Taking a 'Big Data' approach to data quality in a citizen science project.

Area of Science:

• Biodiversity science
• Ornithology
• Computational ecology

Background:

• Experimental data is ideal for causation in biodiversity but not scalable.
• Citizen science data (e.g., eBird) is abundant but noisy and heterogeneous.
• Understanding population-level patterns requires extensive spatial and temporal data.

Purpose of the Study:

• To present a Big Data methodology for enhancing the quality of citizen science bird observation data.
• To demonstrate how to overcome data limitations in large-scale biodiversity monitoring.
• To enable novel population-level analyses using improved citizen science datasets.

Main Methods:

• Utilizing eBird's data submission design for inherent data quality.
• Applying a 'sensor calibration' method to account for individual observer variation.
• Employing species distribution models to address data gaps and improve completeness.

Main Results:

• The Big Data approach significantly improves the accuracy and completeness of citizen science data.
• Individual observer variability in detection and identification can be effectively measured and corrected.
• Data gaps are successfully filled, enabling more comprehensive analyses.

Conclusions:

• Big Data techniques are crucial for leveraging noisy citizen science data in ecology.
• Enhanced eBird data quality supports reliable population-level biodiversity research.
• This approach opens new avenues for understanding bird population dynamics and distributions.