological changes in tree species in India, with a focus on Kerala, using citizen-science data collected from 2015 to 2023. The project examines the relationship between climate change and tree phenology by identifying trends, seasonal shifts, and geographic variations. Our final deliverables include visualizations, statistical analyses, and code that can be reused for similar ecological studies.

Objectives

1. Analyze phenological changes: Investigate how trees respond to climate change and seasonal transitions.
2. Identify key patterns: Focus on the timing of phenological stages (leaves, flowers, fruits) for the top 30 observed tree species.
3. Visualize shifts: Create interactive and static visualizations to convey changes in onset weeks, geographic clustering, and seasonal variability.
4. Provide reusable tools: Develop and share scripts, cleaned datasets, and workflows for future analysis.

---

Deliverables

1. Data Cleaning and Preparation

• Input Data:
  • Citizen-submitted observations from SeasonWatch (2015–2023).
  • ~177 species with detailed phenological stage observations.
• Cleaned Data:
  • Filtered dataset with accurate geocoding, standardized state names, and adjusted missing values.
  • Historical (pre-2020) and comparative (post-2020) datasets for reference.

2. Visualizations

• Interactive Visualizations: Created with Flourish Studio, including:
  • Seasonal shifts in onset weeks.
  • Geographic clustering of phenological stages.
• Static Visuals: Heatmaps, time-series plots, and summary tables saved to: /data/VISUALIZATIONS-fall 2024/Kerala Visuals

3. Statistical Analysis

• Summary statistics, regression analysis, and survival modeling to answer base questions:
  • How are trees changing due to climate change?
  • What is the onset timing for flowering and fruiting in tropical species?
  • What is the probability of transitioning between seasonal states?

4. Final Report

• Comprehensive document with:
  • Visualizations.
  • Interpretations of patterns and trends.
  • Recommendations for conservation efforts.
  • Delivered in PDF format.

---

Getting Started

Prerequisites

1. Python Environment:
   • Python 3.9 or higher.
   • Required libraries:
     • pandas
     • numpy
     • matplotlib
     • geopandas (v0.9.0)
     • shapely (v2.0.1)
     • googlemaps
     • seaborn
2. Geospatial Tools:
   • Shapefiles available in india_map folder for geographic visualizations.
   • Google Maps API key for geocoding (optional).
3. Data Files:
   • Cleaned datasets stored in the /data directory.

Installation

1. Clone the repository:

   git clone https://github.com/your-org/seasonwatch-project.git
   cd seasonwatch-project

2. Set up a virtual environment:

   python -m venv env
   source env/bin/activate  # On Windows: env\Scripts\activate
   pip install -r requirements.txt

3. Download the cleaned datasets and place them in the /data directory.

Running the Code

1. Data Cleaning:

   python scripts/data_cleaning.py

   Cleans raw observation data, standardizes formats, and generates cleaned datasets.

2. Geocoding: If using Google Maps API, update your API key in config.py:

   GOOGLE_API_KEY = 'your-api-key'

   Run geocoding script:

   python scripts/geocoding.py

3. Visualization Generation: Generate visuals:

   python scripts/visualizations.py

   Outputs saved in /data/VISUALIZATIONS-fall 2024.

4. Statistical Analysis: Perform analyses and generate summary tables:

   python scripts/analysis.py

---

Directory Structure

seasonwatch-project/
├── data/
│   ├── raw_data.csv
│   ├── cleaned_data.csv
│   ├── VISUALIZATIONS-fall 2024/
│   │   ├── Kerala Visuals/
│   │   └── ...
├── scripts/
│   ├── data_cleaning.py
│   ├── geocoding.py
│   ├── visualizations.py
│   └── analysis.py
├── india_map/
│   ├── shapefiles/
│   └── ...
├── README.md
├── requirements.txt
└── config.py

---

Blockers Faced and Solutions

Blockers

1. Google API Costs:

   • Budget constraints limited the number of geocoding requests.
   • Solution: Implemented a caching mechanism to minimize API calls and explored free alternatives like OpenStreetMap.

2. Anomalies in Data:

   • Missing or incorrect data for several observations.
   • Solution: Used statistical imputation techniques and cross-referenced with the SeasonWatch tree phenology handbook to standardize missing values.

3. Processing Speed:

   • Geocoding and data cleaning scripts were slow due to large datasets.
   • Solution: Optimized scripts using try-except blocks for error handling and parallel processing where possible.

4. Prior Data Loss:

   • Previous teams dropped too many rows during data cleaning.
   • Solution: Reviewed raw data meticulously and identified only 20,000 rows with missing values, preserving as much data as possible.

---

Next Steps for Future Teams

1. Enhance Visualizations:

   • Improve interactivity in Flourish Studio and integrate additional features, such as user filtering by state or species.
   • Explore advanced visualization libraries like Plotly or D3.js for greater customization.

2. Expand Analysis:

   • Incorporate additional years of data beyond 2023 to track long-term trends.
   • Perform deeper survival and Markov modeling to understand tree state transitions.

3. Climate Correlation:

   • Integrate external climate datasets (e.g., rainfall, temperature) to analyze correlations with phenological changes.

4. Optimize Geocoding:

   • Automate retries for failed geocoding requests and further explore OpenStreetMap for cost-free options.

5. Machine Learning Models:

   • Apply machine learning techniques to predict phenological stages based on climate and temporal data.

6. Documentation:

   • Update README and inline comments regularly for new tools or methods added to the project.

Contributors

• Team Members: Cecily Wang-Munoz, Aditya Chopra, An Ngo (Sue), Brenda Kim

---

Acknowledgments

This project uses data from SeasonWatch and insights from the SeasonWatch tree phenology handbook. Special thanks to BU SPARK for supporting the project.