2. Data Frequency and Temporal Aggregation
Sampling Frequency
| Metric | Value |
|---|---|
| Raw Sampling Rate | Every 10 minutes |
| Data Points per Hour | 6 measurements |
| Daily Data Points | ~144 measurements per 24-hour period |
Pre-Aggregated Data on Platform
The AirQo Analytics platform provides pre-computed temporal averages: hourly, daily, weekly, and monthly. These aggregations are calculated from the raw 10-minute measurements to facilitate analysis.
Handling Missing Data
Our approach to data gaps:
- No interpolation — We do not perform backfilling, forward filling, or imputation of missing values. When a monitor fails to capture data during a specific period, the value remains null/empty in the dataset.
- Rationale — Air pollution levels can change rapidly due to traffic, weather, and local emissions, making interpolated values unreliable for exposure assessments.
Data Completeness Threshold
For robust analysis, we recommend applying a data completeness criterion:
- Recommended threshold — Use only data from monitors that provide ≥50% of expected measurements for your analysis period.
- Hourly analysis — If a monitor has fewer than 12 readings in a 24-hour period, consider excluding that day.
- Daily averages — If fewer than 72 measurements (50% of 144 expected) are available, the daily average may not be representative.
Recommended Analytical Workflow
For time-series analysis:
- Download hourly (or raw 10-minute) data from the platform in quarterly batches.
- Check data completeness for each monitor and time period.
- Apply your completeness threshold (we recommend 50% minimum).
- Compute daily averages from hourly data where applicable.
- Aggregate to weekly or monthly levels as needed.
- Document and report your data quality control steps.